Corsair is proud to announce the launch of the upgraded RM1200e with Platinum-certified efficiency, the new flagship of the best-selling RMe power supply family. It joins the latest generation of the RMe PSU series as a high-powered option for system builders who value the reliability and features of the RMe series.

The RM1200e is also the first Corsair power supply to feature a native 12V-2x6 connection with PinProtect per-pin overcurrent protection (OCP). Designed to help protect high-end graphics cards, PinProtect monitors the on flowing through each power pin in real time and reduces power to the connection if the current exceeds a preset threshold for too long, triggering a protective GPU shutdown. If this occurs, users should completely power down their system, check for loose or improperly seated connections, and only restart once the connections are secure. This helps reduce the risk of damage caused by excessive current.

Reddit user DallasGrave, who recently soldered a dual 8-pin board to an RTX 5090's 16-pin header, is back with a more ambitious take on the same idea. The new "Quad 2x8 Header Conversion" mounts four 8-pin PCIe connectors right on the card, tapping the contact points of the 12V-2x6 header, so the 16-pin connector can either stay in use alongside it or be skipped entirely. Each 8-pin carries three 12 V conductors, which spreads the load over twelve wires instead of six.

According to the modder, current per conductor is "almost perfectly halved" compared to the native connector, landing at roughly 4.1 to 4.4 A per wire at 600 W, with wire temperatures dropping by around 20°C. The sense pins are set for 600 W, and the board has apparently been tested up to 800 W on a shunt-modded card, where the PCB itself reportedly sat at about 45°C. Videos covering installation, live current data, and a 24-hour sustained load test are apparently being shot now, and custom backplates are in the works to make the mod look more factory-made. Versions for nearly every card, Founders Edition included, are also going into production.

NVIDIA today quietly added another card to its pro-viz product lineup with the RTX PRO 5500 "Blackwell" workstation GPU edition. Featuring 84 GB of GDDR7 memory, it is equipped with a GB202 die that includes 21,760 CUDA cores. This is the same CUDA core count as the consumer GeForce RTX 5090 SKU, but it is now certified for professional workloads and has a much higher memory capacity. The 84 GB of GDDR7 is connected to the GB202 with a 416-bit memory bus that delivers about 1,398 GB/s of memory bandwidth. This would put each GDDR7 memory chip at about 26 Gbps per pin, which is a standard GDDR7 memory speed, but not the fastest available on the market. Additionally, to achieve the massive 84 GB capacity, there need to be 28 memory modules. NVIDIA placed 14 GDDR7 modules on the front side with the GPU die and 14 GDDR7 modules on the back of the PCB, effectively sandwiching the PCB.

Compared to the rest of the RTX PRO lineup, the new RTX PRO 5500 "Blackwell" sits just a tick below the flagship RTX PRO 6000 SKU, which has a slightly higher CUDA core count of 24,064 and a 96 GB memory capacity. However, the RTX PRO 5500 is also much more powerful than the RTX PRO 5000 in the 5000 tier. The RTX PRO 5000 "Blackwell" has a much lower CUDA core count of 14,080 cores, with up to 72 GB of memory. Technically, the new SKU is much closer to the RTX PRO 6000 than the RTX PRO 5000 SKU, just with a slightly lower core count and memory capacity. The official specification sheet follows, courtesy of PNY.

DLSS 5 seems to be contributing to increased power draw, and we have our first victim here. Videocardz reports a new melted-connector case on an MSI GeForce RTX 5090 Gaming Trio OC, which surfaced during DLSS 5 Neural Rendering testing in NBA 2K27, currently the only officially supported game. A reader nicknamed erek documented the incident, noting DLSS 5 pushed the card's power draw from around 450 W to over 600 W, which is above MSI's 575 W rating. A GPU-Z screenshot showed a peak board power reading of 613.5 W, which erek says held steady rather than spiking briefly.

Erek reported the card stopped working shortly after, and a burning smell led him to find that the connector's plastic had melted and fused to the socket. We already reported last week that DLSS 5 can push the RTX 5090's power draw unusually high, so this seems to be the first reported case where that added power draw has burned the notorious RTX 5090 power connectors. However, it should be remembered that GPU-Z tracks total board power rather than just the auxiliary connectors, since the PCIe slot also supplies power, so the reading doesn't necessarily mean the connector carried the full load.

