News Posts matching #GPU

Return to Keyword Browsing

PS5 Emulator KytyPlus Gets Silent Hill 2 Remake to Boot on Integrated Graphics

The KytyPlus PS5 emulator has reportedly rendered the Unreal Engine 5-powered Silent Hill 2 remake on an integrated GPU. Reddit user Legal_Help_31 shared a test on an AMD Radeon 840M with 512 MB of dedicated VRAM plus shared system memory, paired with a Ryzen AI 5 340 and around 15 GB of shared RAM. Performance is still not where we would want it, of course. The game's content warning screen runs at around 4 FPS, and it looks like the game only starts and sits there for several minutes without crashing. The test includes no gameplay at this time. The developer's own GitHub page calls the work experimental and says it is not a playability claim. FSR upscaling can apparently be used as well.

KytyPlus is a high-level emulation (HLE) fork of the KytyPS5 project. It reportedly runs without Sony's console firmware and uses Vulkan for rendering. The original KytyPS5 reportedly still crashes when launching Silent Hill 2 Remake, even on stronger hardware, so KytyPlus is a few steps ahead there. It is also impressive that this technically ran on an iGPU, but a boot screen at 4 FPS tells us very little about whether this ever becomes playable. The emulation scene has moved quickly in recent weeks, though, with KytyPS5 showing tremendous progress in Demon's Souls Remake and Astro Bot, among other games. KytyPlus developers are apparently now working to improve the performance in Silent Hill 2 Remake.

Beelink ME Pro NAS Gets Intel Core Ultra X7 358H Upgrade

Beelink already detailed its updates to its ME Pro NAS solutions in May 2026, but it has apparently seen fit to upgrade the internals on its two- and four-bay NAS units once again, this time upgrading the NAS to include Intel's Core Ultra X7 358H line-up. This means the NAS will be able to leverage the powerful Intel Arc B390 iGPU, effectively allowing the NAS to pull double duty as a network-attached render machine or AI accelerator for private cloud applications and similar.

In addition to the Intel Core Ultra X7 358H, the updated version of the NAS will also be available with the Intel Core Ultra 5 325H, which is a much less powerful CPU. The Beelink ME Pro NAS was previously shown to come in both two- and four-bay configurations, and the Chinese version will ship with Feinui fnOS to manage file backups and remote access, and it will have built-in AI video management and organization capabilities. Tencent's Marvis will also be installed for file search, categorization, and archiving. It's currently unclear if the upgraded Beelink ME Pro NAS will launch with those AI features included when it hits international markets, but an October launch is expected.

NVIDIA Smooth Motion Reaches GeForce RTX 30-Series "Ampere" GPUs Through Mods

Modders have gotten NVIDIA's Smooth Motion frame generation running on GeForce RTX 30 series cards, VideoCardz reports. Smooth Motion is NVIDIA's driver-level frame generation feature, similar in spirit to AMD's AFMF, and NVIDIA only exposes it officially on RTX 40-series and newer through the NVIDIA App. Ampere owners have been left out, at least officially. Two unrelated projects are apparently working to close that gap.

The first comes from developer pipotoufikxyz-lgtm, who expanded the earlier dlssg_for_sm86 project, itself an adaptation of NVIDIA's DLSS-G runtime for the SM86 architecture that RTX 30 GPUs use. The update adds Smooth Motion and Vulkan support, and it reportedly went live on September 19, tested on an RTX 3070. The second, NVSmooth30, comes from a developer going by ItsAdeline and takes a different route: rather than building on the existing DLSS-G package, it's a clean-room compatibility layer aimed directly at NVIDIA's NvPresent64.dll Smooth Motion path. Currently, NVSmooth30 has been tested on an RTX 3050 laptop GPU, and separately on an RTX 3070.

Micron's 2 GB GDDR7 Memory Chips Reportedly Reach End of Life

Micron has reportedly discontinued its 2 GB GDDR7 memory chips, the same modules used on several GeForce RTX 50 series cards. Spotted by UNIKO's Hardware, Micron's 28 Gbps D8FHL chip (MT68A512M32DF-28:A) is now listed as end of life on the company's own website. Every 2 GB GDDR7 chip in Micron's catalog, both the 28 Gbps and 32 Gbps variants, now carries the same EOL status, leaving 3 GB modules as the only option going forward.

