Your beloved PCs deserve a regular clean as part of their maintenance schedule. If you miss a seasonal clean or two it is easy for dust to build up and transform from a slight blemish on the system's appearance to something that actually hinders cooling and performance. For all your dusty electronics cleaning tasks, a capable air duster can save the day, and we've spotted a limited-time deal on the powerful Wolfbox MF60 Compressed Air Duster. It is at one of its best prices ever, reduced by 39% from $49.99 to $30.39, for a limited time on Amazon.

• Check out this deal on Amazon

No matter how clean you keep your home, electrical equipment always attracts dust. With PC airflow cases pulling in air from outside to keep components cool, there's always a fresh intake of detritus every time it is used. Laptop fans can be an even worse dust magnet if they are regularly carried around in bags, rested on soft furnishings - an issue exacerbated by the small fans and vents on portables.

This is where a capable dust buster can be useful. It can make maintenance effortless and convenient. The Wolfbox MF60 Compressed Air Duster, for example, is a cordless rechargeable device weighing just half a pound. Users can select one of its three fan speeds, and its brushless motor will kick in to deliver up to 72.4 m/s wind speed using its 110,000 RPM fan.

For less strenuous dust busting work, or on more delicate items, users can dial in the fan at 25K RPM, or 60K RPM. On the lowest setting you'll get up to 240 minutes of air dusting from a single charge. Charging up from empty to full will take 3.5 hours. A USB charging cable is supplied in the box.

Another attraction of the Wolfbox MF60 Compressed Air Duster is the nozzle selection in the box. These include a flat nozzle for blowing dust off PC cases and desks, a long nozzle for deep cleaning computer keyboards and car vents, a round all-purpose nozzle, and there's a swim ring/air mattress nozzle for outdoor inflation and camping.

If you're looking for more savings, check out our Best PC Hardware deals for a range of products, or dive deeper into our specialized SSD and Storage Deals, Hard Drive Deals, Gaming Monitor Deals, Graphics Card Deals, gaming chair, or CPU Deals pages.

Noctua is known for making high-end, high-quality fans and air coolers, which it has been doing for over 20 years, recently expanding into AIOs. However, it appears that there were a few questions regarding its PC case compatibility database, which some PC DIYers say don’t match the data from the manufacturer’s website. The company cleared this up in a blog post, saying that it has undertaken its own measurements and it has determined that 56 of the more than 100 cases it has tested showed significant deviation from the details officially listed online.

The company gave two diverging situations of inaccurate official measurements. The first one is when the case is larger than what is actually written, which would allow PC builders to attach a larger cooler. Noctua said that the deviations aren’t minor, either. For example, the NZXT H2 Flow case is listed as being compatible with a maximum 75mm CPU cooler height, but the actual case can accommodate an air cooler up to 79mm. There’s also the Thermaltake TR100, which is 2.7mm larger than what the Thermaltake website says, and the Corsair Frame 4000D, which gives you 10mm of extra clearance.

This is a good thing, as any cooler you bought based on their website’s measurement means that it will definitely fit, but Noctua said that this is a missed opportunity. That’s because you could’ve bought a larger cooler that delivered more performance in the same case.

On the flip side, the company warned that there were also some cases that were actually smaller versus the numbers in their specifications. It cited two examples — the Asus Prime AP201 and the Lian Li B4-mATX. The former is 1.7mm shorter while the latter is 3.5mm shorter than what’s listed on their specific product pages.

This could to be a major issue, especially if you bought items that fitted the available space exactly, based on the manufacturer’s measurements. It would mean that these parts would either be touching the case panels, or worse, not allow them to close at all. If that’s the case, you’ll have to spend extra effort to return either the part or the case and get something that fits better.

The safest solution to this is for you to buy the PC case before you shop for the remaining parts of your build. This is especially true if you’re looking at the best mini-ITX cases, as the tight tolerances mean that you don’t have much margin if the numbers aren’t accurate. But by having the case physically available and measuring it yourself, you’re at least confident that what you’re buying will fit. Alternatively, you can check out Noctua’s compatibility search page, with the clearances of the cases in its database verified by its team.

This compatibility tool will give you peace of mind when shopping for the best cooler and is useful for confirming measurements even if you end up buying from another brand. But this begs the question — are the variations between official and actual measurements limited to the air coolers? Or are the numbers for other components, like the GPU, PSU, and other key components also inaccurate? Noctua’s website only covers coolers, so you’ll have to rely on other pages (or your own measurements) if you want some confirmation of a case’s specifications and clearances for other parts like GPUs and PSUs.

The fearless TrashBench has been testing direct die water cooling of graphics cards. The inevitable twist here is that they removed the metal waterblock from the equation. Despite initial leaky results and much concern of water damage to the expensive parts thrown into the mix, some of the end results are eyebrow-raising in a good way.

“In my pursuit of the perfect water block, I realized, why do we need metal at all?” queried the antipodean host of the punk-rock GPU death lab. “Why can't we just pump water directly over the bare silicon? Makes sense to me.”

TrashBench began this project well aware that leaks might be a problem. Thus, a non-working GeForce RTX 3060 was chosen for the initial feasibility tests. It was measured up for a 3D printed water block which would direct the coolant directly over the GPU die. Our hardware adventurer covered any surface-mount components beside the GPU with nail polish to prevent water damage/shorting.

Some tried and trusted plumbing measures were then applied, with the design incorporating washers, gaskets, and hose clamps (worm-screw equipped jubilee clips). Water pipe fittings were melted into the block, which was fastened using the retaining clamp mechanism that the air cooler had previously used.

This initial test sprang a leak, or two, so TrashBench decided to add epoxy adhesive to secure and seal the 3D printed block to the GPU. This worked, but then some water was spotted oozing from the tube fittings, so they got a dose of epoxy too. At last, we had the first non-leaky prototype…

After some live runs using a powered-up GTX 980 were successful, it was time for the main event with fitting, testing, and benchmarking using an RTX 2060 Super. On the way, it was noticed that the choice of 3D print material used on TrashBench’s Bambu Lab 3D printer wasn’t trivial. Some materials leaked from the edges, and others were actually slightly porous. During live power tests, the TechTuber wisely used a PCIe riser cable so any leaks wouldn’t immediately drip onto the motherboard.

Was it worth it?

If you like to live dangerously, it could be argued that the TrashBench results show some value in this cooling methodology. The headlining technique was compared against the stock cooler and a clamped-on AiO cooling solution.

Going by the numbers above, you can see that direct water cooling could be interesting, as long as you can ensure long-term confidence about leakage. In the video, you can also see some interesting extras, like running this system using -28°C coolant. The same trick is also tested using an Intel i5-7600K CPU. That was chosen as “it’s not a particularly hot CPU, but if it dies, I don’t care.”

The eventual conclusion was that chamber size matters (most clearly illustrated by the CPU cooling test), but pumping water directly over silicon can work.

Two new Geekbench listings for Nvidia's upcoming RTX Spark laptop superchip have surfaced, confirming a second, cut-down SKU exists with 18 CPU cores, at least in the testing phases. Both parts score remarkably close to each other, posting impressive multi-core numbers, but their single-core performance isn't special. Still, these are preliminary results on unreleased hardware, so take these numbers with a grain of salt.

The model name for the 20-core (10+10) SKU was shown as "OEMQAJ 766_MIS Product Name DV" with its clocks reported at 4.0 GHz. This chip scored 2,570 points in the single-core test and 23,126 points in the multi-core test. The second listing showed up as "OEMQAJ OEMQAJ Product" with this SKU's 18 cores divided across 10+8 core clusters, clocked at 3.9 GHz. It scored 2,541 points in the single-core test, so within 1% of the 20-core part, while the multi-core was 21,776 points, which is about 6% faster than the full-fat chip.

When you compare these numbers to other Windows-running SoCs on the Geekbench database, these multi-core numbers are among the best. The Snapdragon X2 Elite Extreme currently holds the record for the highest multi-core score with 25,075 points, so both SKUs of the RTX Spark lose to it. But when we switch to the AMD AI Max+ 395, which scores 20,609 on average in the same test, even the 18-core Nvidia chip is 5.6% faster here, and the 20-core variant is about 12% faster.

Compared to Apple's lineup, the 20-core RTX Spark is 3.3% faster than the 14-core variant of the M3 Max, which scores 22,385 points in the multi-core bench. Considering that this is pre-release hardware with unoptimized drivers, we can expect it to even beat the maxed-out M4 Pro and the base M4 Max (which hover around 25,000 points) when it officially launches. Moreover, the 18-core SKU's 21,776 points are equal to, if not more than, the base M4 Pro, which also nets around 21,700 points in the multi-core test.

That being said, the swan song falls apart when we bring up the single-core numbers. Every Apple Silicon chip from the base M3 onwards beats the 2,570 points the 20-core RTX Spark achieved in this run. Only the M2 Max with its 2,477 points is bested by the Nvidia offering, and that too by just 3.75%. Apple's current-gen base M5 is a whopping 41% faster here. On the Windows side, the Snapdragon X2 Elite and X2 Elite Extreme score roughly 3,000 points, but AMD and Intel's top-end models take a beating.

The Ryzen AI Max+ 395 only achieves 2,429 points in the single-core test, making the 20-core RTX Spark 5.8% faster, and the 18-core variant 4.6% faster. The same goes for the Ryzen AI 9 HX 370 because its 2,432 points are within the margin of error of the Strix Halo chip. Intel's highest single-core score is achieved by the Core Ultra 9 275HX at 2,467 points, making it 3% slower than the 18-core RTX Spark and 4.2% slower than the 20-core SKU. However, Intel's actual fastest chip, the Core Ultra X9 388H, scores 2,767 points, making it about 7% faster than even the 20-core RTX Spark.

All in all, this is a much better showing for Nvidia's upcoming laptop Superchip than previous leaks. It's clear that driver optimization is slowly unlocking the full potential of the silicon, and by the time RTX Spark is in the hands of reviewers, we expect it to compete fiercely with every Windows machine. Nvidia is working hard to ensure a proper launch with a commitment to making DRM and anti-cheat work on day one.

The mid-range CPU segment has become more competitive than ever, with both Intel and AMD refreshing their lineups and pushing aggressive pricing. In one corner, we have the Intel Core Ultra 7 270K Plus, an important reset for Intel that prioritizes class-leading productivity performance over efficiency, while also offering solid value. In the other corner is AMD's newly launched Ryzen 7 7700X3D, a more affordable, slightly lower-clocked alternative to the 7800X3D.

This faceoff is particularly interesting because the two chips take very different approaches despite costing about the same price. The Core Ultra 7 270K Plus packs 24 cores and excels in productivity workloads, while also delivering Intel's strongest gaming performance in years. The Ryzen 7 7700X3D, meanwhile, relies on AMD's proven 3D V-Cache technology to deliver gaming performance within 5% of the 7800X3D, although it trails significantly in single- and multi-threaded workloads.