Corsair (Nasdaq: CRSR) is proud to announce the launch of ThermalProtect, a 12V-2x6 GPU power cable featuring innovative technology that monitors the temperature of a GPU's power cable in real time to help prevent damage to the GPU.

Corsair ThermalProtect
ThermalProtect enables additional Over Temperature Protection (OTP) for GPUs by actively monitoring the 12V-2x6 cable. Housed discreetly inside a cable comb 30 mm from the connection point, the hardware offers a clean, unobtrusive appearance. If the sensor detects extreme temperatures, ThermalProtect activates, causing the GPU to shut down to prevent potential damage.

ASML says it has boosted the light source power in its EUV lithography systems to 1,000 watts, up from roughly 600 W today. According to a Reuters report, the company claims the increase could enable up to 50% more chip output by the end of the decade. More source power translates directly into higher throughput. ASML says customers could process around 330 wafers per hour by 2030, compared to about 220 wafers per hour today, lowering the cost per chip. EUV light at a 13.5 nm wavelength is generated by firing a CO₂ laser at microscopic droplets of molten tin, creating plasma that emits EUV radiation. The light is then collected and directed through precision optics supplied by Carl Zeiss AG. To reach 1,000 W, ASML doubled the droplet rate to roughly 100,000 per second and shifted to a two-pulse laser shaping approach instead of a single pulse. The company says there is a path toward 1,500 W and potentially even 2,000 W over time.

ASML is the only supplier of commercial EUV scanners, used by major chipmakers including TSMC, SK Hynix, and Intel. The tools are considered critical to advanced node production and have been subject to export restrictions to China. The move is aimed at maintaining ASML's lead as U.S. companies, such as Pat Gelsinger's xLight EUV startup and Chinese efforts to develop competing lithography systems continue to ramp. Chinese companies have reportedly been sourcing parts from older ASML machines through secondary markets. Huawei is leading the charge, working to build a homegrown AI supply chain as a way around foreign tech restrictions. The company set up a large semiconductor manufacturing facility in Guanlan focused on 7 nm chips for its own processors. The Chinese government had initially aimed to have a working prototype from this effort by 2028.

Flashing alternative vBIOS versions onto GPUs is a common practice used by modders to increase power limits on GPUs, and it has only become more common since NVIDIA started clamping down on overclocking options available to users via software tools. The recently announced ASUS ROG Matrix Platinum has become an attractive source for one such high-wattage vBIOS, although it has proven difficult to get the aforementioned vBIOS onto certain cards. One user on the HardwareLuxx forum, however, has managed to get the 800 W vBIOS from the Matrix Platinum GPU onto an ASUS ROG Astral GeForce RTX 5090 (read our review of the latter here), although the mod only seems to be necessary on the black version of the card.

It turns out that in order to flash the Matrix vBIOS onto the black Astral 5090 required physical modifications to the GPU PCB itself—this is due to the rear-mounted fan being wired with the front middle fan on that particular version of the card, as opposed to on a separate header as it is on the white version. The modification in question involves swapping a pull-down resistor in the serial in signal line of the EEPROM to a pull-up resistor. Obviously, this mod alone requires more than a little bit of soldering experience, but there's also the question of cooling. The stock version of the ROG Astral is designed with a 600 W power limit, and the Matrix vBIOS lifts that to 800 W, so realistically, any overclocking attempts made with the modded card requires an aftermarket cooling solution to handle the increased power and corresponding heat.

Corsair is delighted to announce the new HXi SHIFT Series of fully modular, low-noise, high-efficiency power supplies that make building high-performance PCs easier and smarter. HXi SHIFT Series PSUs feature a built-in iCUE LINK System Hub and a side-mounted SHIFT connector interface, simplifying installation, reducing cable clutter, and enabling direct control of iCUE LINK devices. With Cybenetics Platinum efficiency, ATX 3.1 certification, and PCIe 5.1 compliance across 1000 W, 1200 W, and 1500 W models, HXi SHIFT delivers clean, reliable power for next-generation gaming and creator PCs.

The integrated iCUE LINK System Hub inside the PSU removes the need to mount and power a separate hub, enabling builders to connect up to 24 iCUE LINK devices straight to the power supply. This approach means fewer components and cables are needed, resulting in faster builds, easier troubleshooting, and a cleaner interior that looks as good as it performs. The side-mounted SHIFT interface channels all the modular connections to the PSU's left panel for clear access in modern cases. Routing becomes straightforward and tidy, thanks to Type-5 Micro-Fit modular cabling. The included debossed PVC cable kit features low-profile combs that lay flat and keep cable bundles compact. The result is professional-grade cable management that can improve airflow and make for a more appealing final build.