Micron only started supplying GDDR7 to NVIDIA in a meaningful way earlier this year, and its chips were never the primary source for RTX 50 cards. Samsung and SK Hynix still cover the bulk of the supply, so this shouldn't cause an immediate shortage. Still, losing even a partial supplier during an already tight memory market isn't reassuring, and it says something about where these companies would rather put their manufacturing capacity these days.

Qualcomm Unveils the Snapdragon 8 Elite Extreme Gen 6 and Snapdragon 8 Elite Gen 6

Qualcomm Technologies, Inc. unveiled today its newest premium mobile lineup - the Snapdragon 8 Elite Extreme Gen 6 Mobile Platform and Snapdragon 8 Elite Gen 6 Mobile Platform. Designed to meet the evolving needs of consumers and OEM partners, both platforms power a new generation of mobile experiences.
Chris Patrick, Senior Vice President and General Manager of Mobile Handsets, Qualcomm Technologies, Inc"Today marks a milestone as we introduce not one, but two new flagship mobile platforms, delivering agentic AI experiences that are immediate, personal, and adaptive," "As the agentic AI age evolves, our multi-flagship strategy allows us to push the boundaries of what's possible for mobile while bringing our latest AI innovations to a broader range of premium devices. By designing two innovative platforms, we're giving OEMs greater choice as we collectively shape the capabilities of the next generation of intelligent, personalized, mobile experiences."

AMD Shows Off Ray Tracing Tech That Slashes GPU Memory Use By 47x

AMD has detailed a new ray tracing technique called tetrahedral cages, and according to a blog post on the company's GPUOpen developer site, it can cut BVH memory use by roughly 47x while speeding up BVH updates by around 90x. AMD demoed the tech on a large scene with about 25,000 independently animated plants, running at 60+ FPS at 1080p on a Radeon RX 9070 XT. In a conventional setup, every uniquely animated object needs its own acceleration structure updated every frame, which gets expensive fast in dense scenes. AMD says the same demo would normally require up to 80 GB of GPU memory and over 300 ms per frame for BVH updates using standard methods. With tetrahedral cages, that dropped to about 1.7 GB of memory and roughly 3.3 ms per frame.

The trick is that instead of animating the dense mesh directly, AMD wraps it in a much simpler deformable cage. The dense geometry and its BVH stay static and reusable, and only the cage itself needs to be animated each frame, so update cost scales with the cage's complexity rather than the triangle count. The tradeoff is losing per-vertex animation control, which is why AMD is positioning it for grass, foliage, and other dense but not individually detailed objects. AMD says the technique can be combined with Microsoft's DirectX ray tracing too. Still, this is only a research showcase in its current state, and there's no word yet on when or if it might show up in games.

Manufacturers Reportedly Planning Both 4 GB and 6 GB GDDR7 Chips For 2027-28

The NVIDIA GeForce RTX 5090 currently ships with 2 GB GDDR7 modules, but even 3 GB modules have seen limited adoption due to their high cost. However, that could change soon. According to Weibo leaker Golden Pig Upgrade, memory manufacturers are reportedly planning 4 GB and 6 GB GDDR7 chips for a 2027 to 2028 window, a jump that could reshape how much VRAM budget and mid-range GPUs can carry without needing a wider memory bus. A 128-bit card using four memory packages currently tops out at 8 GB with today's 2 GB chips. Move to 4 GB chips and that same card could hit 16 GB, or 24 GB with 6 GB modules. Scale that up to a 256-bit interface with eight packages, and you're looking at 32 GB or 48 GB respectively, without touching the PCB layout or bus width. Of course, going the other way is possible too, a narrower memory bus combined with higher density, faster memory chips, can lead to similar bandwidth, with lower relative cost.

This would matter most at the low end, where narrow 128-bit buses have made generous VRAM capacities either awkward or expensive to pull off. Videocardz, citing supply chain sources, notes that today's 3 GB GDDR7 chips are already unusually expensive due to extreme demand, and that cost has reportedly contributed to delays around the rumored RTX 5050 9 GB and the RTX 50 Super refresh. If 4 GB chips end up cheaper to produce at scale, they could become the more practical choice for mainstream cards, while 6 GB modules might make more sense for workstation and other high-memory GPUs first. It's worth noting that the leaker's timeline only points to sometime in 2027 or 2028, so there's no confirmation this lines up with NVIDIA's RTX 60 series or AMD's next Radeon generation specifically.

