At the heart of the GeForce RTX 4090 is the gigantic AD102 silicon, which we broadly detailed in an older article. Built on the 4 nm silicon fabrication process, this chip measures 608 mm² in die-area, and crams in 76.3 billion transistors. We now have our first look into the silicon-level block diagram of the AD102, including the introduction of several new components.

The AD102 features a PCI-Express 4.0 x16 host interface, and a 384-bit GDDR6X memory interface. The Gigathread Engine acts as a the main resource allocation component of the silicon. Ada introduces the Optical Flow Accelerator, a component crucial for DLSS 3 to generate entire frames without involving the graphics rendering machinery. The chip features double the number of media-encoding hardware engines as "Ampere," including hardware-accelerated AV1 encode/decode. Multiple accelerators mean that multiple streams of videos can be transcoded (helpful in a media production environment), or transcoding is performed at twice the FPS rate (each encoder takes turns at encoding a single frame).

During its GTC 2022 session, NVIDIA introduced its new generation of gaming graphics cards based on the novel Ada Lovelace architecture. Dubbed NVIDIA GeForce RTX 40 series, it brings various updates like more CUDA cores, a new DLSS 3 version, 4th generation Tensor cores, 3rd generation Ray Tracing cores, and much more, which you can read about here. However, today, we also got a new Ada Lovelace card intended for the data center. Called the L40, NVIDIA updated its previous Ampere-based A40 design. While the NVIDIA website provides sparse, the new L40 GPU uses 48 GB GDDR6 memory with ECC error correction. Using NVLink, you can get 96GBs of VRAM. Paired with an unknown SKU, we assume that it uses AD102 with adjusted frequencies to lower the TDP and allow for passive cooling.

NVIDIA is calling this their Omniverse GPU, as it is a part of the push to separate its GPUs used for graphics and AI/HPC models. The "L" model in the current product stack is used to accelerate graphics, with display ports installed on the GPU, while the "H" models (H100) are there to accelerate HPC/AI installments where visual elements are a secondary task. This is a further separation of the entire GPU market, where the HPC/AI SKUs get their own architecture, and GPUs for graphics processing are built on a new architecture as well. You can see the specifications provided by NVIDIA below.

The AD102 silicon on which NVIDIA's new flagship graphics card, the GeForce RTX 4090, is based, is a marvel of semiconductor engineering. Built on the 4 nm EUV (TSMC 4N) silicon fabrication process, the chip has a gargantuan transistor-count of 76.3 billion, a nearly 170% increase over the previous GA102, and a die-size of 608 mm², which is in fact smaller than the 628 mm² die-area of the GA102. This is thanks to TSMC 4N offering nearly thrice the transistor-density of the Samsung 8LPP node on which the GA102 is based.

The AD102 physically features 18,432 CUDA cores, 568 fourth-generation Tensor cores, and 142 third-generation RT cores. The streaming multiprocessors (SM) come with special components that enable the Shader Execution Reordering optimization, which has a significant performance impact on both raster- and ray traced graphics rendering performance. The silicon supports up to 24 GB of GDDR6X or up to 48 GB of GDDR6+ECC memory (the latter will be seen in the RTX Ada professional-visualization card), across a 384-bit wide memory bus. There are 568 TMUs, and a mammoth 192 ROPs on the silicon.

NVIDIA has been dropping a trail of breadcrumbs in its GeForce twitter feed over the past week surrounding "Project Beyond," which from the get go was interpreted as its debut of the GeForce RTX 40-series, based on the "Ada" graphics architecture (named after Ada Lovelace, credited with being the mother of computer programming). In its latest tweet, the company posted a picture of the famous "Diagram for the computation by the Engine," believed to be a computer program she wrote for the Charles Babbage number engine (an analog computer from the early 19th century).

The teaser video shows a copy of Lovelace's chart lying on a modern PC desk. One of the two monitors has a sticky note with a phone number scribbled on it—(208)-629-7538. Punters on social media decoded this not to be a random phone number, but a well-crafted hint about the vital specs of the top GeForce RTX 4090. "208" could denote 2.08 times the performance (or at least FP32 compute throughput) of the RTX 3090 "Ampere." 629 could be the die-size in mm², and 7538 could mean 75.38 billion transistors (the rumored transistor count of the 5 nm "AD102" silicon on which the RTX 4090 is based).