While the Ryzen 7 7700X3D is aimed at gamers looking for X3D performance at a lower price, it faces stiff competition from Intel's aggressively -priced Core Ultra 7 270K Plus, as well as its own sibling, the Ryzen 7 7800X3D, which is often available for only a little more.

So, is AMD's gaming-focused approach enough to beat Intel's well-rounded Arrow Lake Refresh processor, or has Intel finally found the right balance of price and performance to reclaim the mid-range crown? In this six-round faceoff, we compare these two sub-$350 CPUs to find out which one comes out on top.

Features and Specifications: Intel Core Ultra 7 270K Plus vs AMD Ryzen 7 7700X3D

Taking a look at the technical specifications, it's clear that each chip adopts a distinct strategy for the mid-range market. While Intel has optimized its architecture for a higher core and thread count along with improved interconnect speeds, AMD continues to leverage its stacked cache to maintain gaming dominance on a long-lived platform.

The Intel Core Ultra 7 270K Plus is based on the original Arrow Lake-S family, utilizing the same microarchitecture that we saw on the 265K, built using TSMC’s 3nm process. It matches the flagship 285K with a 24-core configuration (8 Lion Cove P-cores and 16 Skymont E-cores) and operates with a 125W TDP. The chip can scale to a 250W Maximum Turbo Power (MTP), which is the maximum power limit that this CPU can consume for short periods when running at maximum turbo boost frequencies

Intel has also standardized several enthusiast-level performance tweaks for Arrow Lake Refresh, most notably a 900 MHz increase in die-to-die frequency and a 400 MHz fabric speed boost. It also officially supports faster DDR5-7200, along with 20 lanes of PCIe Gen 5 and 76MB of total cache. Possibly the only major concern with the Core Ultra 7 270K Plus is the LGA 1851 platform. With Intel's next-generation Nova Lake processors expected to launch later this year, the current platform will have a relatively short lifespan, restricting any future upgrade options.

The AMD Ryzen 7 7700X3D serves as a more affordable, lower-clocked variant of the 7800X3D, featuring the same eight Zen 4 cores with a peak boost clock of 4.5 GHz. Its primary architectural strength is its 104MB of total cache including 64MB of vertically stacked L3, which is specifically tuned to eliminate memory latency bottlenecks.

The chip is rated at a 120W TDP with a 162W Package Power Tracking (PPT) limit, though it frequently operates well below these levels during gaming. A major selling point for Team Red is motherboard flexibility as the 7700X3D can be used with existing AM5 motherboards (600 and 800-series chipsets), and AMD has committed to supporting the socket through at least 2029.

⭐ Winner: Intel Core Ultra 7 270K Plus

While AMD offers a more stable platform and superior gaming efficiency thanks to its large cache size, Intel’s hardware package is more comprehensive for the price. The 270K Plus offers triple the core count, significantly higher frequencies, and faster memory support.

Gaming Benchmarks and Performance: Intel Core Ultra 7 270K Plus vs Ryzen 7 9700X

To evaluate the gaming performance of these two chips, we conducted our testing at 1080p resolution using an Nvidia GeForce RTX 5090. We used this setup to prevent the GPU from becoming the bottleneck, allowing each CPU's gaming performance to shine through. By removing the graphics card as a limiting factor, we can see exactly how the 7700X3D’s 3D V-Cache stacks up against the higher core count and clock speeds of the 270K Plus.

We used identical systems for testing. You can read more about our full system configuration in our Ryzen 7 7700X3D review and Core Ultra 7 270K Plus review.

Across a comprehensive 16-game geomean, the AMD Ryzen 7 7700X3D holds its position as the superior gaming chip, delivering a 174.3 FPS average. This represents a 5.3% performance lead over the 270K Plus, which produced a 165.6 FPS average. While Intel's latest flagship has significantly narrowed the gap compared to the original Arrow Lake silicon, AMD’s 3D V-cache architecture continues to provide a higher performance in the majority of tested titles.

Even when examining the 1% low frame rates, which is the perceived smoothness while gaming, the 7700X3D maintains a slight edge with a 118 FPS average compared to the 115 FPS from the 270K Plus.

AMD's stacked cache gives the 7700X3D a clear edge in games that are sensitive to memory latency. For instance, in Minecraft RT, the 7700X3D delivered 144.5 FPS average, dwarfing the 89.7 FPS produced by the 270K Plus for a massive 61.1% performance advantage. We see similarly dominant leads for AMD in F1 2024, where it leads by 30.4% (201.8 vs. 154.7 FPS), and in Final Fantasy XIV, where it holds a 21.7% lead (177.7 vs. 146.0 FPS). Even in esports titles like Counter-Strike 2, the 7700X3D maintains a comfortable 7.3% lead with 707.9 FPS over Intel’s 660.0 FPS.

Intel’s Core Ultra 7 270K Plus is no slouch, however, as it leverages its higher boost frequencies and the new iBOT runtime translation layer to take the lead in several games. Intel manages a win in Hogwarts Legacy, producing 135.3 FPS for a 9.1% advantage over AMD’s 124.0 FPS, a result directly attributed to iBOT. Team Blue also edges out victories in The Last of Us Part 1 with 175.6 FPS (6% lead) and Spider-Man 2 with 207.2 FPS (5% lead).

In many other modern titles, the two processors are essentially locked in a dead heat. For example, in DOOM: The Dark Ages, the 270K Plus delivers 200.4 FPS while the 7700X3D is right beside it at 200.0 FPS. The story is similar in Flight Simulator 24, where Intel's 120.5 FPS and AMD's 114.9 FPS result in a negligible difference during actual gameplay. These results indicate that while the 7700X3D is the more consistent gaming choice, especially in cache-heavy titles, the 270K Plus is a highly competitive gaming chip that delivers impressive results for its $300 price tag.

The biggest difference between these two architectures is noticeable when analyzing power consumption and efficiency metrics. Based on our 16-game CPU power geomean, the 7700X3D is clearly way more efficient, drawing an average of just 60.9 watts while gaming. In comparison, the 270K Plus consumed 107.3 watts on average, which is a significant 76.2% increase in power. This results in a massive gap in gaming efficiency where the 7700X3D delivers 2.86 FPS per watt, making it roughly 85.7% more efficient than the 270K Plus, which trails at 1.54 FPS per watt.

Intel does take the lead when it comes to raw frequency, with the 270K Plus maintaining an impressive 5,247 MHz average clock speed across our gaming suite. This is over 700 MHz faster than the 7700X3D, which averaged 4,505 MHz. Despite the much higher clocks and power draw of the Intel chip, its thermal management remains surprisingly competitive when paired with the right cooling solution. The 270K Plus averaged 59°C during gaming, though it still runs warmer than the 7700X3D, which stays at a cooler 55°C average.

When factoring in the cost of the silicon, the competition for the best bang for your buck is incredibly tight. The Ryzen 7 7700X3D provides a value of 0.53 FPS per dollar, narrowly edging out the Core Ultra 7 270K Plus, which sits at 0.52 FPS per dollar.

⭐ Winner: AMD Ryzen 7 7700X3D

The 7700X3D’s overall higher average frame rates and excellent efficiency makes it the more attractive option for a dedicated gaming build. Since raw gaming performance, efficiency, and thermals all favor Team Red in this category, the Ryzen 7 7700X3D takes the win for this round.

Productivity Performance: Intel Core Ultra 7 270K Plus vs AMD Ryzen 7 7700X3D

AMD's X3D processors are widely known for their gaming prowess. However, they have historically struggled to maintain the same level of dominance in productivity performance. The Ryzen 7 7700X3D is no exception, as its cache-focused architecture and lower clock speeds result in relatively weak single- and multi-threaded results.

In contrast, Intel has managed to maintain its standing in the productivity segment even when its gaming performance faced challenges. The 270K Plus is a productivity workhorse that is capable of delivering excellent productivity performance making it a standout choice for users who need a balanced system for both work and play.

In multi-threaded workloads, the performance gap between these two processors is pretty massive, primarily driven by Intel's higher core count advantage. Across our multi-threaded performance ranking geomean, the Intel chip delivers a score of 626 compared to the 272 produced by the 7700X3D, representing a staggering 130% performance lead for Team Blue. This level of parallel processing power places the 270K Plus in a completely different performance tier, making it more comparable to much more expensive flagship processors.

Individual benchmarks further highlight this lopsided victory for Intel across various professional tasks. In Cinebench 2024’s multi-core test, the 270K Plus scores 2,509 points, which is roughly 135.6% faster than the 1,065 points achieved by the 7700X3D. This trend continues in POV-Ray, where Intel leads by approximately 169.5% (15,697 vs. 5,823 PPS), and in V-Ray 6, where it maintains a massive 129.5% advantage (45,016 vs. 19,615). Even in intensive video encoding tasks via HandBrake x265, the Intel chip more than doubles the performance of its AMD rival, delivering 29.9 FPS compared to just 14.3 FPS for the 7700X3D. These results demonstrate that for heavy rendering or data-crunching workloads, Intel is the undisputed leader in this price bracket.

A similar trend can be seen when examining single-threaded performance, a crucial metric for general system responsiveness and applications that do not scale across multiple cores. In our single-threaded geomean, the 270K Plus scores 278 points, which is 42.5% faster than the 195 achieved by the 7700X3D. Specific tests like Cinebench 2024’s single-core benchmark show the 270K Plus maintaining a 39.7% lead over the 7700X3D (145 vs. 103.8 points), while the lead grows to a staggering 82.6% in POV-Ray’s single-core test (1,138 vs. 623 PPS). Even in audio encoding, the Intel chip finishes the Lame Extended task in 68.61 seconds, whereas the 7700X3D trails at 91.53 seconds.

⭐Winner: Intel Core Ultra 7 270K Plus

Ultimately, the 270K Plus simply blows the competition out of the water when it comes to productivity. While the 7700X3D is a highly efficient and specialized CPU for gaming, it cannot match the raw horsepower that Intel offers. For any user whose daily routine involves video editing, 3D rendering, or heavy multitasking, the 270K Plus is the obvious choice and should be the clear favorite for a multi-purpose system.

Overclocking: Intel Core Ultra 7 270K Plus vs AMD Ryzen 7 7700X3D

While both manufacturers provide tools to squeeze extra performance from their silicon, the Intel Core Ultra 7 270K Plus is built as an enthusiast-friendly, fully unlocked processor. The AMD Ryzen 7 7700X3D, on the other hand, is a more restricted processor designed primarily for out-of-the-box gaming efficiency.