Intel Introduces "PISA" Virtual ISA for Greater GPU Software Portability

Intel today updated its Compute Runtime with support for Portable Instruction Set Architecture (PISA) format, a new virtual ISA and programming model for Intel GPUs. Unlike CPUs, where software commonly targets established ISAs such as Arm or x86, GPUs have their own native instruction sets and hardware-specific architectures. As GPU generations introduce new hardware capabilities, execution features, and native instructions, compilers targeting the machine ISA have to account for differences between architectures. A standardized virtual ISA changes this by introducing an abstraction layer between compiler output and the GPU's native machine code. Essentially, PISA provides a common compiler target that allows generalized code to be written once and then finalized for multiple compatible Intel GPU architectures.

With PISA, Intel aims to provide a stable and portable ISA for compilers that target Intel GPUs. This means that a workload compiled to PISA can potentially run across multiple compatible Intel GPU architectures without requiring architecture-specific code generation for each generation. The PISA representation can then be finalized into native machine instructions for the target GPU, allowing the same workload to take advantage of the capabilities of newer hardware where supported, without requiring changes to the original PISA code. This does not mean that every PISA workload will automatically run on every Intel GPU, however, as compatibility depends on the target architecture and the features and instructions used by the program. PISA is also not Intel's first virtual GPU ISA, as the company's existing vISA has long served as an intermediate representation for its GPU compiler stack, while PISA introduces a newer portable ISA and programming model.

Acer Expects PC Prices to Rise by Up to 20% in Q4, Peak Around Mid-2027

Acer Chairman and CEO Jason Chen says PC prices could rise by another 5% to 20% in the fourth quarter of 2026, according to Taiwan's Economic Daily News. Chen reportedly said that DDR4 and DDR5 memory are no longer in short supply and that CPU supply is mostly fine, with only some low-end and mid-range chips still tight. Prices haven't followed this change, though. Memory and SSD costs are still high, so systems with more RAM or storage will probably see bigger increases. Chen expects the average selling price of PCs to peak around mid-2027, with a possible drop in the second half of that year.

Even if component costs fall earlier, it takes several months for cheaper parts to reach finished PCs. Chen also reportedly said suppliers keep pushing through price increases, some of which he considers hard to justify. He also pointed to expanding Chinese memory production, and by the sound of it, he doesn't think the shortage can run until 2030, as many in the industry have claimed. That capacity is still ramping up, though. Chinese giant CXMT is reportedly preparing to enter the NAND flash market with an R&D line, and YMTC is expanding with new fabs.

AI Chip Packaging Pushes Beyond Reticle Limits, with CoWoS-L Set to Remain the Mainstream Advanced Packaging Solution Through 2028

TrendForce's latest research into advanced semiconductor packaging reveals that GPU suppliers and hyperscale CSPs are continuing to develop increasingly powerful and versatile AI chips, making larger package sizes a key development priority for 2.5D packaging. Although TSMC's CoWoS-L faces competition from Intel's EMIB-T, its advantages in technology maturity and yield are expected to keep it the mainstream packaging solution for AI chips through 2028.

TrendForce notes that strong AI demand continues to support NVIDIA's push toward integrated GB/VR rack-scale solutions, which incorporate large numbers of chips. NVIDIA's high-end GPU shipments are expected to grow approximately 30% YoY in 2026. AMD, meanwhile, is aggressively promoting its MI400/MI450 series beginning in the second half of 2026, providing further momentum to overall shipments.

PS3 Emulator RPCS3 Gets Major FPS Boost After NVIDIA Driver Bug Fix

The RPCS3 team says it found a workaround for an NVIDIA driver bug that was holding back performance in the popular PlayStation 3 emulator on GeForce cards. Contributor "Yahfz" reportedly tracked the issue down, and the fix apparently boosts frame rates by up to 37% depending on the game and GPU, while also cutting VRAM usage in some titles. AMD GPUs are not affected, since RPCS3 says the bug is specific to NVIDIA's drivers. RPCS3 also said it has tried reaching NVIDIA through developer forums about this and other driver bugs, without much luck, and is hoping for a more direct line of communication going forward.