NVIDIA's next-generation AD102 "Ada" GPU is shaping up to be a monstrosity, with a rumored transistor-count north of 75 billion. This would put over 2.6 times the 28.3 billion transistors of the current-gen GA102 silicon. NVIDIA is reportedly building the AD102 on the TSMC N5 (5 nm EUV) node, which offers a significant transistor-density uplift over the Samsung 8LPP (8 nm DUV) node on which the GA102 is built. The 8LPP offers 44.56 million transistors per mm² die-area (MTr/mm²), while the N5 offers a whopping 134 MTr/mm², which fits in with the transistor-count gain. This would put its die-area in the neighborhood of 560 mm². The AD102 is expected to power high-end RTX 40-series SKUs in the RTX 4090-series and RTX 4080-series.

NVIDIA's next-generation GeForce RTX 4090 "Ada" flagship graphics card allegedly scores over 19000 points in the 3DMark Time Spy Extreme synthetic benchmark, according to kopite7kimi, a reliable source with NVIDIA leaks. This would put its score around 66 percent above that of the current RTX 3090 Ti flagship. The RTX 4090 is expected to be based on the 5 nm AD102 silicon, with a rumored CUDA core count of 16,384. The higher IPC from the new architecture, coupled with higher clock speeds and power limits, could be contributing to this feat. Time Spy Extreme is a traditional DirectX 12 raster-only benchmark, with no ray traced elements. The Ada graphics architecture is expected to reduce the "cost" of ray tracing (versus raster-only rendering), although we're yet to see leaks of RTX performance, yet.

Rumors of NVIDIA's upcoming Ada Lovelace graphics cards keep appearing. With every new update, it seems like the total power consumption is getting bigger, and today we are getting information about different SKUs, including mobile and desktop variants. According to a well-known leaker, kopite7kimi, we have information about the power limits of the upcoming GPUs. The new RTX 40 series GPUs will feature a few initial SKUs: AD102, AD103, AD104, and AD106. Every SKU, except the top AD102, will be available as well. The first in line, AD102, is the most power-hungry SKU with a maximum power limit rating of 800 Watts. This will require multiple power connectors and a very beefy cooling solution to keep it running.

Going down the stack, we have an AD103 SKU limited to 450 Watts on desktop and 175 Watts on mobile. The AD104 chip is limited to 400 Watts on desktop, while the mobile version is still 175 Watts. Additionally, the AD106 SKU is limited to 260 Watts on desktop and 140 Watts on mobile.

That NVIDIA's upcoming GeForce "Ada Lovelace" graphics architecture will feature significant increases in board power across the lineup, is widely reported, with the top AD102 silicon as the RTX 3090-successor allegedly drawing over 400 W. All that power drawn converts to heat that needs to be dissipated, and so we'll see proportionately larger cooling solutions. Kopite7kimi, a reliable source with NVIDIA rumors, claims that the reference design AD102 board features three fans.

To be clear, the reference design board is not the same thing as the Founders Edition product. The reference-design is an internal prototype NVIDIA develops the product around, and it rarely makes public appearances. The Founders Edition is a technically a "custom design" card from NVIDIA, that features higher clock speeds than reference, and is known for a high level for industrial design that sets standards for board partners and their custom-designs. We know from a May 2022 rumor and alleged photo leak that the Dual-Axial Flow-through concept of air-based cooling solutions, will make a comeback with Ada Founders Edition cards. and its design appears to have the usual layout of two large fans; unless NVIDIA somehow found room for a third fan. We'll know soon enough.

An alleged technical drawing of the PCB of reference-design NVIDIA "Ada" AD102 silicon was leaked to the web, courtesy of Igor's Lab. It reveals a large GPU pad that's roughly the size of the GA102 (the size of the fiberglass substrate or package, only, not the die); surrounded by twelve memory chips, which are likely GDDR6X. There are also provision for at least 24 power phases, although not all of them are populated by sets of chokes and DrMOS in the final products (a few of them end up vacant).