For the Intel Core Ultra 7 270K Plus, overclocking is a centerpiece of the experience. As a fully unlocked K-series SKU, it offers users granular control over per-core voltages, power limits, and clock speeds via an unlocked multiplier. While these changes can be done by entering the BIOS, one can also download the Intel XTU (Extreme Tuning Utility) software to fine tune the CPU.

One of the most significant changes seen on Arrow Lake Refresh is that Intel has standardized several high-end performance tweaks including the 900MHz die-to-die frequency bump and the 400MHz fabric speed increase. This means users receive enthusiast-level interconnect performance without necessarily needing a premium Z-series motherboard. That said, Z890 boards are still necessary as they offer sophisticated tools for manual tuning and pushing the CPU to its limits.

In contrast, the AMD Ryzen 7 7700X3D follows the same path as previous Zen 4 X3D processors including a locked multiplier, which prevents traditional manual overclocking. Users are instead limited to automated and semi-automated features like Precision Boost Overdrive 2 (PBO2) and Curve Optimizer. PBO2 allows the CPU to dynamically adjust its frequencies based on available power and thermal headroom, while Curve Optimizer enables more advanced fine-tune voltage offsets for each of the eight Zen 4 cores.

While these tools can lead to sustained higher boost clocks, the sensitive nature of the 3D V-Cache stack leads to thermal challenges that limit frequency headroom. Thus, overclocking gains on the 7700X3D are often minimal compared to the flexibility offered by the Intel chip.

⭐Winner: Intel Core Ultra 7 270K Plus

The 270K Plus is a far more overclocking-friendly product. It offers a vast suite of features that AMD simply cannot match due to its architectural restrictions and locked multiplier.

Power Consumption, Efficiency, and Cooling: Intel Core Ultra 7 270K Plus vs AMD Ryzen 7 7700X3D

This is a crucial section of the faceoff as it highlights the most significant architectural divide between these two processors. While the 270K Plus prioritizes raw performance throughput, the 7700X3D focuses on extreme efficiency. This is immediately evident in their official power ratings where Intel specifies a 125W TDP with a massive 250W Maximum Turbo Power (MTP), while AMD utilizes a 120W TDP with a 162W Package Power Tracking (PPT) limit.

In synthethic multi-threaded workloads, the difference is quite evident. While running HandBrake x265 10-bit encode, the 270K Plus consumes an average of 226W, which is more than triple the 72W required by the Ryzen 7 7700X3D. A similar trend appears in VP9 encoding, where Intel draws 184W compared to AMD’s efficient 65W. Even in lighter tasks like single-threaded y-cruncher AVX workloads, the Intel chip requires 55W while the AMD chip stays at a modest 32W. Idle power consumption also favors Team Red where the 7700X3D idles at 19W and draws 22W during YouTube playback, whereas the 270K Plus sits higher at 29W and 38W, respectively.

When we translate these power figures into efficiency metrics, AMD’s lead tends to stay ahead in most traditional benchmarks. In Cinebench 2024, the 7700X3D produces 14.4 points per watt, significantly outperforming the 10.4 points per watt from the 270K Plus. In HandBrake x265, AMD achieves a superior efficiency rating of 5W per FPS , while Intel trails at 7.56W per FPS.

However, the efficiency scatter plots provide a more nuanced look at how performance scaling impacts total energy consumption. In the Blender Classroom scatterplot, the 270K Plus delivers a much higher performance of roughly 135 samples per minute, but it consumes over 33,000 kJ of task energy. In contrast, the 7700X3D finishes with only around 60 samples per minute but uses significantly less total energy at roughly 4,500 kJ.

The HandBrake x265 scatterplot shows a similar trade-off where Intel reaches 30 FPS but consumes over 55 kJ, while AMD provides 14 FPS at roughly 36 kJ. Interestingly, the Linpack scatterplot shows the 270K Plus reaching over 850 GFLOPS with roughly 11WHr of energy, making it slightly more efficient in terms of performance-per-energy than the 7700X3D, which delivers roughly 340 GFLOPS for 10Whr.

From a cooling perspective, the 250W MTP on the 270K Plus means that it requires a robust cooling solution, like a 360mm AIO, to avoid thermal throttling during extended all-core loads. The 7700X3D is far easier to manage, remaining remarkably efficient and manageable with a mid-range air cooler. While the 270K Plus delivers class leading performance, it does so by significantly compromising efficiency in heavy workloads compared to AMD.

⭐Winner: AMD Ryzen 7 7700X3D

Pricing: Intel Core Ultra 7 270K Plus vs AMD Ryzen 7 7700X3D

The Intel Core Ultra 7 270K Plus was initially introduced with a justified $300 price tag, but you'll now find it between $300 and $320. The AMD Ryzen 7 7700X3D made its debut just last month and carries a $330 MSRP, placing it right under the 7800X3D. Essentially, there is not a major difference when it comes to the chips themselves, however, the true financial divide becomes apparent when we look at the total platform cost.

The Core Ultra 7 270K Plus demands a more substantial investment to reach its full potential. While one can opt for a B860 motherboard, a compatible Z890 motherboard is necessary in order to access premium overclocking and tuning features. These typically start around $150-$200, and can go as high as $600 packed with premium features. To handle the processor′s 250W Maximum Turbo Power(MTP) during heavy productivity tasks, a high−end 360mm AIO liquid cooler ($100-$150) or a dual-tower air-cooler ($50-$100) is recommended.

In contrast, the AMD Ryzen 7 7700X3D offers a much more budget-friendly entry point. A solid B650 motherboard can be found for around $120-$150 and the chip can be effectively cooled with a modest $50 air cooler.

Unfortunately, both platforms are held back by the ongoing global shortage of memory, which has caused prices to shoot up significantly. Since both CPUs require DDR5 RAM, builders are stuck in a RAMpocalypse where a basic 32GB kit costs over $400. This extra cost is bad news for both sides as it cancels out any price advantage.

Overall, the lower upfront platform cost makes the 7700X3D a slightly more attractive option for users prioritizing their budget. Furthermore, the AM5 socket offers a clear upgrade path through at least 2029, ensuring that your motherboard investment remains viable for future CPU generations. That is not the case with Intel, as the LGA 1851 socket is expected to be replaced once the next-generation of Nova Lake CPUs arrive.

⭐ Winner: AMD Ryzen 7 7700X3D

Ultimately, while both the chips are available at a very similar price range, the AMD Ryzen 7 7700X3D is the smarter financial play for a majority of users. Its significantly lower platform costs, superior gaming efficiency, and guaranteed platform longevity provide a level of value that the more power-hungry Intel refresh cannot quite match.

Bottom Line: Intel Core Ultra 7 270K Plus vs AMD Ryzen 7 7700X3D

The Ryzen 7 7700X3D and Core Ultra 7 270K Plus are designed for very different users, even though they sit in roughly the same price range. Like previous X3D chips, AMD's latest processor is built with gaming in mind, using its large 3D V-Cache to deliver excellent frame rates while keeping power consumption impressively low. Intel, on the other hand, combines a 24-core design with higher clock speeds and platform improvements to create one of the strongest productivity processors in its class without giving up much gaming performance.

The final score is tied at three rounds apiece. AMD comfortably wins gaming performance, power efficiency, and platform value, three factors that matter a lot for buyers shopping in this price segment. The AM5 platform also provides a significantly longer upgrade path, allowing users to drop in future processors without replacing their motherboard.

Intel, however, deserves recognition for what it has accomplished with the 270K Plus. It delivers outstanding single- and multi-threaded performance, offers a fully unlocked overclocking experience, and narrows the gaming gap to just a few percentage points in many modern titles. If your workload includes video editing, software development, 3D rendering, or other heavily threaded applications, the 270K Plus is the obvious choice and justifies its higher power draw.

For everyone else, the Ryzen 7 7700X3D is the more compelling CPU. It offers faster gaming performance, exceptional efficiency, lower platform costs, and a future-proof AM5 ecosystem. Unless your workload regularly extends beyond gaming into demanding productivity applications, AMD's latest X3D chip is the easier processor to recommend.

⭐ Winner: Tie

More CPU Faceoffs

• AMD Ryzen 9 9950X3D2 vs Ryzen 9 9950X3D
• Intel Core i5-14400 vs AMD Ryzen 5 7600X
• AMD Ryzen 7 9850X3D vs Intel Core i9-14900K
• AMD Ryzen 7 9850X3D vs Ryzen 7 9800X3D

From the outside, the Phantom Drive, created by Rootkit Labs, appears like a boring old USB. It has that classic, cheap plastic-feeling look to it that would never make you second-guess its intention; when you plug it in, it shows up as just an 8GB drive. However, under the hood, the Phantomdrive is powered by a microSD card that only unlocks itself when you specifically create a new text file, after which it mounts the secret partition, which comes with AES-256 encryption. It's made for situations and places where you might be forced to reveal the contents of a storage device.

The Phantomdrive is built using the CH569 microcontroller that has a bunch of different hardware blocks — this project uses the USB3, SD/eMMC, and the AES block. There are no NAND chips directly onboard the custom PCB because of the ongoing component crisis. Instead, the Phantomdrive relies on a simple microSD card (that you provide; it doesn't come with one) for both its 8GB decoy partition and the hidden partition. The creator does say they'll make a version with eMMC modules once prices come down.

Of course, if someone were to open up the drive, they'd be able to physically rip the SD card out. However, it still wouldn't be a dead giveaway since many cheap USBs are powered by microSD cards, and SD card adapters are also a thing. Anyhow, the finished drive looks pretty simple from the inside; it comprises a USB port, two buck supplies, a button for firmware update, the microSD card, UART test points, and some support components. The custom firmware now comes into play.

The name of the game is interception. The firmware doesn't know it's being used to handle file transfers; it just reads raw packets of data being sent over the USB bus as is. It's coded to always look for a "password" string, so as soon as you create the unlock.txt file with "password:xyz" inside, it detects the "xyz" bit and copies it over to SRAM, while overwriting the raw data with a bunch of zeroes before it's written to the SD card. The operating system thinks it's created the new file, but the firmware intercepted it midway and sent essentially nothing to the physical storage cells.

The firmware uses the unique internal hardware ID of the microcontroller as the salt, combines it with the password you just typed, and runs the pair through PBKDF2-HMAC-SHA-256 hashes 100,000 times (you can also select 600,000 times) to create a key. Each run induces an intentional 2-3 second delay to deter brute-force attacks that would try to crack the password. The key is used to unlock and lock the hidden partition, and since it's tied to the silicon itself, putting the SD card into another Phantom Drive will not unlock it.

Even if you try to access it directly through a microSD card slot, it will stay hidden, and the 8GB decoy will actually appear like a broken partition. The creator has also provided the option to recompile the firmware on your own to spoof specific vendor IDs, so that the USB appears like it's from different brands. You can make the Phantomdrive pretend to be a cheap 16GB Kingston Datatraveler at the hardware protocol level, at which point no scanning software will flag it as anything more than a cheap USB.

