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.

YEYIAN at the 2024 International CES displayed some really wacky cooling and power solutions. Over in the cooling front, was got to see a super rare 480 mm AIO. This cooler has a 480 mm x 120 mm radiator, which requires four 120 mm in a row. We're not sure if there are any cases that support this radiator size, but we're sure it could become a thing. It would be interesting to compare the cooling performance of a 480 mm x 120 mm AIO with that of a 420 mm x 140 mm (a much more popular form-factor). YEYIAN's AIO coolers come in three sizes—360 mm x 120 mm, 480 mm x 120 mm, and 420 mm x 140 mm. The pump-blocks feature a 2.1-inch IPS LCD display with 480 x 480 pixels resolution, which can be programmed to display anything. Each of the included fans turn at speeds of up to 2600 RPM, pushing up to 72 CFM of airflow, with as high as 23 dBA noise.

Next up, YEYIAN showed off its Orion series full modular ATX 3.1 power supplies. These meet the latest ATX 3.1 and PCIe Gen 5 spec; and offer a native 12V-2x6 power connector, the successor to 12VHPWR with more mechanical durability, which can deliver 600 W of continuous power. The YEYIAN Orion series comes in 850 W, 1300 W, and 1600 W capacities, with the 1300 W and 1600 W models meeting 80 Plus Platinum efficiency; while the 850 W model meets 80 Plus Gold. All three are equipped with a 600 W-capable 12V-2x6; the top 1600 W offers two of these in case you have one of those crazy GALAX HOF RTX 4090 cards.

The theme at be quiet! this CES is all about RGB lighting and white color variants. This is a big aesthetic departure from be quiet! which tends to have understated, mostly black product styling, with minimal use of RGB. Ah well, as long as it's all quiet. Among the cases, we saw all-white editions of the Dark Base 701, and Dark Base Pro 901 cases. The Dark Base 701 white retains its essential design and functionality, but now gets two addressable RGB silicone diffusers along the sides of the front panel. The interiors are all white, as are the included fans. The Dark Base Pro 901 white lacks any such RGB lighting elements, it's more of just a whitewash.

As we said, be quiet! is betting big on RGB and white color scheme this year, and has unveiled white variants of its Silent Wings 4, Light Wings, and Pure Wings 3 fan series, across all available sizes. These see white take over the frames and impellers. be quiet! collaborated with SignalRGB, the experts in elaborate RGB lighting, to create an art piece—a wall of 192 Light Wings White fans, all controlled through a single machine. The fans work like a mosaic to create actual images when viewed from a distance. Very creative!

The SG10 is an ambitious product designed to push the boundaries of fanless technology and create a gaming case capable of cooling high-performance CPUs and GPUs without the use of fans. This project collaboration between Streacom and Calyos is also aimed at showcasing the application of loop heat pipe technology in PC cooling and represents the first step in creating a viable mainstream high-performance solid-state alternative to water cooling solutions.

The spiritual successor to the SG10 began life as a Kickstarter project launched by Calyos as the NSG S0, and whilst having extensive expertise in cooling, the additional challenges of consumer product manufacturing proved to be a stumbling block. A search ensued to find a partner with suitable experience to get the project back on track, and with experience in both case manufacturing and fanless cooling, Streacom was the natural fit. As can be understood just by glancing at the image, the SG10 represents a complete redesign of the NSG S0. Not a single asset or component was carried over from the original case. This includes the Evaporator that pumps the coolant around the loop and the Condenser that radiates the heat, both core components that make this performance possible. Everything has been entirely re-engineered and improved far beyond the original specification and performance.

Twitter/X user @momomo_us has unveiled official documentation detailing a new GC-HPCE power connection standard to supply additional GPU power throughout motherboard. This connector can deliver over 600 Watts of power, surpassing 12VHPWR connector's capacity. The GC-HPCE power connector features four groups of pins. Sixteen pins are dedicated to power delivery, while the remaining twelve facilitate communication between the motherboard and the graphics card. Its size is comparable to the regular PCIe x1 connector. Positioned in alignment with the primary PCIe x16 slot on a motherboard, the GC-HPCE power connector is situated behind the x16 slot, typically where the motherboard chipset heatsink is found. This strategic placement ensures the graphics card can be easily slotted into both connectors, simplifying installation. The connector's design negates the need for an additional locking mechanism, as the x16 slot's existing lock sufficiently secures the graphics card.

First showcased at Computex, this connector was featured in several prototype motherboards and graphics cards, emphasizing its potential to enhance cable management and aesthetics. By eliminating supplementary power cables from the graphics card and channeling power through the motherboard, the graphics card's appearance remains uncluttered, and cable management behind the motherboard tray becomes more streamlined. Intriguingly, this connector is a familiar design. It's a modified version of the High Power Card Edge (HPCE) standard prevalent in the server industry. While ASUS has been the primary proponent of this connector, it remains to be seen if other companies will adopt this standard for their consumer products. Additionally, routing power throughout the motherboard will require manufacturers to include additional power regulating circuitry, potentially driving motherboard costs up. It is also not an agreed upon industry standard, which could block some future GPU upgrades from happening.You can check out ASUS'es implementation of the Back-to-Future (BTF) motherboard that supports this connector below.

Channel Well Technology, or CWT in short, is a major OEM of PC power supplies and notebook power bricks. Several major PSU brands source their PSUs from CWT, across all market segments. The company apparently also sells certain reference PSUs of its own in the retail channel, the way Seasonic and Fortron (FSP) do, and showcased its latest ATX 3.0 + PCIe Gen 5 compliant PSU models. The CTT series of enthusiast-segment reference PSUs comes in 1650 W and 1350 W variants, and boast of 80 Plus Titanium efficiency, and two 16-pin 12VHPWR connectors, each capable of 600 W continuous output.

The CTB Series are the company's top SFX-L PSU models, and meet SFX12V 4.0 (ATX 3.0 analogue) standards, along with PCIe Gen 5. These come in a 1000 W variant with 80 Plus Gold efficiency, and a 1200 W variant with 80 Plus Platinum, both feature a 16-pin 12VHPWR connector. These use 120 mm fans. The CSX Series PSUs target the traditional SFX form-factor, and come in two high-wattage models—1000 W with 80 Plus Platinum efficiency, and 850 W with 80 Plus Gold. Both models offer a 600 W-rated 16-pin 12VHPWR connector.

SilverStone showcased a large selection of new and upcoming PSUs at Computex 2023. The star attraction at their booth was the HELA 2050R Platinum (model: SST HA2050R PM), which is, well, hela powerful. This 2050 W PSU comes with two 16-pin 12VHPWR connectors each capable of 600 W continuous output, besides close to a dozen 6+2 pin PCIe power connectors. It includes a flattened, ribbon-type 12VHPWR cable that the company claims is "super flexible." The PSU meets both ATX 3.0 and PCIe Gen 5 standards. It meets Cybenetics Platinum switching efficiency. SilverStone plans to launch the HELA 2050R Platinum toward the end of June, 2023. This model will be joined by the HELA 1650R (SST HA1650R PM) later this year. This 1650 W model retains much of the feature-set of its 2050 W sibling, but in a lower depth body. You still get two 16-pin 12VHPWR connectors, the same Cybenetics Platinum switching efficiency, and those flat black 12VHWPR flexible cables. SilverStone plans to launch the HELA 1650R in September 2023.

be quiet!, a German company focusing on creating PC components and accessories, had a strong presence at Computex 2023. Showing off its cases and power supplies, the company introduced a few additions to its portfolio of offerings. For starters, be quiet! has introduced a new case series called Shadow Base. The new Shadow Base lineup is focused on maximum airflow and maximum performance, where the company has made a new open mesh design that allows for "exceptionally high airflow." Inaugurating model of the series is the Shadow Base 800, which comes in three flavors: standard, DX, and FX. The standard edition is just a regular case without flashy RGB; the DX model has an RGB strip, while the FX model is tailored for maximum RGB illumination across the case. With up to E-ATX motherboard support, the case supports three or four 140 mm fans, depending on the variant. You can see the specification sheet below.

As COMPUTEX once again opens its door, MSI, the world leader in gaming, content creation, business & productivity, and AIoT solutions, is all geared up to dominate the show floor with a full slate of new lineup and its Best Choice Award honorees. "We are thrilled to be back at COMPUTEX," said Sam Chern, MSI Vice President of Marketing. "We expect to see users' tech experiences reinvented and MSI is at the forefront. Our COMPUTEX 2023 lineup showcases the ultimate fusion of breakthrough technology and everlasting aesthetics that surpasses all limits and indulges global users with the best possible experience like never before."

This year, MSI themes its booth around The Leap to Singularity, envisioning a future in which the technological singularity brings forth superintelligence to improve human lives. Among the highlights on display will be the Raider GE78 HX Smart Touchpad, Stealth 16 and Creator Z17 laptops powered by up to 13th Gen Intel Core i9 processors and NVIDIA GeForce RTX 40 Series graphics, making them powerhouses for gamers and creators alike.

While NVIDIA has launched high-end GeForce RTX 4090 and RTX 4080 GPUs from its Ada Lovelace family, middle and lower-end products are brewing to satisfy the entire consumer market. Today, according to the kopite7kimi, a well-known leaker, we have potential information about the configuration of the upcoming GeForce RTX 4060 Ti graphics card. Featuring 4352 FP32 CUDA cores, the GPU is powered by an AD106-350-A1 die. On the die, there is 32 MB of L2 cache. To pair, it has 8 GB of GDDR6 18 Gbps memory, which should be enough to power games at 1440p resolution, which this card is aiming for.

The design of the cards reference PG190 PCB is supposedly very short, making it ideal for ITX-sized designs we could see from NVIDIA's AIB partners. Interestingly, with a TDP of 220 Watts, the reference card is powered by the infamous 16-pin 12VHPWR connector, capable of supplying 600 Watts of power. This choice of connector is unclear; however, it could be NVIDIA's push to standardize its usage across all products in the Ada Lovelace family stack. While the card should not need the full potential of the connector, it signals that the company could only be using this type of connector for all of its future designs.

The innovative manufacturer of high-end multi-GPU workstations and professional liquid cooling solutions for individuals and workgroups requiring GPU computation, announces a slim waterblock of top-tier efficiency for GIGABYTE GeForce RTX 4090 GAMING OC 24 GB and AORUS RTX 4090 MASTER 24G GPUs.

Given the fact that the latest 4090 GPUs by Nvidia are extremely power hungry, the Comino team did its best to deliver the highest performance 4090 waterblock design, specifically adding:

Everyone's favorite GPU-news leaker Kopite7kimi has updated his tweet from April 1 with more specific board part numbers, and suddenly the information contained there—which could have been misinterpreted as an April 1st joke—now all seems to add up with our own posting from last month about memory bus widths. The update seems to indicate the boards will now feature 21 Gbps memory, which is the same as what we saw on the recently made available RTX 3090 Ti cards, and Videocardz goes further to speculate the 3090 Ti could have been a dry run for the upcoming cards, including with a similar 600 W TDP rating to follow. Note also that the leaker is shying away from referring to these as the RTX 4080/4090 series, leaving room in case NVIDIA decides to jump in naming scheme for reasons including marketing and what the competition decides.

After the data center-oriented Hopper architecture launch, NVIDIA is slowly preparing to transition the consumer section to new, gaming-focused designs codenamed Ada Lovelace. For starters, the source claims that NVIDIA is using the upcoming GeForce RTX 3090 Ti GPU as a test run for the next-generation Ada Lovelace AD102 GPU. Thanks to the authorities over at Igor's Lab, we have some additional information about the upcoming lineup. We have a sneak peek of a few features regarding the top-end GeForce RTX 4080 and RTX 4090 GPU SKUs. According to Igor's claims, NVIDIA is testing the PCIe Gen5 power connector and wants to see how it fares with the biggest GA102 SKU - GeForce RTX 3090 Ti.

Additionally, we find that the AD102 GPU is supposed to be pin-compatible with GA102. This means that the number of pins located on GA102 is the same as what we are going to see on AD102. There are 12 places for memory modules on the AD102 reference design board, resulting in up to 24 GB of GDDR6X memory. As much as 24 voltage converters surround the GPU, NVIDIA will likely implement uP9512 SKU. It can drive eight phases, resulting in three voltage converters per phase, ensuring proper power delivery. The total board power (TBP) is likely rated at up to 600 Watts, meaning that the GPU, memory, and power delivery combined output 600 Watts of heat. Igor notes that board partners will bundle 12+4 (12VHPWR) to four 8-pin (PCIe old) converters to enable PSU compatibility.

During the International Solid-State Circuits Conference (ISSCC) 2022, Intel gave us a more significant look at its upcoming Ponte Vecchio HPC accelerator and how it operates. So far, Intel convinced us that the company created Ponte Vecchio out of 47 tiles glued together in one package. However, the ISSCC presentation shows that the accelerator is structured rather interestingly. There are 63 tiles in total, where 16 are reserved for compute, eight are used for RAMBO cache, two are Foveros base tiles, two represent Xe-Link tiles, eight are HBM2E tiles, and EMIB connection takes up 11 tiles. This totals for about 47 tiles. However, an additional 16 thermal tiles used in Ponte Vecchio regulate the massive TDP output of this accelerator.

What is interesting is that Intel gave away details of the RAMBO cache. This novel SRAM technology uses four banks of 3.75 MB groups total of 15 MB per tile. They are connected to the fabric at 1.3 TB/s connection per chip. In contrast, compute tiles are connected at 2.6 TB/s speeds to the chip fabric. With eight RAMBO cache tiles, we get an additional 120 MB SRAM present. The base tile is a 646 mm² die manufactured in Intel 7 semiconductor process and contains 17 layers. It includes a memory controller, the Fully Integrated Voltage Regulators (FIVR), power management, 16-lane PCIe 5.0 connection, and CXL interface. The entire area of Ponte Vecchio is rather impressive, as 47 active tiles take up 2,330 mm², whereas when we include thermal dies, the total area jumps to 3,100 mm². And, of course, the entire package is much larger at 4,844 mm², connected to the system with 4,468 pins.

As you're most likely aware of, NVIDIA introduced a new power connector with its RTX 30-series Founders Edition graphics cards and at the time it was something of a controversy, especially as none of its AIB partners went for the connector. As it turned out, the connector was largely accepted by the PCI-SIG, with a few additions which lead to the 12VHPWR connector. The main difference between the two was the addition of a small set of sense connectors, for a 12+4-pin type connector. It has now been confirmed that the 12VHPWR will work with NVIDIA's Founders Edition cards, although this isn't a huge surprise as such, but rather good news for those that happen to own a Founders Edition card and are looking to invest in a new PSU.

However, what's more interesting in the news about the 12VHPWR connector is that it will operate in two distinct modes. If the 4-pin sense connector isn't connected to the GPU, the PSU will only deliver 450 Watts to the GPU, presumably as some kind of safety precaution. On the other hand, if the sense connector is used, the same cable can deliver up to 600 Watts, which would allow for a combined card power draw of up to 675 Watts for next generation GPUs. It's possible that we'll see cards with multiple power thresholds that will be negotiated on the fly with the PSU and we might also see PSU's that can force a lower power state of the GPU in case the overall system load gets too high. It'll be interesting to see what the new standard delivers, since so far not a lot of details have been released with regards to how the sense function works in detail.