As modern homes and small businesses become increasingly dependent on connected devices and 24/7 digital access, the risks of sudden power outages-such as data loss, hardware damage, and operational disruption-are more critical than ever. Goldenmate, a brand committed to intelligent power solutions, has launched a next-generation line of UPS (Uninterruptible Power Supply) systems powered by lithium iron phosphate (LiFePO4) technology. These systems ensure consistent, secure backup power across essential devices, including network systems, workstations, and home servers.

Whether it's a home office or a small studio, Goldenmate UPS units provide instant response when power drops, keeping key electronics operational without interruption.

MSI, a global leader in high-performance server solutions, is showcasing its enterprise-grade, high-performance server platforms at ISC 2025, taking place June 10-12 at booth #E12. Built on standardized and modular architectures, MSI's AI servers are designed to power next-generation AI and accelerated computing workloads, enabling enterprises to rapidly advance their AI innovations.

"As AI workloads continue to grow and evolve toward inference-driven applications, we're seeing a significant shift in how enterprises approach AI deployment," said Danny Hsu, General Manager of Enterprise Platform Solutions at MSI. "With modular and standards-based architectures, enterprise data centers can now adopt AI technologies more quickly and cost-effectively than ever before. This marks a new era where AI is not only powerful but also increasingly accessible to businesses of all sizes.

NVIDIA has prepared a "Blackwell" Titan equivalent—RTX PRO 6000 Blackwell GPU for professional visualization and local AI solutions. Based on the GB202 silicon, the RTX PRO 6000 is the closest solution to the full power of NVIDIA's prosumer-oriented Blackwell SKUs. With 24,064 CUDA cores on board, this configuration is just 512 CUDA cores shy from the complete GB202 24,576-core configuration. This is likely due to the yield defects, meaning that this is perhaps the highest CUDA core count model we will see based on GB202. Additionally, at 24,064 core RTX PRO 6000 Blackwell is carrying 2,340 more cores than the top-end consumer GeForce RTX 5090 with 21,760 CUDA cores. Based on the 600 W TGP, the new RTX PRO 6000 Blackwell should have lower clocks, given its bigger core count to maintain the TGP for its massive memory configuration. Remember, GDDR7 memory modules consume power, too.

With 96 GB of GDDR7 memory and ECC memory correction on board, it will primarily target professionals in 3D rendering, simulations, and local AI development. To cool the beastly configuration, NVIDIA opted for a double-flow-through cooler used on the RTX 5090, with an open-air design. Operating on a PCIe 5.0 x16 interface with four DisplayPort 2.1 connectors, the design remains double-slot in thickness. VideoCardz obtained pictures of it, which you can see below. Interestingly, the color accents of the new cooler are darker. We can expect to see more from NVIDIA and its new RTX PRO 6000 Blackwell GPU from the upcoming GTC 2025 on March 17.

Days before releasing, Chinese leakers on Chiphell are showcasing a custom variant of AMD's upcoming Radeon RX 9070 XT with what appears to be a 12V-2x6 power connector. The custom AIB model is Sapphire Radeon RX 9070 XT NITRO+, which features a triple-fan cooling configuration and a departure from Sapphire's older GPUs that used eight-pin power connectors. Despite proving to be problematic for NVIDIA, this power connector could do well with AMD's Navi 48 XT GPU SKU due to its power consumption envelope. With an entire GPU projected to use 304 W total board power, the 12V-2x6 connector could easily power this configuration without the need to overheat and possibly melt. If it manages to power 600 W TGP NVIDIA GeForce RTX 5090, then powering a custom Radeon RX 9070 XT shouldn't be an issue.

Even if the card experiences a heavy overclock, power limits will remain within a usable range so as not to cause any trouble. Sapphire's reason for ditching the older, reliable eight-pin power connectors is unknown, but the use of the new 12V-2x6 here isn't expected to be problematic either. Some previous VBIOS records in our GPU DB indicate that the GPU could boost the Navi 48 XT SKU to 2,520/2,518 MHz, which is about 120 MHz higher than the stock AMD Radeon RX 9070 XT boost clock.