There are Linux-based C tools you can use in case of emergency, however, to recover your data, such as physical damage to the board. The transfer speeds are also very limited because the project is built over USB 2 — 20 MB/s reads and 9 MB/s writes in AES-CTR mode. The speeds fall to 10 MB/s reads and 6 MB/s writes in AES-XTS, but at least you have the freedom to choose between the two according to your needs.

You can buy the Phantomdrive from Rootkit Labs for $50, but you can also build your own, since this is an open-source project. The prebuilt device comes with two extra shells because the pins wear off and break quickly. Otherwise, the GitHub repo provides all the resources, including the PCB layout files and schematics, the BOM for every part, and the custom firmware, alongside its related tools.

The creator of Phantomdrive, Ryan Walker, explains the goal behind the project as both utilitarian and political. It's built using entirely open-source software and with inexpensive, easily accessible components. They wanted to create something that could "slip past authoritarian government representatives, corrupt police, and anyone else that doesn't respect basis encryption rights."

Micron used to produce some of the best RAM available to mainstream users through its Crucial brand. Despite the company's exit from the consumer market through its now-defunct Crucial brand, Micron had pledged to continue honoring warranties on existing Crucial products. A recent case on Reddit, however, highlights ongoing challenges with Micron's warranty process, where it reportedly offered a customer a ludicrous payout for an RMA. Only after further communication did Micron provide an improved offer. Another user also chimed in with a similar story about an SSD return with Micron. We have reached out to Micron for comment.

Redditor Key_Tailor6948, who shared the email exchange with Tom's Hardware, recently shared their RMA experience with Micron's warranty process after encountering a failing memory module from a 48GB (2x24GB) DDR5 memory kit. While the Redditor did not specify the exact model of the memory kit, Crucial used to sell two such kits in this capacity: the Crucial Pro DDR5-6000 C48 and the Crucial Pro DDR5-5600 C46. The last known pricing for the memory kits was $489.99 and $459.99, respectively, during the memory shortage.

According to the Redditor, Micron allegedly said in an email that the company is no longer offering replacement products for warranty claims. Instead, Micron purportedly stated it would provide a refund based on the original purchase price, including taxes where applicable. However, the proposed refund amount was $241.86.

The major issue at hand is that the reimbursement Micron offered was based on the cost of the 48GB memory kit purchased before the recent memory shortage and the resulting price hikes. Nowadays, the most affordable 48GB DDR5 memory kit available on the market, such as a DDR5-6400 C32 one, which is faster than Micron’s discontinued offerings, starts at a staggering $649.99. The refund Micron proposed amounts to just 37% of the current market value for a somewhat comparable product. The Redditor's math about being 17% of market value and estimating similar-capacity memory kits going for “roughly $1,400" is a bit off based on verifiable pricing, but the proposed payment was still well short of current market value.

Beyond the disappointment over the low reimbursement, the Redditor expressed frustration that they had to send back the entire 48GB memory kit, even though only one of the two modules was faulty. This meant parting with a memory module that was still fully functional. Unfortunately, this is a common industry practice. Most manufacturers require the return of the full memory kit for warranty claims. It is exceedingly rare for companies to allow returns or replacements of only a single memory module from a multi-module kit, regardless of whether the other module is working perfectly. (EDIT 8/11/2026 : Micron's warranty states that it will either replace a product under warranty or refund the current market value of the product or the amount of the original purchase cost, whichever is lower. Therefore, Micron's offer is within the bounds of its warranty.)

The Redditor was also not happy about covering the shipping costs to return the defective memory module. This is another policy that varies widely among brands: some manufacturers cover all shipping fees, others split the cost, so you cannot really single out Micron in that aspect.

It would seem that Micron subsequently offered a more favorable resolution. The company agreed to send the user three Crucial Pro DDR5-6400 C32 16GB (CP16G64C32U5B) memory modules as a replacement. This solution matches the original capacity of 48GB and also delivers better performance than the user's original memory kit. The user accepted the offer.

Meanwhile, Crucial SSD owners are seemingly experiencing similar RMA practices from Micron. A Reddit user named LeanaIsTheBest highlighted in the same thread that Micron is allegedly refusing to provide direct replacements for a failing 4TB SSD. Instead, the company ostensibly offered a gift card equivalent to the original purchase price of the 4TB SSD as compensation. However, the company provided an alternate option to receive the same amount in cash through PayPal or Venmo, with the condition that the user pays the 6% transaction fee.

Micron wound down its consumer business back in December 2025. It is reasonable to think the chipmaker has reshaped its manufacturing factories to cater to more profitable markets, such as data centers and the AI sector. Micron probably is not maintaining inventory or production lines for consumer products, so direct replacements are unlikely. For a company that reported revenue of $41.46 billion in its last earnings call, it isn't a great look for Micron to not reimburse the replacement cost for loyal customers who have purchased its Crucial products.

The owner of what is claimed to be “the original engineering copy of the Intel 8080 rubylith mask” has shared a social media shout-out, looking for a skilled restorer. From the shared photograph, this important artifact from the history of computing looks like it would benefit from remounting and reframing. Hopefully, the hand‑crafted, large‑scale sheet of red film (Rubylith) has remained in good condition under glass, though. The framed artifact is likely genuine and original, as its current owner is thought to be related to Internet Hall of Fame inductee Dan Lynch, a pivotal figure in the early Internet’s success.

In you unfurl the tweet above, you can see Tom Lynch, a self-described arborist and the owner of an AI infrastructure startup, standing behind the framed chip artwork. Folks comment on the Rubylith looking just like the one that the Intel Trinity: Andy Grove, Robert Noyce, and Gordon Moore, were photographed beside in 1978.

Rubylith is simply an adhesive, peelable red film that found favor in graphic arts and chip design before designers went digital. Engineers could cut away parts of the film to define where light would be exposed during photolithography processes. Rubyliths would often be marked with pens, tape, and overlays, especially as designs were iterated and refined ahead of tape-out.

The Intel 8080’s primary architect was Federico Faggin, and it was manually drafted in the age before modern CAD. This human-scale draft is likely 100x magnification compared to the finished processor die. Even scaled this large, it would still be quite an intricate drawing as it had to map around 5,000 transistors, traces, etc., into an approximate 20- x 16-inch sheet. The commercial Intel 8080 release ended up being manufactured on a 6 μm silicon gate process. It originally ran at 2.0 MHz clocks, but later revisions would boast clock speeds up to 3.125 MHz.

Intel’s 8080 was an important 8-bit microprocessor for both the company and the advancement of personal computing. This 8-bit CPU was the one chosen for the Altair 8800, was the original target architecture for the CP/M operating system, and would be a big influence on the later x86 architecture. We also recently wrote how the 8080 bottlenecked the Space Invaders arcade design, unintentionally resulting in the space shoot-em-up’s thrilling increase in pacing as aliens were zapped from the sky.

The AI-driven memory apocalypse has made computer parts so expensive that the average buyer has stopped building PCs. Multiple research firms have already said that, but we’re now seeing it in action in the market. Uniko’s Hardware shared on X four crazy combos of RTX 5090 GPUs being sold on the Taiwanese e-commerce platform PChome24h.

Some of the GPUs that come with a lot of "freebies" include the Asus ROG Astral GeForce RTX 5090, Gigabyte Aorus RTX 5090 Master, MSI GeForce RTX 5090 Gaming Trio OC, and a Gigabyte Aorus RTX 5090 AI Box external graphics card. These items are priced in NTD, but we’ll list them below, including their retail prices in dollars (NTD 32.25 = USD $1 exchange rate) and the included items.

As we can see, even though GPU prices are quite inflated, many of them come with enough freebies that you can start building one high-end system and three to five entry-level gaming rigs. They’re still priced relatively high compared to Amazon or Newegg in the U.S., though, but we have to keep in mind that Taiwan is a different market, so there will likely be variations in prices plus VAT.

While this might be an amusing piece of trivia for some people, it’s actually evidence of a disturbing trend in the PC building industry. Multiple research firms have been saying that the market will contract by double digits this year, with the entry-level PC market being hit the hardest, and that it’s not expected to bounce back until 2029. These combos are just evidence of that, at least in Taiwan.

That’s because sellers often put slow-moving products in a combo to help get their inventory moving. Given the number of motherboards, entry-to-mid-range GPUs, and other components included in these combos, it’s safe to say that these items aren’t selling and retailers are doing all these crazy deals just to free up capital.

As the AI-driven memory crisis has pushed up the prices of memory and graphics cards, many have already paused or delayed their purchases and upgrades. Only those with deep pockets who can afford to spend thousands of dollars on parts and combos like these are still buying PC parts.

A powerful processor is at the heart of every exceptional gaming PC, and nothing rivals the performance of AMD's Ryzen 7 9800X3D. If you are building a new system or updating an existing one, this Ryzen 7 9800X3D bundle, which includes an MSI B850 Gaming Plus WiFi motherboard and an MSI MAG Coreliquid A13 240 AIO liquid cooler, is for you. This curated package is available for $458.99, a massive $160 off the regular price of purchasing the components separately.

• Check out the Ryzen 7 9800X3D bundle deal on Newegg

The Ryzen 7 9800X3D is the best gaming processor that money can buy right now. With an octa-core, 16-thread design and a massive 96MB of L3 cache, the Ryzen 7 9800X3D not only delivers top-notch gaming performance, but it also excels in productivity tasks. Whether you are assembling a system for gaming or serious work, the Ryzen 7 9800X3D provides a rock-solid platform for your build.

While the exclusive bundle does not include any DDR5 memory, it is not a big drawback. Thanks to its design, the Ryzen 7 9800X3D maintains stellar gaming performance regardless of whether you pair it with slow memory or even a single memory module. Therefore, you do not need to spend a fortune on premium memory, which is a big plus in this ongoing memory shortage.

On the motherboard side, Newegg's premium bundle includes a feature-rich MSI B860 Gaming Plus WiFi motherboard. It boasts a 12+2+1-phase power delivery subsystem that maximizes the Ryzen 7 9800X3D's performance while also offering ample headroom for overclocking for those who want to extract extra performance.

The B860 Gaming Plus WiFi comes equipped with four DDR5 memory slots that can accommodate up to 256GB of memory with blazing data rates up to DDR5-8200. Meanwhile, storage options arrive in the shape of three high-speed M.2 slots (one at PCIe 5.0 and two at PCIe 4.0) and four standard SATA III ports. The motherboard also has a PCIe 5.0 x16 expansion slot for pairing the Ryzen 7 9800X3D with one of the best graphics cards.