The numbers RPCS3 posted on X are pretty striking. A Ryzen 7 9850X3D paired with an RTX 5070 reportedly saw Gran Turismo 5 jump from 107 FPS to 147 FPS, a 37% gain, while VRAM usage dropped from 7.1 GB to 4.8 GB. Other pairings showed smaller gains: a Core i9-13900K with an RTX 3080 gained around 20% in one Red Dead Redemption test scene, and even an aging GTX 1060 with an i7-7700HQ reportedly saw roughly 10-20% improvements in Saints Row titles. Actual gains will apparently vary a lot by game and hardware, so it's probably worth tempering expectations a bit until more testing rolls in. Nonetheless, it looks like NVIDIA users emulating PS3 games get a solid free performance bump, though a proper fix from NVIDIA's side would still be the better long-term outcome.

Apple M5 Ultra Scores Big GPU Gains in Leaked Geekbench Benchmark

Apple officially announced the M5 Ultra in late August 2026, alongside the new Mac Studio desktop PC, claiming that it would offer significant increases in both AI and GPU-accelerated workloads. A new leaked set of Geekbench scores seems to validate many of Apple's claims in the GPU performance department, with the new GPU in the M5 Ultra coming out around 41% faster, on average.

The Apple M5 Ultra GPU scored a total of 360,019 points in the Geekbench 7 GPU benchmark, while the M3 Ultra only clocked in with 255,009 points in the same benchmark. Even compared to an NVIDIA GeForce RTX 5080, the Apple Silicon SoC's GPU impresses, trading blows with the 5080 in creative workflows, physics simulations, and photo editing, in some cases, beating out the gaming GPU by as much as 22%—while also losing by as much as 100% in some cases.

NVIDIA DLSS 5 Neural Rendering Now Runs in a Web Browser Via WebGPU

A developer going by MAAN has demonstrated NVIDIA DLSS 5 Neural Rendering running inside a web browser using WebGPU, and according to their post on X, it works on macOS too. A live demo is up now, and MAAN says people can try it with their own models as well. This is odd because NVIDIA doesn't officially support DLSS through WebGL or WebGPU at all. DLSS normally runs through NVIDIA's NGX interface or the Streamline SDK, which sits between a game and DirectX or Vulkan and needs motion vectors, depth data, and the rendered frame handed over directly. None of that maps cleanly onto a browser's pipeline, so it's not clear which approach the developer has used to get it running on a web browser.

The developer hasn't shared technical details either, but they have confirmed that the source code will be shared on GitHub soon. Performance is another question mark. VideoCardz says loading the Neural Rendering model alone took a second or two on an RTX 4090, and running on Apple Silicon apparently works but isn't fast. That probably rules out real-time gaming for now, though the concept could still be useful for 3D model previews, where a slower response time matters less. It's an early proof of concept more than anything, but getting DLSS 5 working outside NVIDIA's own supported paths entirely is still a neat trick.

AMD Plans 10% Price Hike Across GPUs, Chipsets, and Possibly CPUs

AMD is preparing for a series of price hikes in response to rising costs of acquiring silicon from its supply chain. According to ChannelGate, AMD is working on a 10% price increase across its products due to the upstream supply chain squeeze. Since most of AMD's silicon is manufactured by Taiwanese semiconductor giant TSMC, any change in wafer pricing forces AMD to pass these costs onto customers. Reportedly, TSMC is increasing the pricing of its silicon wafers, and as a result, AMD must raise its product costs to maintain margins and keep the business afloat. The price increases will affect almost every product AMD offers, including GPUs, possibly CPUs, and even motherboard chipsets.

For CPUs and GPUs, AMD has already implemented price hikes in the past, driven by a combination of demand and high DRAM pricing. However, this is one of the first instances where AMD is increasing the pricing of its motherboard chipsets, which will require motherboard makers to adjust their pricing accordingly. If TSMC implements a 10% price hike across its nodes, from 2 nm N2 to some older mature nodes, AMD will have to pass these costs onto partners and consumers, who will ultimately bear the difference. As a fabless chip designer, AMD cannot dictate or negotiate for lower rates for wafers from TSMC.