A monitoring adapter prototype for NVIDIA's problematic 12VHPWR/12V-2x6 power connectors has surfaced in Asian markets, potentially offering RTX 5090 owners a stopgap solution amid ongoing concerns about thermal issues and power delivery weaknesses. The prototype features a circuit board with individually routed +12V lines through shunt resistors, enabling precise current measurement across each power line while maintaining ground and sense pin functionality. The adapter incorporates voltage monitoring capabilities and an apparent alarm function designed to trigger during overload conditions. A USB port is present on the board for user-accessible data output to a custom display. Notably, the current design iteration appears to omit temperature monitoring functionality, focusing exclusively on current distribution metrics.

These monitoring solutions merely detect rather than resolve the fundamental design issues reportedly affecting NVIDIA's high-end graphics cards. NVIDIA has maintained silence regarding the reported thermal issues and power supply inconsistencies affecting their flagship GeForce RTX 5090 cards despite growing user concerns about connector safety and performance stability. The emergence of third-party monitoring solutions proves the demand for greater transparency regarding power delivery characteristics, particularly for cards operating at extreme power limits. The RTX 5090, with its factory power limit of 600 W, represents the most power-hungry GPU, getting its massive power through a single 12V-2x6 connector interface.

Recent engineering prototype leaks reveal NVIDIA's power delivery design for its RTX 5090 flagship GPU featured a jaw-dropping quad 12V-2x6 connector configuration. The prototype board, while not publicly photographed, is significantly different from the final version we have today. The quad-connector design serves multiple purposes beyond raw power delivery. Each 12V-2x6 connector, capable of delivering up to 600 W individually, creates a theoretical power ceiling far beyond reasonable requirements. More likely, the implementation allows for separate voltage rails and enhanced power stability, crucial for the card's reported 575 W TDP - a 125 W increase from the RTX 4090's specifications. This overengineered piece follows NVIDIA's historical engineering practices, where early prototypes typically feature expanded power delivery systems that are later optimized for production.

The timing of this leak happens just in time with ongoing discussions about 12V-2x6 connector reliability, particularly following documented failures. While the quad-connector design may seem excessive, it distributes power load more evenly across multiple connection points, potentially addressing thermal concentration issues that plagued single-connector implementations. Board partners like ASUS have already implemented sophisticated current monitoring solutions in their custom designs, using shunt resistors to detect overcurrent conditions across individual pins. Established hardware leaker, Panzerlied confirmed the prototype's existence on Chinese forums, alongside images of other engineering samples featuring similar power delivery experiments across multiple GPU generations.

ZOTAC's website has been updated with technical specifications for its GeForce RTX 5090 GPU-based custom models—the Hong Kong-based hardware company is perhaps the first NVIDIA board partner to publicly disclose these details. According to VideoCardz, this discovery was made by a loyal reader—product pages for SOLID, SOLID OC and AMP Extreme INFINITY seemed to have appeared online within the past couple of days. A resultant report suggests that Team Green has only recently communicated (potentially) finalized specs with its AIBs.

The aforementioned discoverer of ZOTAC's GeForce RTX 5090 spec sheets noticed an unusual memory clock figure on the SOLID OC model's listing—30 Gbps instead of 28 Gbps. This was an error—ZOTAC has since amended that particular data point (see VideoCardz's screenshots below)—their presumably more expensive AMP Extreme INFINITY card's memory clock spec is set at 28 Gbps. Interestingly, ZOTAC's upcoming flagship is the first example of an NVIDIA GPU configured with a power consumption rating of 600 W. It is not immediately apparent whether this TDP figure is an out-of-the-box default—VideoCardz reckons that the GeForce RTX 5090 AMP Extreme INFINITY will arrive with dual-BIOS functionality. A high-performance mode could be user selected. Will rival flagship GeForce RTX 5090 custom cards roll out with similar TDPs? TechPowerUp anticipates the emergence of pre-launch technical details—from other brands/manufacturers—over the next two weeks.

ASRock is presenting its complete range of power supply units at CES 2025, featuring the Taichi, Phantom Gaming, Steel Legend, and Challenger PSU series. Fully compliant with Intel ATX 3.1 and PCIe 5.1 standards, these PSUs can handle peak power loads ranging from 200% to 235%. They are equipped with 12 V 2x6 cables, capable of delivering up to 600 W through a single cable. Additional highlights include a 5 V BOOST function for improved stability, 100% Japanese capacitors, Infineon CoolMOS, and SiC SBD technology.

The ASRock Taichi series power supply unit is certified for 80 Plus and Cybenetics Titanium efficiency and also received Cybenetics LAMBDA A+ ultra-quiet certification. These PSUs are available in 1650 W and 1300 W models, feature fully modular cables, two native PCIe 12 V 2x6 cables, all-Japanese capacitors, Infineon CoolMOS, SiC SBD, NTC sensor to monitors connector temperatures in real-time and are backed by a 10-year warranty.

NVIDIA's unannounced GeForce RTX 5090 graphics card has leaked, confirming key specifications of the next-generation GPU. Thanks to exclusive information from VideoCardz, we can see the packaging of Inno3D's RTX 5090 iChill X3 model, which confirms that the graphics card will feature 32 GB of GDDR7 memory. The leaked materials show that Inno3D's variant will use a 3.5-slot cooling system, suggesting significant cooling requirements for the flagship card. According to earlier leaks, the RTX 5090 will be based on the GB202 GPU and include 21,760 CUDA cores. The card's memory system is a significant upgrade, with its 32 GB of GDDR7 memory running on a 512-bit memory bus at 28 Gbps, capable of delivering nearly 1.8 TB/s of bandwidth. This represents twice the memory capacity of the upcoming RTX 5080, which is expected to ship with 16 GB capacity but 30 Gbps GDDR7 modules.

Power consumption has increased significantly, with the RTX 5090's TDP rated at 575 W and TGP of 600 W, marking a 125-watt increase over the previous RTX 4090 in raw TDP. NVIDIA is scheduled to hold its CES keynote today at 06:30 pm PT time, where the company is expected to announce several new graphics cards officially. The lineup should include the RTX 5090, RTX 5080, RTX 5070 Ti, RTX 5070, and an RTX 5090D model specifically for the Chinese market. Early indications are that the RTX 5080 will be the first card to reach consumers, with a planned release date of January 21st. Release dates for other models, including the flagship RTX 5090, have not yet been confirmed. The RTX 5090 is currently the only card in the RTX 50 series planned to use the GB202 GPU. Pricing information and additional specifications are expected to be revealed during the upcoming announcement.

According to two of the most accurate leakers, kopite7kimi and hongxing2020, NVIDIA's GeForce RTX 5090 and RTX 5080 will feature 575 W and 360 W TDP, respectively. Previously, rumors have pointed out that these GPU SKUs carry 600 W and 400 W TGPs, which translates into total graphics power, meaning that an entire GPU with its RAM and everything else draws a certain amount of power. However, TDP (thermal design power) is a more specific value attributed to the GPU die or the specific SKU in question. According to the latest leaks, 575 Watts are dedicated to the GB202-300-A1 GPU die in the GeForce RTX 5090, while 25 Watts are for GDDR7 memory and other components on the PCB.

For the RTX 5080, the GB203-400-A1 chip is supposedly drawing 360 Watts of power alone, while 40 Watts are set aside for GDDR7 memory and other components in the PC. The lower-end RTX 5080 uses more power than the RTX 5090 because its GDDR7 memory modules reportedly run at 30 Gbps, while the RTX 5090 uses GDDR7 memory modules with 28 Gbps speeds. Indeed, the RTX 5090 uses more modules or higher capacity modules, but the first-generation GDDR7 memory could require more power to reach the 30 Gbps threshold. Hence, more power is set aside for that. In future GDDR7 iterations, more speed could be easily achieved without much more power.

Here is the first picture of what is very likely the GeForce RTX 5090 "Blackwell," the successor to the RTX 4090 "Ada." The picture, its massive GPU, and layout appear to confirm the weekend's bare PCB leak. The RTX 5090 is powered by the "GB202" silicon, the largest gaming GPU based on the "Blackwell" graphics architecture. The silicon in the picture has the ASIC code "GB202-300-A1." From this ASIC code, we can deduce that the RTX 5090 may not max out the silicon (i.e. enable all SM present on it), as maxed-out NVIDIA ASICs tend to have the variant designation "450."

The "GB202" ASIC is surrounded by sixteen GDDR7 memory chips, which reportedly make the 32 GB memory size of the RTX 5090. The chip count, coupled with the large GPU package size (high pin-count), confirm that the "GB202" features a 512-bit wide memory bus. Assuming a memory speed of 28 Gbps, this memory bus should yield a stellar memory bandwidth of 1,792 GB/s. The GPU and memory are surrounded by the card's 24-phase VRM solution. This draws power from a single 16-pin 12V-2x6 power connector. NVIDIA will likely max out the 600 W continuous power-delivery capability of the connector, and give the card a TGP of around 500-550 W, if not more.

Thanks to the renowned NVIDIA hardware leaker kopite7Kimi on X, we are getting information about the final versions of NVIDIA's first upcoming wave of GeForce RTX 50 series "Blackwell" graphics cards. The two leaked GPUs are the GeForce RTX 5090 and RTX 5080, which now feature a more significant gap between xx80 and xx90 SKUs. For starters, we have the highest-end GeForce RTX 5090. NVIDIA has decided to use the GB202-300-A1 die and enabled 21,760 FP32 CUDA cores on this top-end model. Accompanying the massive 170 SM GPU configuration, the RTX 5090 has 32 GB of GDDR7 memory on a 512-bit bus, with each GDDR7 die running at 28 Gbps. This translates to 1,568 GB/s memory bandwidth. All of this is confined to a 600 W TGP.

When it comes to the GeForce RTX 5080, NVIDIA has decided to further separate its xx80 and xx90 SKUs. The RTX 5080 has 10,752 FP32 CUDA cores paired with 16 GB of GDDR7 memory on a 256-bit bus. With GDDR7 running at 28 Gbps, the memory bandwidth is also halved at 784 GB/s. This SKU uses a GB203-400-A1 die, which is designed to run within a 400 W TGP power envelope. For reference, the RTX 4090 has 68% more CUDA cores than the RTX 4080. The rumored RTX 5090 has around 102% more CUDA cores than the rumored RTX 5080, which means that NVIDIA is separating its top SKUs even more. We are curious to see at what price point NVIDIA places its upcoming GPUs so that we can compare generational updates and the difference between xx80 and xx90 models and their widened gaps.

NVIDIA CEO Jensen Huang never misses an opportunity to remind us that Moore's Law is cooked, and that future generations of logic hardware will only get larger and hotter, or hungrier for power. NVIDIA's next generation "Blackwell" graphics architecture promises to bring certain architecture-level performance/Watt improvements, coupled with the node-level performance/Watt improvements from the switch to the TSMC 4NP (4 nm-class) node. Even so, the GeForce RTX 5090, or the part that succeeds the current RTX 4090, will be a power hungry GPU, with rumors suggesting the need for two 16-pin power inputs.

TweakTown reports that the RTX 5090 could come with two 16-pin power connectors, which should give the card the theoretical ability to pull 1200 W (continuous). This doesn't mean that the GPU's total graphics power (TGP) is 1200 W, but a number close to or greater than 600 W, which calls for two of these connectors. Even if the TGP is exactly 600 W, NVIDIA would want to deploy two inputs, to spread the load among two connectors, and improve physical resilience of the connector. It's likely that both connectors will have 600 W input capability, so end-users don't mix up connectors should one of them be 600 W and the other keyed to 150 W or 300 W.

Phanteks is thrilled to unveil two new products tailored for PC enthusiasts and builders: the AMP GH power supply unit and Polar ST CPU coolers. The AMP GH power supply unit combines advanced technology with user-friendly features, including individual textured cables, PCIE Gen 5.1 GPU support, a 120 mm hydraulic bearing fan, compact ATX size, ATX 3.1 specifications, and hybrid fan control with a zero RPM mode for silent operation under low loads.

The Polar ST CPU coolers offer high performance and exceptional value with sintered powder copper heat pipes, M25-120 fans, efficient design, and sleek aesthetics that fit seamlessly on any motherboard. The series includes the Polar ST4, an entry-level cooler with four heat pipes, and the Polar ST5, featuring an all-black heatsink with five heat pipes for optimal performance.

According to Benchlife.info insiders, NVIDIA is supposedly in the phase of testing designs with various Total Graphics Power (TGP), running from 250 Watts to 600 Watts, for its upcoming GeForce RTX 50 series Blackwell graphics cards. The company is testing designs ranging from 250 W aimed at mainstream users and a more powerful 600 W configuration tailored for enthusiast-level performance. The 250 W cooling system is expected to prioritize compactness and power efficiency, making it an appealing choice for gamers seeking a balance between capability and energy conservation. This design could prove particularly attractive for those building small form-factor rigs or AIBs looking to offer smaller cooler sizes. On the other end of the spectrum, the 600 W cooling solution is the highest TGP of the stack, which is possibly only made for testing purposes. Other SKUs with different power configurations come in between.