While the Ryzen 7 9800X3D is not a processor that runs very hot, you can never have too much cooling. The MSI MAG Coreliquid A13 240, a solid 240mm AIO liquid cooler, uses a high-performance pump, a thick radiator with wider water channels for a higher flow rate, and two pre-installed 120mm PWM fans that are quiet and performant.

The Ryzen 7 9800X3D bundle will be available on Newegg for the entire month. The massive 26% discount translates to real, tangible savings, bringing the bundle down to just $458.99. It is a rare opportunity to secure the best gaming processor with a very decent motherboard and 240mm AIO liquid cooler at a price that you probably will not find elsewhere. Given the popularity of the Ryzen 7 9800X3D, the bundle will likely sell out very fast, so do not think too much before swiping your credit card.

If you're looking for more savings, check out our Best PC Hardware deals for a range of products, or dive deeper into our specialized SSD and Storage Deals, Hard Drive Deals, Gaming Monitor Deals, Graphics Card Deals, gaming chair, or CPU Deals pages.

The best RAM on the market generally utilizes integrated circuits (ICs) from industry giants such as SK hynix, Samsung, or Micron. Recently, however, Chinese chipmaker ChangXin Memory Technologies (CXMT) has emerged as a strong competitor amid the ongoing global memory shortage. In Colorful's latest overclocking exhibition, CXMT's ICs achieved speeds up to DDR5-8800 on AMD's X870E platform, keeping pace with the best overclocking ICs, such as SK hynix's A-die and M-die.

Many often regard CXMT as an underdog in the memory industry, particularly when compared to the "Big Three." Founded only a decade ago, the ambitious Chinese chipmaker eventually became China's largest domestic memory manufacturer and is now the fourth-largest memory maker globally.

The recent memory shortage has accelerated CXMT’s rise to fame. The chipmaker has been silently gaining ground in the memory market. Its momentum has been especially evident in the DDR5 segment despite the company having only unveiled its first DDR5 ICs in November 2025. Many top Chinese memory brands, including Asgard, Colorful, Gloway, and KingBank, are widely using CXMT ICs, and even Corsair has done so to a certain extent.

Colorful showed off an incredible overclock with its iGame Shadow II DDR5-6000 C34 memory kits, pushing both the 32GB (2x16GB) and 48GB (2x24GB) configurations far beyond their rated speeds. These kits reached DDR5-8812 and DDR5-8817, respectively, on Colorful's iGame X870E Vulcan W OC motherboard. Unfortunately, the screenshots provided by Colorful were too blurry to discern detailed specifications such as memory timings or DRAM voltage. For context, some high-end DDR5-8800 memory kits typically operate at timings around 42-55-55-140 with a voltage of 1.45V.

The noteworthy aspect is that Colorful achieved the overclocking feat on 16GB memory modules. Most, if not all, DDR5-8600 and faster memory kits currently on the market come with 24GB memory modules. The reason is that many use SK hynix’s 24Gb (3GB) M-die ICs, produced on the chipmaker's advanced 1a-nm (~14nm) process node with Extreme Ultraviolet (EUV) lithography. These M-die ICs are the go-to choice for memory vendors thanks to their superior density, power efficiency, and ability to scale to higher frequencies.

In comparison, CXMT’s accomplishments are notable when considering the technological challenges. Due to U.S. export restrictions on China, CXMT does not have access to cutting-edge EUV equipment, so its 16nm node, which it uses for DDR5 ICs, leverages Deep Ultraviolet (DUV) machines instead. CXMT’s 16nm process is roughly in the same alley as the 1z nm (~14–15nm) nodes previously used by the Big Three, so CXMT is easily a few years behind the top-tier process nodes.

The reality is that CXMT will have to make do with what it has. The chipmaker is already developing a 1a-nm node but on DUV. It is far from prime time, though.

In addition to seeing CXMT ICs in mainstream memory kits, companies such as HP, Asus, and Acer have already started selling consumer devices, such as laptops or desktops with CXMT memory. Even Apple reportedly wants to buy memory chips from CXMT.

CXMT recently went public via IPO and saw its stock price skyrocket by 466%, raising over $8.6 billion. Without access to EUV machinery, CXMT will focus primarily on memory production and is rapidly building factories across China. The ambitious Chinese chipmaker hopes to steal 30% of the market share from the Big Three by 2030. It is just a matter of time before we start seeing CXMT ICs in all the prominent consumer memory brands.

We tested gaming performance with both NVMe and SATA SSDs across 11 games to determine if upgrading to a faster drive is worth it. Solid-state drives (SSDs) have been a staple of PC gaming for years, but their importance grew significantly with the launch of the PlayStation 5 and Xbox Series X|S in 2020. Both consoles feature high-speed NVMe SSDs and storage architectures designed to fully leverage their capabilities, enabling faster asset streaming and reducing traditional storage bottlenecks.

As a result, game developers have increasingly developed their titles for SSDs, with many modern games now listing them as a minimum system requirement. But how exactly do games use SSDs during gameplay, and does upgrading to a faster SSD actually improve gaming performance? In this article, we'll examine SSD activity in modern games and determine whether higher storage speeds translate into a better gaming experience. Given the surge in SSD prices, will a slower drive deliver comparable gaming performance to high-end alternatives?

Load Times

Load Times

While this article focuses primarily on SSD activity and performance during gameplay, loading times remain one of the most significant advantages of solid-state storage. It's well known that games can take several minutes to load from an HDD, whereas modern SSDs typically reduce load times to just a few seconds.

The gap between SSDs is much smaller. In some titles, PCIe NVMe drives can load games several times faster than SATA SSDs, but in many cases, the difference is relatively small. Naturally, different generations of PCIe SSDs have even smaller differences, with flagship drives often delivering only marginal improvements to loading times despite their substantially higher sequential throughput.

Disk I/O Operations in Games

Disk I/O Operations in Games

Before examining the performance results, it's important to understand the types of disk I/O operations that games typically generate. Two key metrics are block size and queue depth. The block size of an I/O operation defines the size of the chunk of data a game will read at one time, while queue depth is the number of pending I/O requests that a storage device has queued at any one time.

During gameplay, the vast majority of storage activity consists of random read operations as game assets are streamed from storage into memory. Sequential reads are less common, and write activity is generally minimal.

Microsoft provides several storage optimization recommendations for developers in its DirectStorage documentation. Although these guidelines are intended for DirectStorage-enabled titles, many of the underlying principles apply to games in general, even if they do not use DirectStorage. Among Microsoft's recommendations are the use of block sizes of at least 32k and sufficiently high queue depths to maximize storage throughput.

Modern games typically employ 32k or 64k random I/O requests. However, despite Microsoft's recommendations, most games continue to operate at relatively low queue depths.

Performance Comparison

Performance Comparison

Now we will examine SSD activity and compare performance across different SSD generations. The games included in our testing were selected because they either exhibited higher storage bandwidth than usual, had greater peak or average SSD utilization, had support for DirectStorage, or because they are popular titles that many gamers would be interested in seeing analyzed from a storage-performance perspective.

To place as much stress as possible on each SSD, all testing was conducted with 16GB of system memory. Using a smaller memory capacity reduces the amount of game data that can be held in RAM, forcing games to rely more heavily on the SSD than they would in a 64GB configuration. It also better reflects the configurations found in many mainstream gaming PCs. That said, we observed little difference in SSD activity when comparing 16GB and 64GB memory configurations.

When interpreting the SSD activity data below, it is important to understand that 100% utilization does not necessarily mean an SSD is reaching its maximum rated bandwidth. For example, a PCIe 4.0 SSD reporting 100% utilization is unlikely to be delivering its full 7.5 GB/s of sequential read throughput. In gaming workloads, SSDs can become fully utilized while transferring data at significantly lower bandwidths.

This is because the speeds advertised by SSD manufacturers are based on sequential read performance, whereas games primarily rely on random reads. While random-read throughput can approach sequential-read speeds under certain conditions, doing so requires sufficiently large block sizes and high queue depths. Although modern games often use relatively large block sizes, they typically do not use the high queue depths needed to fully saturate an SSD's theoretical bandwidth.

We have also chosen not to test any traditional hard drive disks. There are many documented examples of modern games performing poorly, taking an unbearably long time to load, or having visual issues when installing the game on an HDD. There are some games where this may not be an issue, but we still do not recommend playing modern games on an HDD.

The testing was conducted in gameplay. No built-in benchmarks were used. We found that SSD activity increases with higher settings, so that is where we focused our testing. We set each game to its absolute maximum settings.

We used HWiNFO to measure SSD activity and set it to the lowest polling rate possible to capture activity accurately without incurring overhead.

Test system

• MSI GeForce RTX 5090 Vanguard SOC
• Ryzen 7 9800X3D
• 16GB G.Skill Trident Z5 DDR5 6000 MHz CL30
• Crucial T700 Gen5 SSD
• Sabrent Rocket 4 Plus-G Gen4 SSD
• Adata XPG SX8200 Pro Gen3 SSD
• Samsung 870 EVO SATA III SSD
• Asus ROG Strix B850-F Gaming WiFi
• Corsair Nautilus 360 RS AIO Cooler
• MSI MPG Ai1600TS PCIe 5 PSU
• HAGS enabled
• Windows 11 25H2 (Build 26200.8737)
• Nvidia driver 610.62

Forspoken

Forspoken

Forspoken was the first game to support DirectStorage on PC. DirectStorage is a feature that aims to allow games to make full use of high-speed storage devices while reducing the CPU overhead associated with making reads from disk.

Looking at the table below, you can see that it does exhibit a fairly high peak and average bandwidth. In fact, it’s the highest average bandwidth we have seen in any game we have tested. Average utilization is also quite high, and the game hits 100% SSD utilization frequently.

How does this translate to performance?

While there is no difference in performance between the PCIe NVMe SSDs, you can see that having a PCIe SSD can give you a large boost in the 1% lows compared to a SATA SSD. We see a 39% increase in 1% lows at 4K, a 51% increase at 1440p, and a 22% increase at 1080p.

The higher peak and average bandwidth used by the PCIe drives resulted in higher 1% lows in this game, which makes for a smoother gameplay experience.

Alan Wake 2

Alan Wake 2

Alan Wake 2 makes extensive use of modern rendering technologies, including path tracing, mesh shaders, and RTX Mega Geometry. In a recent interview, Remedy explained that the game operates with a fixed memory budget, continuously streaming the assets required for the current scene from the SSD while evicting assets that are no longer needed. This approach keeps memory usage under control despite the game's highly detailed textures and geometry, making Alan Wake 2 one of the most storage-intensive titles available today.

The SSD activity increases with resolution in this game. We see an average of over 200 MB/s on each drive at 4K, which is rare for a game. We also see a whopping 5.1 GB/s of peak bandwidth on a PCIe 5.0 SSD at 4K.

How does this impact performance?

Unlike Forspoken, we do not see any difference in performance here for any of the drives tested.

Although the faster SSDs reach higher peak bandwidth, these peaks are only brief bursts rather than sustained transfers. In other words, the game is not continuously streaming multiple gigabytes of data every second. Instead, it transfers smaller amounts of data over very short intervals, resulting in high instantaneous bandwidth readings. As a result, the average bandwidth is the more meaningful metric, and it remains relatively similar across all of the SSDs we tested.

The average utilization on the SATA SSD can reach over 80% at 4K, but it is not enough to degrade its performance in comparison to the PCIe drives.

Ratchet and Clank: Rift Apart

Ratchet and Clank: Rift Apart

The disk bandwidth through this sequence is actually lower than the previous two games. This makes sense, as the game uses GPU decompression, which increases the effective bandwidth flowing through the PCIe bus beyond what CPU decompression would. However, this increase in effective bandwidth is not reflected in the data that is being read from disk.

When reading the performance charts below, it is important to note that PCIe SSDs reduce portal transition times by roughly 2-3 seconds compared to SATA SSDs. On SATA drives, Ratchet therefore spends more time within each portal sequence. As a result, average performance and 1% lows can generally be slightly higher on SATA SSDs, since there is less active rendering during these extended transition periods. However, there is one exception.

While the SATA SSD has an advantage at 1440p and 1080p due to the extended time spent inside of each portal, the 1% lows take a hit at 4K. The SATA drive does not handle the first portal transition in the sequence particularly well at 4K, and so the PCIe drives have about a 38% advantage in 1% lows.

This portal sequence is the most storage-intensive part of the game. In normal gameplay, the game uses significantly less bandwidth, and so there is no difference in performance at all between the SATA SSD and the PCIe SSDs.

Spider-Man 2

Spider-Man 2

Spider-Man 2 was the second game to use GPU decompression of assets on PC. This is the third DirectStorage title we have tested so far.

Since the game uses GPU decompression, the effective bandwidth after decompression will be considerably higher than what we see here, but none of the SSDs we tested are being stressed in this game.

Performance is virtually identical on all drives in this game.

Starfield

Starfield

Starfield is an open world game that constantly reads data from disk. It also frequently hits 100% disk utilization as you traverse the world.

In previous games, 100% SSD utilization was accompanied by much higher peak bandwidth. Here, the lower peak bandwidth suggests the game uses a very low queue depth for its read operations. As you move through the open world, it continuously streams data from the SSD in frequent, small transfers, keeping average bandwidth very low.

Once again, we see no improvement in performance going from the SATA SSD to any of the faster PCIe NVME SSDs.

DOOM: The Dark Ages

DOOM: The Dark Ages

DOOM: The Dark Ages was the first game to list an NVMe SSD as a requirement. Not only is it listed as such on the PC system requirements, but Bethesda has a support page that reiterates that installing the game on an NVMe SSD is required.

The developers at idSoftware mentioned that the game uses a proprietary ultra-fast continuous sector streaming tech that virtually eliminates load times between levels.

The drives we tested are hardly being stressed in DOOM: The Dark Ages.

Given the disk utilization figures, it comes as no surprise that all of the tested drives deliver identical gameplay performance.

Although this article focuses on in-game performance rather than loading times, we also found little difference in load times when using a SATA SSD. Loading screens were not noticeably longer, and we observed no visual artifacts or streaming issues while playing on the slower drive.

Based on our testing, it is unclear why Bethesda lists an NVMe SSD as a requirement. A SATA SSD provided an indistinguishable gameplay experience, with no measurable performance drawbacks or observable streaming issues.

Dead Space Remake

Dead Space Remake

Dead Space Remake made headlines when it launched with abhorrent performance and stuttering issues. But how much of that was due to streaming from slower storage? Does a faster SSD help performance?

Although average disk utilization and bandwidth are relatively low, the game frequently reaches 100% utilization. Yet even during these periods, peak bandwidth remains well below what PCIe SSDs can achieve under high-queue-depth random workloads. This indicates the game is issuing read requests at a very low queue depth, leaving much of the SSD's available performance untapped. Additionally, the difference in average bandwidth between the SSDs is minimal.

All drives provide the same performance. There is nothing to be gained from a faster SSD in this title.

Cyberpunk 2077

Cyberpunk 2077

We included Cyberpunk 2077 in our testing because the game features a large and seamless open world, and it is one of the most popular games ever released.

The average bandwidth is by far the lowest we have seen in the games tested so far. This is not entirely surprising. Although the game was released in the same year as the PlayStation 5 and Xbox Series X|S, development began well before those consoles were released. As a cross-generation title, it also needed to run on the previous generation of consoles, which relied on mechanical hard drives. As a result, the game's streaming system was designed to perform well on significantly slower storage devices.

There is no noticeable difference in performance across the drives tested.

The Elder Scrolls IV: Oblivion Remastered

The Elder Scrolls IV: Oblivion Remastered

The Elder Scrolls IV: Oblivion Remastered modernizes the look of the original game with improved visuals. Built on Unreal Engine 5, it utilizes Nanite virtualized geometry, Lumen global illumination, and virtual shadow maps. This is an open-world game, and our first UE5 title is tested in this article.

None of the SSDs we tested reached 100% peak utilization and average utilization is extremely low, too.

As is the case with most of the games we have tested, the SATA SSD holds its own and there is no difference in performance across all of the drives.

007 First Light

007 First Light

007 First Light is a recently released game that uses DirectStorage. In a recent interview, the developers at IO Interactive revealed that the real-time level streaming system in the Glacier engine was redesigned for 007 First Light. The new system uses dynamic loading and unloading to allow for seamless transitions between sequences.

It’s clear that this new streaming system does not place a heavy burden on storage devices.

Given the I/O activity in this game, it is not surprising to see that all SSDs we tested perform the same.

F1 25

F1 25

We included F1 25 because it exhibits higher peak bandwidth and utilization than most games. The game’s high-speed driving requires rapid streaming of trackside assets, as the environment changes quickly and the player covers large distances in very short periods of time.

The game exhibits higher peak and average bandwidth than many of the games we have tested, but it is still not enough to truly stress a SATA SSD.

Once more, all of the drives show identical performance.

What About the Visuals?

What About the Visuals?

Many readers may be wondering if playing on a SATA drive resulted in any visual anomalies, such as slower loading textures, additional pop-in, or other visual artifacts. We did not notice any such issues in the 11 games we tested. There are some games that exhibit pop-in, but that is consistent across all storage devices. All of the drives we used in our testing provided an identical visual experience.

The Verdict

The Verdict

Out of all the games tested, only two showed measurable performance differences between SATA and PCIe SSDs: Forspoken and Ratchet & Clank: Rift Apart.

In Forspoken, moving from a SATA SSD to any PCIe SSD – regardless of generation – resulted in a significant improvement in 1% lows.

Ratchet & Clank: Rift Apart showed more nuanced results. During the portal sequences at 1440p and 1080p, average framerate and 1% lows were slightly higher on the SATA SSD, as slower loading increased the time spent inside the portals. Since the portals contain less on-screen complexity, framerates are naturally higher within them. However, at 4K resolution, PCIe SSDs delivered better 1% lows. Outside of the portal sequence in normal gameplay, there was no measurable performance difference between any of the drives.

Across all other titles, PCIe SSDs provided no meaningful performance advantage, and there were no observable differences between PCIe generations. In the vast majority of games tested, a SATA SSD delivers comparable performance.

MORE: Best SSDs

MORE: Best External SSDs

MORE: Best SSD for the Steam Deck

To optimize how software runs on hardware, instruction latency analysis looks at the time it takes for low-level instructions to execute on a processor, either to optimize the architecture or to optimize applications to run on a particular architecture. One hardware researcher, Christopher Domas (@xoreaxeaxeax on GitHub), is taking a different approach with the CPU Deoptimization leaderboard, which looks not to make Assembly instructions run as fast as possible, but as slow as possible to measure the single instruction with the highest latency.

The winner here is fxrstor64, which took 62 seconds, or over 198 billion cycles, to complete. This instruction restores the state of registers used for SIMD calculations to a 512-byte memory location. To achieve the highest (slowest) score, Domas first used their own mmiotic tool to find a high-latency area in the internal PCIe fabric, then forced the CPU to load a 512-byte state from MMIO (Memory-Mapped I/O), essentially processing all of those 512 bytes as slow as possible. That took 74 billion cycles or just over 23 seconds to complete.

Then, they went further "by starving the fabric while the load is in flight." They did so by using a series of 4-byte reads from another high-latency MMIO register, overwhelming the CPU's PCIe root complex and forcing the state restore to queue behind frivolous read ops. The next step is to use the AMX instructions available in Intel's Sapphire Rapids for xrstore64. The state area is increased from 512 bytes to 8KB, which could cause the instruction to hang for more than 1 trillion cycles.

The x86 leaderboard is live on GitHub now, and it looks like Domas has ARM and RISC-V leaderboards planned, as well. There are a few rules for the runs. Domas says any setup is fine, so long as only the execution of a single instruction is scored. Interruptible instructions aren't allowed, nor is scoring emulated instructions run on the handler. All times are normalized based on the CPU's base clock, and the platforms were all run without hardware modifications.

We're dealing with assembly code here, so the ranking is less about the specific instruction and more about what you're doing with that instruction.

Domas primarily used two CPUs for testing: the Intel Core i7-8559U and AMD Ryzen 7 5800H (inside the Trigkey S5). For the rdmsr instruction, however, they used a VIA Eden chip, which was a series of embedded processors from the early 2000s. The rdmsr command is used to read a model-specific register, or MSR. According to Domas, VIA "uses an undocumented register at 0x133 that gives wildly high response time." That command took 202 microseconds or 161,602 cycles to execute.

This is not the developer's first foray into wild experimentation with low-level instructions. An earlier project, called movfuscator, is a C compiler that solely uses the mov (move) command.

Software performance expert, scientist, and GitHub developer Daniel Lemire compared historical computer memory prices and showed that decades of falling prices were undone in just a matter of months because of the massive demand for HBM. According to his X post, this event is a historical anomaly, and he cannot think of an instance where technology hardware prices reverted to levels from decades before.

“RAM on a per unit basis is about as expensive as it was in 2007,” Lemire said. He also added, “I do not think anyone can predict what will happen, but I am guessing that we either find a way to build AI systems without so much memory, or we find really clever ways to make much more memory much faster.”

We also checked third-party data compiled and maintained by David Shim for the Stanford DAM Project. The latest price for DDR5 memory sits between $13.28 and $11.41 per GB — the last time RAM hit this price was in 2008, when the price for DDR2 hovered by $15 and 11.00 per GB. These are based on nominal USD values, though. If we take 2024 inflation into account, the current amount for DD5 is between $12.74 and $10.94 per GB, which was last achieved in 2011, when DDR3 prices were at $11.85 per GB.

Technological advances, especially in manufacturing and miniaturization, meant that tech became cheaper over time. This has been true ever since the first general-purpose digital computer was created 80 years ago. ENIAC cost the government $400,000 in 1946, meaning it would probably cost around $6,850,297.44 today. However, its computing power equaled roughly 5,000 additions per second — by comparison, one of the cheapest smartphones you can get today, the Moto G Play, is on sale for $99.99. This smartphone is powered by a Snapdragon 680 processor with AI performance of 3.3 TOPS — that is 3.3 trillion operations per second. We can’t really compare the ENIAC to a modern processor, but it at least shows us the scale by which technology has advanced, and prices have dropped.

The massive increase in memory prices in recent times wasn’t caused by technological regression or a bottleneck in supply — instead, it’s the massive demand for HBM driven by the AI race. “The memory output is increasing by around 20% per year. Now, normally, that would be fantastically fast and amazing for any large, mature industry, but ask yourself, ‘Is the demand increasing by 20% a year?” Elon Musk said during the last SpaceX earnings call. “No, the demand is increasing by 200% a year, maybe higher.”

Other experts and industry leaders agree with Musk. The chairman of the SK Group, the holding company that owns memory chip manufacturer SK hynix, said that RAM prices are “abnormally high,” and that the industry must take steps to increase supply and lower prices. Even RAM that uses chips from Chinese manufacturer CXMT, which some people look to as a cheaper alternative, generally tracked the prices of modules that use chips from the big three (Micron, Samsung, and SK hynix).

Aside from computers, other industries are already being affected by the shortage, including graphics cards, smartphones, gaming consoles, and even automobiles. We do not know how long the current memory crisis will last, but the situation is unsustainable and “something has to give.” The scary thing is that nobody knows what would happen if and when that time comes.

Amazon Web Services is cracking down on internal use of EC2 instances among its engineers. In May, the company reportedly met with engineers and told them to reduce CPU waste to ensure AWS has enough CPU capacity to meet customer demand, The Information reports. The message comes as demand for CPUs in the data center has hit a fever pitch, with the traditional eight-to-one or four-to-one ratio of GPUs to CPUs moving closer to parity.

EC2 instances make up a large chunk of the modern internet, and they're used in private deployments, as well. Traditionally, AWS engineers have been able to spin up their own instances for development, leveraging the relatively low CPU utilization required for web infrastructure to use more virtual machines. Now, engineers say they're waiting days to get access when they previously could get access within hours. One engineer told The Information that they've never had to wait this long for an instance, even after several years of working at Amazon.

Amazon deploys several different types of CPUs in EC2 instances, including AMD and Intel options and its relatively new Graviton5 chip. Graviton5 is Amazon's most powerful CPU to date, and it uses an Arm-based architecture along the lines of Nvidia's Vera CPU and Arm's own AGI.

The increased demand for CPU comes on the back of AI agents, a new paradigm in productivity that even companies as large as Amazon are struggling to reckon with. Last month, for instance, a coding agent blew through $1.8 million in token costs at Amazon, surpassing a development budget by 860%.

Much of the AI infrastructure currently in place is designed around inference, a workload that's accelerated by GPUs. With the four-to-one ratio, the CPU served as a way to keep the GPUs fed, and nothing more. However, agentic workloads are much more complex. They often involve tool calls that run on the CPU, as well as more complex orchestration of inference on GPUs. This is what has brought CPUs center stage in the agentic era. Intel, AMD, and others have echoed what we've heard from memory and storage companies over the past several months: the demand is so high for CPUs that most companies will take whatever they can get.

The major players are capitalizing on that demand. AMD just recently unveiled its portfolio of Zen 6 'Venice' CPUs for the data center, marking the first time AMD has launched a new architecture in the data center before the client market in decades. Nvidia has also pivoted its messaging away from accelerators and toward its new Vera CPU, vying to stake its claim in an expanding market of agentic AI infrastructure.

Although Amazon is wrestling with CPU capacity across its internal engineers and external customers, The Information reports that shortages are largely a problem for spot instances. A consultant told the outlet that contracted capacity hasn't experienced any shortages.

Following publication, an Amazon Web Services spokesperson reached out with the following statement:

"Demand for AWS services, including EC2, is incredibly strong and growing. Even with this heavy demand, we continue to satisfy the overwhelming majority of compute needs for both our internal and external customers. We work closely with internal teams to meet their compute needs while ensuring they use EC2 resources as efficiently as possible, such as reclaiming idle instances, right-sizing, and scaling up and down as needs change – just as we’ve always done. These efficiencies help manage capacity for internal and external customers alike, and any suggestions that these long-standing efforts reflect new capacity constraints is simply wrong.

This premise is sensationalized. Frugality is in our DNA since Day 1. We have always encouraged our teams to operate efficiently. We also share best practices for how customers can optimize resources to external customers. This isn’t a new directive, and encouraging efficient use of resources isn’t unique to Amazon."

We pressed Amazon on if engineers had been given deadlines to reduce their compute usage, as The Information originally reported. An AWS spokesperson responded: "This narrative on EC2 is inaccurate. As part of our normal business operations, and backed by our leadership principle of frugality, we are always driving efficiency across our resources to ensure teams are optimizing capacity."

Tom’s Hardware is renowned for the benchmark tests and performance data we collect on every CPU, GPU, SSD, laptop, and more that runs through our labs — hundreds of products every year, thousands of tests, tens of thousands of data points. Why? Because details matter, and we know that Tom’s Hardware readers value that data.

So for the launch of Tom’s Hardware Premium, we built a new tool called Bench, which offered one-stop access to all of that data. But the tool? We confess, it was merely okay. All those tests were hard to sort through as one giant list, the data was fairly static, and it was hard for the user to put a given score in context to their own rig.

Today, we’re pleased to introduce you to Bench 2.0, a ground-up rethink of our benchmarking tool that makes it the most powerful, most intuitive benchmark browser in the business. Bench 2.0 has tons of new features designed to make the data easier to understand, easier to sort through, and far more usable. Best of all, we've spent dozens of hours re-testing our entire slate of CPUs and GPUs, too.

The most obvious change is the introduction of a detailed hierarchy for tests, which were presented as one long (very long) list in Bench 1.0. Instead, you can pick out just gaming tests or just synthetic benchmarks, or zero in on a given app, whether it’s for content creation or encoding or whatever. We’ve also added detailed product pages, with specs, benchmark test results, and pricing information.

And this is cool: Pricing information culled from our proprietary Hawk tool allowed us to create a neat new visualization mode: Bang for the Buck charts, which plot performance against cost, a crucial data point in today’s expensive market. In both bar charts and these new graphics, mouse over the graphics to interact with them. You can click a given data point to set it as a baseline for comparison, to at a glance see how everything rates against, say, a GeForce RTX 4070 Ti.

Speaking of, did you buy an RTX 4070 Ti? Flag it as your own! You can store your products and mark them as actively in use or in storage. Products you own will show up as blue dots or bars in the charts to showcase your performance against the hottest new gear. You can also pick products you’d like to buy, creating a watch list of what’s on your radar. It’s all tucked away in the 'My Rig' section, accessible from the little suitcase icon in the top right corner.

Finally, we’ve built a better way to share some of this data: The 'Share Results' drop-down has options to let you download an image of whatever chart you’re looking at or create an embeddable, fully interactive widget.

Oh, and don’t miss the settings options at the top right of the page as well, just below My Rig: We’ve built light and dark modes, added sound effects (and the ability to turn them off), and given you a peek into our plans for Version 3 of Bench, which will be expanded to include our prebuilt gaming PC test results, and more. Have fun with the Easter Eggs as well. As befits a tool running on Tom’s Hardware, this one does, in fact, run Doom.

Thoughts? Suggestions? Complaints? We’re all ears. Leave a comment and let us know what you think.

Bench 2.0 is live now, exclusively for Tom's Hardware Premium subscribers.

Nvidia has a booth at the annual QuakeCon, which is happening from August 6 to 9 at the Gaylord Texan Resort & Convention Center in Grapevine, Texas, where the company is selling RTX 50-series GPUs at their original launch price. The company said in its blog post that several Founders Edition graphics cards are available at MSRP while supplies last, and that there will also be several prizes and freebies, including the GeForce Trading Cards Series 1.

“If you’re attending this year’s event between August 6th and 9th, head to the GeForce booth ASAP to get in on the action,” Nvidia said in its blog update. “And if you want a GPU upgrade, our team is bringing Verified Priority Access IRL to QuakeCon, enabling you to purchase Founders Edition GeForce RTX 5090, 5080, and 5070s at MSRP while supplies last.”

The firm officially announced the RTX 50-series in Las Vegas at CES 2025, with the RTX 5090 priced at $1,999, the RTX 5080 at $999, the RTX 5070 Ti at $749, and the RTX 5070 at $549. However, the supplies of these GPUs were limited — the 5090s and 5080s went out of stock on the same day that they arrived on store shelves. It took several months for the RTX 50-series to finally dip below MSRP, but this only lasted a few months as the memory shortage took hold and caused VRAM prices to skyrocket.

This means that the best graphics cards for gaming are quite expensive at the moment, with Tom’s Hardware’s GPU price tracker showing the best price for the RTX 5090 sitting at $4,381 — more than double the $1,999 MSRP that Nvidia set for the GPU. Even the RTX 5080, 5070 Ti, and 5070 aren’t immune to these price increases, with the most affordable options for the graphics cards sitting at $1,289 (29% over MSRP), $989 (32% over MSRP), and $629 (more than 14% over MSRP), respectively.

Unfortunately, this isn’t likely the worst that we will see when it comes to GPU pricing. There has been some disturbing news that prices for RTX 50-series GPUs jumped by 30% in South Korea as wafer costs from TSMC have increased, and GDDR7 modules push past $20. While this hasn’t reached the U.S. at the moment, there is fear that retailers in the country will follow suit soon.

Such is the state of the PC building industry that RTX 50-series GPUs at MSRP have now become a come-on to desperate gamers who just want to upgrade their gaming PCs for a reasonable price. Unfortunately, “Verified Priority Access IRL” is only available during the convention and only until supplies last — once there’s no more stock on site, you’d have no choice but to troll the interwebs for a deal or a secondhand unit if you refuse to pay for more than the MSRP. Alternatively, if you believe that you’ve got the skills, Nvidia is also hosting Quake III Arena RTX Remix and DOOM: The Dark Ages | Revelations challenges, where the best players will each receive an RTX 5070 Founders Edition GPU.

The AI boom means that no matrix math FLOPS are disposable, and that means older Nvidia GPUs with Tensor Cores are getting a new lease on life. Services are popping up that will outfit your RTX 2080 Ti with 22GB of VRAM, doubling its original memory pool and making it more useful for modern LLM and diffusion workloads. If you’re hard up for compute and you don't have an RTX 2080 Ti to modify, however, eBay has just the thing. A Hong Kong-based seller will send you a pre-modded 22GB 2080 Ti in exchange for $499.

The listing only promises that you'll receive a "Turbo," i.e., blower-style, RTX 2080 Ti with the 22GB mod. Don't get too picky about the particular brand of card you might receive, as the seller says: "The GPU brand could be Gigabyte, MSI, ASUS, Leadtek or others. It depends on what we have in hand."

The eBay listing photos show a Gigabyte blower-style card along with a large box of presumably modified cards ready to ship and a GPU-Z screenshot confirming the availability of 22528 MB of memory from a running card.

The seller only has 99.6% lifetime feedback on eBay, which is relatively low, but feedback for this specific listing shows that at least 38 happy buyers have received these cards and that they’re working as described.

Assuming you can count yourself as one of those happy buyers, a $500 22GB RTX 2080 Ti might be the best VRAM bang for your buck that you can find these days. That relatively large memory pool, combined with the RTX 2080 Ti's 616 GB/s of memory bandwidth and Tensor Cores, means that it's still useful for local LLM tasks, although its raw compute capacity and limited reduced-precision data type support compared to more modern products might make demanding diffusion workloads leisurely.

Among Turing cards, the 24GB Titan RTX still commands about $800 on eBay at today's prices, and a Quadro RTX 6000 with the same amount of VRAM is about $900. The local AI enthusiast's favorite RTX 3090, which has 24GB of faster GDDR6X offering 936 GB/s of memory bandwidth, looks to be selling for about $1200. Ampere RTX Pro cards with even more VRAM rapidly get more expensive from there.

The fact that all of these GPUs are still commanding such high prices many years after their introduction is a testament to the continuing utility and universal availability of Nvidia’s Tensor Core architecture across both consumer and data center products since 2018.

AMD has had matrix math accelerators in its IP arsenal since CDNA 1 in 2020, but their availability has been limited to Instinct data center products until RDNA 4 arrived early last year. Intel included XMX matrix engines in the Alchemist architecture from the start in late 2022, but Arc graphics products have faced considerable challenges beyond the presence or absence of that capability. And Apple only just introduced Neural Accelerators to its GPUs with the M5 family.

All that means that if you’re hard up for VRAM capacity and still need a decent amount of compute to go with it, a 22GB RTX 2080 Ti could be a compelling and relatively budget-friendly way to get there, and you get access to the entirety of the CUDA software ecosystem in the bargain. Not bad for an eight-year-old GPU that might have previously ended up in the e-waste bin by now.

Apple reportedly has around $1 billion in "processor wafers on the shelves awaiting packaging" amid ongoing memory shortages, putting pressure on the firm as it nears the launch of the iPhone 18. Tim Culpan of Culpium says Apple is "scrambling" for memory supply ahead of release, working with suppliers to get memory ready so it can finish packaging the A20 Pro and C2 chips that are reportedly inside the iPhone 18 lineup.

It's an interesting peek behind the curtain of the current supply chain limitations, which reach beyond simply DRAM chips. With fully-integrated SoCs across both desktop and mobile, Apple is uniquely vulnerable to supply chain limitations, not only in DRAM supply, but also in packaging capacity to bring together DRAM chips with processor wafers.

According to Culpan, Apple primarily sources DRAM from Micron, with smaller engagements from SK Hynix and Samsung. Micron (along with the other major memory players) has reportedly sold capacity for DRAM and HBM through 2027. A flurry of long-term agreements (LTAs) have cropped up over the past several months, guaranteeing supply to certain companies for several years.

The A20 Pro is the first chip to use TSMC's N2 process, and it has been progressing well, with TSMC reportedly sitting on around $1 billion worth of Apple wafers awaiting packaging with DRAM. Packaging happens with TSMC's new Wafer-Level Multi-Chip Module (WMCM) process, according to Culpan, which is similar to the CoWoS packaging but for mobile parts.

Culpan reports that "Apple and its assemblers remain confident that they can meet initial demand," but DRAM shortages could put pressure on shipments past the initial launch. Apple reportedly plans to ship around 200 million iPhone 18 units across the lineup throughout the device's lifespan. The iPhone 17 range saw record shipments last year, surpassing 245 million devices shipped in 2025.

Apple is one of several companies reportedly looking to alternative suppliers for DRAM, namely China's CXMT. The Chinese memory supplier has become a hot topic in the halls of the U.S. government over the past several weeks, with Apple reportedly lobbying for clearance as the government weighs the option to ban CXMT.

This week, Samsung presented its vision for next-generation memory and storage solutions both for AI and general-purpose applications at the Future of Memory and Storage (FMS) 2026 conference. The company introduced zHBM, zNAND-O, and BV-NAND technologies at the event, and while they are aimed at very different applications, they share one common ingredient: they all rely on wafer bonding.

Here's a breakdown of each type of technology that the company announced.

• zHBM: Samsung’s vision for custom HBM ultra-high-performance memory that will sit on top of logic dies.
• zNAND-O: NAND memory that can be bonded atop a logic device to maximize bandwidth, reduce latency, and minimize power consumption.
• BV-NAND (aka Samsung’s 10^th Generation V-NAND): Samsung’s take on bonding NAND arrays atop all the circuitry needed to operate them. This is essentially next-generation 3D NAND technology that is used today by almost everyone, including Kioxia/Sandisk, SK Hynix, and YMTC (which was the first company to adopt such a method).

This article is a preview of the type of content that readers can expect from our subscription service, Tom's Hardware Premium. If you're interested in more technical content that spans across the hardware industry, subscribe for as little as $29 per year, or $7 per month.

zHBM: Placing HBM stack on top of logic

Around the time we first heard about HBM4's 2,048-bit interface, we read a 'semi-official' report that SK hynix and its partners were considering vertical integration of HBM4 stacks on top of logic chips. To a large degree, companies were considering installing these memory stacks on top of their processors as they doubted that it would be possible and feasible to use interposers to connect HBM4 stacks to AI accelerators. Over time, it turned out that integration of memory on top of logic is complicated when it comes to cooling, power delivery, and developing interposers that can handle a 2,048-bit memory interface. Samsung's zHBM appears to be a continuation of that effort.

Samsung's zHBM concept places HBM directly on top of an AI accelerator instead of on its perimeter, which greatly reduces the distance data must travel between compute and memory. The company says this architecture can increase bandwidth and power efficiency and projects that it can provide 8x the 'performance of HBM5', while its advanced wafer bonding is expected to enable over 10x higher memory density, 3x better energy efficiency, and more than 50% lower thermal resistance than HBM5.

While Samsung's claims about the zHBM look incredibly impressive, they are vague. For example, when the company claims that a 'next-generation interface system incorporating zHBM' will deliver 'approximately eight times the performance of HBM5,' it never explicitly states whether this refers to memory bandwidth, actual application performance due to a combination of improvements over standard HBM, or something else. Had Samsung intended to compare raw bandwidth, it would likely have used more precise wording, such as '8x higher bandwidth.'

Instead, the general term 'performance' may have many meanings. The same applies to references of improved power efficiency, higher memory density, and lower thermal resistance compared to HBM5, a standard that has not been fully defined or ratified. To make matters even more imprecise, Samsung mentions thermal resistance, which depends on implementation rather than the standard. Finally, Samsung didn't reveal which die stacking/bonding technology it will use for zHBM.

In any case, Samsung says that zHBM will also support customer-specific designs and will enable custom IP to be integrated into the interconnect layer (which plays a role similar to the base die, though it is definitely not a base die per se) between the HBM stack and the AI processor, which essentially means that zHBM is not necessarily an industry-standard solution, so it might be implemented with additional perks.

For now, Samsung hasn't disclosed any timeframes for when it plans to offer its zHBM solutions. The only clue is that it compares it to HBM5, which is likely to see the light of day sometime in the late 2020s or early 2030s.

zNAND-O(on-device): Put storage close to compute

Samsung also unveiled zNAND-O, a high-performance NAND memory currently in development that will enable putting four or eight stacks of NAND devices on top of logic dies. zNAND-O(n) device is based on the company's V-NAND platform (or rather, BV-NAND) and is meant to combine high storage density with improved I/O performance and low latency.

Samsung claims that the tech is suitable for edge AI systems running real-time, data-intensive inference workloads, though, of course, there are many other applications that can benefit from high-performance on-package storage.

Since the technology is in development, Samsung hasn't shared much information about it, even keeping the schematics of how it plans to put zNAND-O layers on top of a logic under wraps. We can speculate that since Samsung mentions 'improved I/O performance and low-latency,' it may be planning to exploit both the inherently parallel architecture of NAND memory with a high-performance interface, though we will not try to dive into details for the sake of not making the discussion too speculative.

BV-NAND: Good old V-NAND gets a new treatment

With BV-NAND, Samsung is essentially introducing a new brand name for its next-generation 3D NAND that bonds the NAND array on top of the I/O and logic wafer. This, in turn, produces the layer itself using high-performance logic process technologies and enables higher I/O speeds.

Samsung formally announced its 400-layer-class V-NAND (aka 10^th Generation V-NAND) some time ago, and even disclosed that its areal density would be 28 Gb/mm² with a maximum I/O speed reaching 5600 MT/s. This likely stipulates the usage of a pinout that is different from the 3D NAND packages used today. While the areal density of Samsung's 10^th Gen TLC V-NAND is slightly lower when compared to that of Kioxia/Sandisk's BiCS10 TLC NAND, its maximum I/O speed is significantly higher, which perhaps justifies the new branding.

Considering the fact that Samsung rarely sells its memory to third parties and prefers to market client SSDs itself, a new brand for memory can be a sales driver, even if actual high-end drive performance is limited by a PCIe 5.0 x4 interface rather than by 3D NAND memory I/O. Then again, the performance of mid-range PCIe Gen5 SSDs that rely on a quad-channel NAND controller depends on 3D NAND I/O today, so marketing them under a new flash memory brand is a certain way to attract attention and then sell decent performance.

Samsung's new brand strategy

As the new name of the FMS tradeshow implies, the announcements touch upon very distinct industry needs. In the case of Samsung, we are talking about high-performance data center-grade AI accelerators; applications that need storage with latency that is lower than ~50 – 60 µs of modern high-end PCIe SSDs, though we will see how it compares to Z-NAND, and essentially a new generation of conventional 3D NAND with a new brand.

Except for the 400-layer-class Samsung BV-NAND that is set to hit the market in the foreseeable future, the company's new tech, like zHBM and zNAND-O, is years away.