Gnome 51 Launches With Smoother Desktop Performance and Myriad QoL Updates

It's been just over six months since Gnome 50 launched into beta, in the process, ditching X11 and issuing a number of updates to Mutter, its compositor, including VRR and fractional scaling. Now, the Gnome Foundation has officially announced the public release of Gnome 51, which features myriad updates, both to the user interface and to the underlying tech, all of which is aimed at making Gnome easier to adopt and use as an average computer user and make "the desktop feel smoother and more responsive than ever."

Most notably, Gnome 51 improves frame scheduling, which should make the system feel more responsive, especially under load; faster screen capture with the screenshot tool, and streamlined support for NVIDIA GPUs, which will now all be consolidated into a single, modern management UI. Some of the other major changes include new settings dialogues for smoother fingerprint enrollment, an Auto Rotate option in display settings, for portable devices with accelerometers, and more intuitive window snapping with a new center snap option. Gnome 51 will also let users automatically disable the touchpad when a mouse is connected.

SteamOS is Getting Native NVIDIA GPU Driver Support

Valve is officially adding NVIDIA GPU driver support to SteamOS, as confirmed by the company's software developer Pierre-Loup Griffais. Until now, AMD has been the primary choice for SteamOS, both for CPUs and GPUs, since all of Valve's products are based on AMD hardware. However, this is changing as Valve aims to expand SteamOS to more PCs, allowing gamers to realize the Steam Machine dream on their own hardware, even if it is based on a different GPU manufacturer. When IGN asked Valve's developer about the progress of NVIDIA driver support, the response was:
Pierre-Loup Griffais for IGNYeah, we're working on it hard... We've done a few things like we've enabled the driver now in the main tree of SteamOS, so you can get things to light up on certain generations, but in terms of the performance difference between that driver and where the state of the art is, there's definitely still a little bit of a rift that we're trying to work on, but it's getting better day-to-day.

Sonnet Launches Thunderbolt 5 eGPU Box with 850 W PSU and Triple-Slot GPU Support

Sonnet, the US-based accessory maker who previously launched the Echo 20 and 21 Thunderbolt 5 docking stations with 10 GbE, has officially released the Breakaway Box 850 T5, a Thunderbolt 5-based eGPU box with an industrial aesthetic and a built-in 850 W PSU. The Breakaway Box 850 T5 is available from Sonnet for $599.99.

The Breakaway Box 850 T5 is designed to accept triple-slot GPUs with dimensions up to 345 × 155 × 70 mm, and the included PSU features three 8-pin connectors that can be adapted to a 12v-2×6 connector in the case of the latest NVIDIA and certain AMD Radeon GPUs. The enclosure itself weighs 6.4 kg and measures 210 × 400 × 250 mm, and it is backwards-compatible with Thunderbolt 4 and USB4 in addition to Thunderbolt 5.

First Certified Ultra96 HDMI 2.2 Cables Hit the Market From Silkland and ELECOM

Certified Ultra96 HDMI cables, required to unlock the full HDMI 2.2 specification, are now available for retail purchase. VideoCardz spotted Silkland taking orders for its new H96 series, which appears to be among the first certified Ultra96 cables available to consumers, offering up to 96 Gbps of bandwidth. ELECOM also revealed its own Ultra96 cables on September 8, advertised as the first certified products of their type in Japan, though those aren't expected to ship until mid-November. Silkland's cables, by comparison, opened for orders on September 14. Currently, there's nothing on the market that can actually use that bandwidth, so these cables are targeted towards future products.

HDMI 2.2 was finalized back in June 2025, and it enables some genuinely big jumps on paper, like uncompressed 8K at 60 Hz or 4K at 240 Hz with full 10-bit and 12-bit color. It also edges out DisplayPort 2.1's 80 Gbps ceiling. Actual HDMI 2.2 hardware, meaning GPUs, TVs, and monitors with the right silicon, isn't expected until 2027 at the earliest. Even then, AMD's upcoming RDNA 5 GPUs reportedly won't hit the full 96 Gbps tier, apparently topping out closer to 64 or 80 Gbps instead.