We witnessed NVIDIA testing a 900-watt version of the Ada Lovelace AD102 GPU SKU, which never saw the light of day, so we should take this testing phase with a grain of salt. Often, the engineering silicon is the first batch made for the enablement of software and firmware, while the final silicon is much more efficient and more optimized to use less power and align with regular TGP structures. The current highest-end SKU, the GeForce RTX 4090, uses 450-watt TGP. So, take this phase with some reservations as we wait for more information to come out.

During the Vision 2024 event, Intel announced its latest Gaudi 3 AI accelerator, promising significant improvements over its predecessor. Intel claims the Gaudi 3 offers up to 70% improvement in training performance, 50% better inference, and 40% better efficiency than Nvidia's H100 processors. The new AI accelerator is presented as a PCIe Gen 5 dual-slot add-in card with a 600 W TDP or an OAM module with 900 W. The PCIe card has the same peak 1,835 TeraFLOPS of FP8 performance as the OAM module despite a 300 W lower TDP. The PCIe version works as a group of four per system, while the OAM HL-325L modules can be run in an eight-accelerator configuration per server. This likely will result in a lower sustained performance, given the lower TDP, but it confirms that the same silicon is used, just finetuned with a lower frequency. Built on TSMC's N5 5 nm node, the AI accelerator features 64 Tensor Cores, delivering double the FP8 and quadruple FP16 performance over the previous generation Gaudi 2.

The Gaudi 3 AI chip comes with 128 GB of HBM2E with 3.7 TB/s of bandwidth and 24 200 Gbps Ethernet NICs, with dual 400 Gbps NICs used for scale-out. All of that is laid out on 10 tiles that make up the Gaudi 3 accelerator, which you can see pictured below. There is 96 MB of SRAM split between two compute tiles, which acts as a low-level cache that bridges data communication between Tensor Cores and HBM memory. Intel also announced support for the new performance-boosting standardized MXFP4 data format and is developing an AI NIC ASIC for Ultra Ethernet Consortium-compliant networking. The Gaudi 3 supports clusters of up to 8192 cards, coming from 1024 nodes comprised of systems with eight accelerators. It is on track for volume production in Q3, offering a cost-effective alternative to NVIDIA accelerators with the additional promise of a more open ecosystem. More information and a deeper dive can be found in the Gaudi 3 Whitepaper.

LIAN LI Industrial Co. Ltd., a leading chassis and PC accessories manufacturer, premieres the LIAN LI 2024 DIGITAL EXPO at 9 am EST on YouTube. In this 2-part video, LIAN LI shares updates on prototype cases, a standing desk, and unveils new concepts for an all-in-one cooler and power supply. The prototypes include the SUP-01, the case that experiments with a front GPU layout, the DAN Cases A3 M-ATX case, and the O11 VISION BACK CONNECT. LIAN LI also unveils the GALAHAD II LCD SHIFT, an AIO that minimizes the tubing, the EDGE Series PSU, a new form factor power supply that solves cable access in dual chamber cases, and the DK-07, a new standing desk with new features. The LIAN LI 2024 DIGITAL EXPO will be followed by a roundtable discussion hosted by Jameson Chen, CEO at LIAN LI, with YouTube influencers Stuart Tonks from GGF Events, Justin Robey from Robytech, and PCMR's founder Pedro Valadas.

The SUP-01, a compact 45-liter case, reimagines the PC layout with a front-positioned graphics card and innovative cooling. Equipped with airflow-optimized mesh panels, it surpasses the LANCOOL 216 in cooling performance. The GPU's unique placement optimizes cooling without sacrificing aesthetics and is supported by adjustable brackets. The airflow configuration strategically utilizes GPU fans and rear fans as intakes, and right-side fans as exhausts, ensuring optimal GPU and CPU cooling. The main chamber accommodates motherboard sizes up to ATX, pre-mounted PCIe 4.0 riser cable, and features clearance for 280 to 360 mm AIOs. Storage options include two SSDs or 3.5" HDDs, a removable drive cage, and magnetic dust filters to maintain system cleanliness. The price is set to be $149.99.