We also spy the 16-pin ATX 3.0 power connector that's capable of delivering up to 600 W of power; and four display outputs, including a USB-C in lieu of a larger connector (such as DP or HDMI). A curious thing to note is that the card continues to have an NVLink connector. Multi-GPU is dead, which means the NVLink on the reference design will likely be rudimentary in the GeForce RTX product (unless used for implicit multi-GPU). The connector may play a bigger role in the professional-visualization graphics cards (RTX AD-series) based on this silicon.

NVIDIA's next-generation GeForce RTX 40 series of graphics cards, codenamed Ada Lovelace, is shaping up to be a powerful graphics card lineup. Allegedly, we can expect to see a mid-July launch of NVIDIA's newest gaming offerings, where customers can expect some impressive performance. According to a reliable hardware leaker, kopite7kimi, NVIDIA GeForce RTX 4090 graphics card will feature AD102-300 GPU SKU. This model is equipped with 126 Streaming Multiprocessors (SMs), which brings the total number of FP32 CUDA cores to 16128. Compared to the full AD102 GPU with 144 SMs, this leads us to think that there will be an RTX 4090 Ti model following up later as well.

Paired with 24 GB of 21 Gbps GDDR6X memory, the RTX 4090 graphics card has a TDP of 450 Watts. While this number may appear as a very power-hungry design, bear in mind that the targeted performance improvement over the previous RTX 3090 model is expected to be a two-fold scale. Paired with TSMC's new N4 node and new architecture design, performance scaling should follow at the cost of higher TDPs. These claims are yet to be validated by real-world benchmarks of independent tech media, so please take all of this information with a grain of salt and wait for TechPowerUp reviews once the card arrives.

A technological race is brewing between NVIDIA and AMD over which brand's GPU reaches the 100 TFLOP/s peak FP32 throughput mark first. AMD's TeraScale graphics architecture and the "RV770" silicon, were the first to hit the 1 TFLOP/s mark, way back in 2008. It would take 14 years for this figure to reach 100 TFLOP/s for flagship GPUs. NVIDIA's next generation big GPU based on the "Ada Lovelace," the AD102, is the green team's contender for the 100 TFLOP/s mark, according to kopite7kimi. To achieve this, all 144 streaming multiprocessors (SM) or 18,432 CUDA cores, of the AD102 will have to be enabled.

From the red team, the biggest GPU based on the next-generation RDNA3 graphics architecture, "Navi 31," could offer peak FP32 throughput of 92 TFLOP/s according to greymon55, which gives AMD the freedom to create special SKUs running at high engine clocks, just to reach the 100 TFLOP/s mark. The Navi 31 silicon is expected to triple the compute unit count over its predecessor, resulting in 15,360 stream processors. Both the AD102 and Navi 31 are expected to be built on the same TSMC N5 (5 nm EUV) node, and product launches for both are expected by year-end.

With the release of Hopper, NVIDIA's cycle of new architecture releases is not yet over. Later this year, we expect to see next-generation gaming architecture codenamed Ada Lovelace. According to a well-known hardware leaker for NVIDIA products, @kopite7kimi, on Twitter, the green team is reportedly testing a potent variant of the upcoming AD102 SKU. As the leak indicates, we could see an Ada Lovelace AD102 SKU with a Total Graphics Power (TGP) of 900 Watts. While we don't know where this SKU is supposed to sit in the Ada Lovelace family, it could be the most powerful, Titan-like design making a comeback. Alternatively, this could be a GeForce RTX 4090 Ti SKU. It carries 48 GB of GDDR6X memory running at 24 Gbps speeds alongside monstrous TGP. Feeding the card are two 16-pin connectors.

Another confirmation from the leaker is that the upcoming RTX 4080 GPU uses the AD103 SKU variant, while the RTX 4090 uses AD102. For further information, we have to wait a few more months and see what NVIDIA decides to launch in the upcoming generation of gaming-oriented graphics cards.

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.

The deluge of NVIDIA leaks continue following the major cyber-attack on the company, with hackers getting away with sensitive information about current and upcoming products. The latest in this series covers the memory bus widths of the next-generation RTX 40-series GPUs based on the "Ada Lovelace" graphics architecture. There is early-information covering the streaming multiprocessor (SM) counts of each GPU, and their large on-die caches.

The top-of-the-line AD102 silicon allegedly has a 384-bit wide memory bus, similar to its predecessor. The next-best AD103 silicon has a 256-bit wide memory bus. Things get very interesting with the AD104, which has a 192-bit wide memory bus. The AD104 is a revelation here, because it succeeds a long line of NVIDIA GPUs with 256-bit memory buses (eg: GA104, TU104, GP104, GM204, etc). This confirms the theory that much like AMD, NVIDIA is narrowing the memory bus widths in the lower segments to cut board costs, and compensate for the narrower bus-width with large on-die caches, high memory data-rates, and other memory-management optimizations.

Possible streaming multiprocessor (SM) counts of the various NVIDIA "Ada Lovelace" client-graphics GPUs surfaced, allegedly pieced together from code seen in the recent NVIDIA cyberattack data-leak. According to this, the top-dog "AD102" silicon has 144 SM, the next-best "AD103" has 84. The third-largest "AD104" silicon has 60. The performance-segment "AD106" has 36, and the mainstream "AD107" has 24. Assuming the number of CUDA cores per SM in the "Ada Lovelace" graphics architecture is unchanged from that of "Ampere," we're looking at 18,432 CUDA cores for the "AD102," an impressive 10,752 for the "AD103," 7,680 cores for the "AD104," 4,608 for the "AD106," and 3,072 for the "AD107."

The recent NVIDIA data-leak springs up information on various upcoming graphics parts. Besides "Ada Lovelace," "Hopper," we come across a new codename, "Blackwell." It turns out that NVIDIA is splitting the the graphics and compute architecture naming with the next generation, not unlike what AMD did, with its RDNA and CDNA series. The current "Ampere" architecture is being used both for compute and graphics, with the streaming multiprocessor for the two being slightly different—the compute "Ampere" has more FP64 and Tensor components, while the graphics "Ampere" does away with these in favor of RT cores and graphics-relevant components.

The graphics architecture to succeed GeForce "Ampere" will be GeForce "Ada Lovelace." GPUs in this series are identified in the leaked code as "AD102," "AD103," "AD104," "AD106," "AD107," and "AD10B," succeeding a similar numbering for parts with the "A" (GeForce Ampere) series. The compute architecture succeeding "Ampere" will be codenamed "Hopper." with parts in the series being codenamed "GH100" and "GH202." Another compute or datacenter architecture is "Blackwell," with parts being codenamed "GB100" and "GB102." From all accounts, NVIDIA is planning to launch the GeForce 40-series "Ada" graphics card lineup in the second half of 2022. The company is in need of a similar refresh for its compute product lineup, and could debut "Hopper" either toward the end of 2022 or next year. "Blackwell" could follow "Hopper."

The rumor mill has begun grinding with details about NVIDIA's next-gen graphics processors based on the "Lovelace" architecture, with Kopite7kimi (a reliable source with NVIDIA leaks) predicting a 71% increase in shader units for the "AD102" GPU that succeeds the "GA102," with 12 GPCs holding 6 TPCs (12 SMs), each. 3DCenter.org extrapolates on this to predict a CUDA core count of 18.432 spread across 144 streaming multiprocessors, which at a theoretical 1.80 GHz core clock could put out an FP32 compute throughput of around 66 TFLOP/s.

The timing of this leak is interesting, as it's only 3 months into the market cycle of "Ampere." NVIDIA appears unsettled with AMD RDNA2 being competitive with "Ampere" at the enthusiast segment, and is probably bringing in its successor, "Lovelace" (after Ada Lovelace), out sooner than expected. Its previous generation "Turing" architecture saw market presence for close to two years. "Lovelace" could leverage the 5 nm silicon fabrication process and its significantly higher transistor density, to step up performance.