Microsoft Rolls Out Emergency Windows 11 Patch, But AMD GPU Driver and File Explorer Crashes Remain

Microsoft today rolled out an out-of-band emergency Windows 11 patch with KB5129195, which is being distributed to everyone. This emergency update is automatically downloading across every PC, as it fixes crucial issues and patches privilege elevation vulnerabilities. The list of fixes in this update includes a patch for CVE-2026-62721, issues with remote desktop connection, and importantly, a partial fix for faulty USB audio. If readers recall, Microsoft's previous September update disrupted USB audio output on many PCs, affecting USB Audio Class 1.0 devices, where systems that used to work produced no sound. In the latest patch rolling out today, Microsoft noted that 8-channel and 3D audio modes are being fixed, but several other issues remain. For example, the lack of volume controls and the complete lack of audio output are still present for some USB audio devices, so this release doesn't completely fix the problem.

Next, there are a few unconfirmed issues in the previous KB5124008 September update that sites such as Windows Latest are tracking. Today's out-of-band KB5129195 doesn't fix them. These include issues with AMD Radeon GPU users, who are reportedly experiencing system freezing, crashing, driver timeouts, black screens, missing display output, and even immediate errors after reinstalling the driver stack. Separately, Windows Latest also spotted users complaining about File Explorer failing to start or crashing immediately, as well as not being able to access file history. This issue is present in some enterprise environments such as Citrix UPM, FSLogix, Horizon, or ProfileUnity ProfileDisks on both Windows 11 24H2 and 25H2.

Intel's Big 12 Xe Core iGPU for "Nova Lake-S" Postponed to "Razor Lake"

Intel's long-rumored "big iGPU" version of the upcoming "Nova Lake-S" desktop processor family might not materialize, according to well-known Intel leaker Jaykihn. Intel had been working on a desktop "Nova Lake-S" version with 12 Xe3P cores and a 16-core (4+8+4) CPU cluster, designed to provide enough graphics power for light gaming and 3D work using the integrated graphics solution. However, the future of this Intel APU is now uncertain, as it has reportedly been postponed to launch with the "Razor Lake" generation, which is set to debut a year after "Nova Lake." Considering that "Nova Lake" is still months away and this SKU has been moved to the next generation, it seems we are quite far from seeing a desktop APU from Intel capable of delivering excellent integrated graphics performance.

Previously, we reported that this specific "big APU" SKU could support DDR5-8000 memory through CSODIMM modules, with a Power Limit 2 (PL2) setting of about 40 W. This builds on earlier rumors that this integrated graphics solution would come with two VCCGT phases for up to 65 W of power. This SKU includes four P-Cores, eight E-Cores, and four LPE-Cores, along with 12 Xe3P GPU cores. This 4+8+4 CPU core and 12 GPU core configuration is one of the few processors where integrated graphics would require two VCCGT phases. In regular SKUs, VCCGT is a single-phase power rail for the integrated graphics inside the processor. However, it is now known that the PL2 setting will only use 40 W, with the 65 W threshold reserved for extreme scenarios like PL3, and the graphics processor's PL4 capped at that 65 W figure. The Xe3P features more microarchitectural changes compared to the Xe3 found in "Panther Lake."

AMD Ryzen 5 5500F Beats Ryzen 5 5500 by Up to 16% in Early Gaming Benchmarks

AMD's new Ryzen 5 5500F is turning out to be more than just a rebadge of the standard Ryzen 5 5500 in gaming. New benchmarks from YouTuber RandomGaminginHD show the $99 chip pulling ahead of its non-F sibling by up to 16% in CPU-bound titles, even though both processors share the same core count and land at roughly the same price. Do note that the testing is done by a mostly unknown third-party reviewer, and we don't have solid data track to back the claim, but for now it remains as a reference for future independent reviews from traditional sources.

The testing paired both chips with an RX 9060 XT 16 GB graphics card and ran through seven games at 1080p. Battlefield 6 showed the biggest gap at 16% in favor of the new 5500F, followed by Cyberpunk 2077 at 11%, GTA V: Enhanced at 9%, and Counter-Strike 2 at 8%. Titles that leaned more on the GPU, like Crimson Desert and Red Dead Redemption 2, saw that gap shrink to single digits or vanish once the RX 9060 XT became the bottleneck instead of the CPU. Of course, these are early numbers from one single test system. The reviewer also does not have a perfect track record when it comes to quality of testing, so it's important to wait for further testing to form a concrete verdict either way.

