According to information provided by Kopite7kimi, the flagship NVIDIA Ada Lovelace GPU SKU is the best graphics processor ever produced.
Such an expensive Ada Lovelace GPU SKU has been discussed previously; it is not the first time you have heard about it. The earlier rumor, which also originated from the same leaker, claimed that the Ada Lovelace GPU lineup would include a Titan-Class graphics card with some
outrageous specifications. Again, the previous version was described as a 900W TBP SKU with the full 18432 CUDA Cores; however, this isn’t the fully activated AD102 GPU.
As per the discussion, the graphics card will be powered by the Ada Lovelace architecture and have the configuration AD102-450-A1, which will have 18,432 CUDA cores and 142 SMs (out of 144 SMs) (out of 18432 CUDA cores). This graphics card will easily surpass the 100 TFLOPs Compute barrier based on a clock speed of about 3 GHz. According to reports, the graphics card has 48 GB of GDDR6X memory connected by a 384-bit bus interface.
It’s surprising that NVIDIA won’t hold back on the VRAM specifications and includes the newest 24 Gbps memory modules, providing the GPU with up to 1.152 TB/s of VRAM bandwidth. Compared to the current RTX 3090 Ti flagship, which has memory dies capable of
21 Gbps, it is a 14 percent improvement in memory bandwidth. The next RTX 4090 is predicted to use the same 21 Gbps memory dies, with the 24 Gbps dies being reserved for the top-tier Ti variant.
The new NVIDIA Flagship AD102 GPU-powered graphics card will consume insane amounts of power, nearly doubling the RTX 3090 Ti’s TDP to up to 800W. If this monster of a card ever becomes a reality, a dual 16-pin connector configuration will be required because a single 16-pin connector can only deliver 600 Watts of power. The graphics card may use the PG137-SKU0.
Based on the Ampere range, we could see that NVIDIA not only chose not to manufacture a Titan graphics card but also effectively replaced it with its BFGPU-class GeForce RTX lineup. The larger capacity cards continued to emerge under the workstation RTX A**** range, which
also received the complete GA102 treatment, although there was no Titan class beyond the RTX 3090 Ti and the RTX A6000. Will a Titan-class GPU be appropriate for Ada Lovelace, or will this particular SKU become the workstation and next-generation gaming BFGPU flagship?
Although we can’t be certain, one thing is certain: such a GPU setup will be insane in terms of its specifications, power consumption, and cost.
NVIDIA GeForce RTX 40 Series Graphics Card Lineup (Rumored):
GRAPHICS CARD | GPU | PCB VARIANT | SM UNITS / CORES | MEMORY / BUS | MEMORY CLOCK / BANDWIDTH | TGP | POWER CONNECTORS | LAUNCH |
---|---|---|---|---|---|---|---|---|
NVIDIA Titan A / GeForce RTX 40? | AD102-450? | PG137-SKU0 | 142/ 18176? | 48 GB / 384-bit | 24 Gbps / 1.15 TB/s | ~800W | 2x 16-pin | TBD |
NVIDIA GeForce RTX 4090 Ti | AD102-350? | TBD | 144 / 18432? | 24 GB / 384-bit | 24 Gbps / 1.15 TB/s | ~600W | 1x 16-pin | TBD |
NVIDIA GeForce RTX 4090 | AD102-300? | PG137/139 SKU330 | 128 / 16384? | 24 GB / 384-bit | 21 Gbps / 1.00 TB/s | ~450W | 1x 16-pin | Q4 2022 |
NVIDIA GeForce RTX 4080 | AD103-300? | PG13*/139 SKU360 | 80 / 10240? | 16 GB / 256-bit | 21 Gbps / 672 GB/s | ~420W | 1x 16-pin | Q4 2022 |
NVIDIA GeForce RTX 4070 | AD104-275? | PG141-310 SKU341 | 56 / 7168? | 10 GB / 160-bit | 18 Gbps / 360 GB/s | ~300W | 1x 16-pin | Q4 2022 |
NVIDIA GeForce RTX 4060 | AD106-***? | TBD | >36 / 4608? | 8 GB / 128-bit | TBD | ~200W | 1 x 16-pin | Q1 2023 |
New Source: wccftech.com, videocardz.com, Twitter: @kopite7kimi