OpenRGB 1.0 Launches for Open-Source RGB Customization on Windows, Linux, and macOS

RGB control software has long been a pain point of gamers, with background apps and clunky UIs frequently coming up in discussions about gaming mice, keyboards, speakers, monitors, among many more. OpenRGB is a project that hopes to cut through some of that noise and provide a cross-platform support for RGB on gaming peripherals and components. OpenRGB just released version 1.0 after months testing release candidates. With the update, the utility expands support to a number of new brands and components and lays the groundwork for more Logitech device support.

For starters, the latest version of OpenRGB adds support for QMK and VialRGB, including Keychron's forks of those protocols, as well as more Gigabyte and MSI motherboards laptops, and motherboard, Clevo laptop keyboards and light bars, various GPUs across a number of controllers, including PNY Epic-X ARGB and PowerColor GPUs, and more.

AI Firms Are Reportedly Buying NVIDIA RTX 5090 GPUs in Bulk for Server Use

New photos shared by HKEPC allegedly show pallets of brand-new, boxed GeForce RTX 5090 cards being unpacked at AI server assembly sites. Notably, these firms are not using the RTX PRO 6000 or dedicated data center accelerators NVIDIA actually markets for that work. The cards in question are standard retail RTX 5090s, the same cards gamers have struggled to find at anywhere close to MSRP, being pulled straight from their packaging and slotted into multi-GPU AI rigs. It must be mentioned that some elements of these photos appear to be altered, or at least edited with AI, so take these with a bit of caution.

Despite that, the current situation is not entirely new. During the crypto boom, miners bought up GeForce cards in bulk while gamers got squeezed out, and NVIDIA eventually responded with Lite Hash Rate models that throttled mining performance without touching gaming specs. That restriction got fully bypassed within about a year. This time, the workload is AI instead of mining, but the outcome for gamers looks familiar. As a result, retail RTX 5090 pricing has reportedly climbed past $5,000 on some listings, more than double its $1,999 launch price. Industry analysts even warned about this possibility months ago, and we now have a good idea why the RTX 5090 is at the price it's at.

ASRock's TempGuard Failed to Stop a PSU-Side Connector Meltdown on a $15,000 RTX PRO 6000 Rig

A repair shop technician has reported another melted 16-pin power connector on an ASRock Taichi TC-1650T PSU, paired with a $15,000 NVIDIA RTX PRO 6000 Blackwell Workstation Edition GPU. The former ships with native 12V-2x6 cables carrying ASRock's own NTC temperature sensor. According to the technician's Reddit post, the connector failed catastrophically on the PSU side and melted into the socket itself, even though both ends of the cable were fully seated. The GPU reportedly survived, and the damage was confined to the power supply side. The technician also says the sensor is mounted on the wrong side of the cable, on the ground line rather than the 12 V side it's supposed to watch, which would explain why it apparently never caught the rising heat in time.

ASRock added the NTC sensor and TempGuard system to its 12V-2x6 cables specifically to shut a system down before a connector melts. This time the system failed to catch the rising temperatures, but it has worked in the past. TempGuard reportedly caught a shunt-modded RTX 5090 pulling over 1,300 W, saving its power connectors from melting. On the flip side, this isn't the first time this exact GPU and PSU pairing has run into trouble either, as a similar melting incident on the same combo surfaced back in March. If the sensor placement is really off on at least some cables, that's a bigger concern than one bad unit, especially when paired with a card that costs as much as a used car.

Cooler Master GPU Shield Adapter Officially Launches in the US

Cooler Master previously showed off the GPU Shield adapter cable at CES 2026, demonstrating the small 12V-2×6 adapter that is able to not only alert the user with a visual and auditory warning when there are anomalies in the current in the GPU power cable, but also actively intervene and reduce current when higher-than-normal current is detected. GPU Shield has officially launched in the US after showing up in China in early August 2026, and it is available on the Cooler Master online shop for $49.99.

In addition to monitoring high current, which is often the cause of the melted 12V-2×6 GPU connectors, GPU Shield also attempts to mitigate the meltdowns in the first place with dual-color connector tips that are an easy indicator to ensure proper cable insertion in the GPU and PSU. The adapter itself comes with 200 mm of total cable length, which should be enough for most DIY PC cases with conventional PSU placement, and it improves on Cooler Master's GPU Shield tech that is built into the PSU itself, since that is often not visible in many modern PC cases.
Return to Keyword Browsing
Nov 22nd, 2026 16:32 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts