Flagship Graphics Card Nvidia Ada Lovelace AD102 Vs AMD RDNA 3 Navi 31

Nvidia and AMD are currently working on their next generations of GeForce RTX and Radeon RX. New rumors appear. They address several issues, including video memory.

The next family of GeForce RTX graphics cards will not necessarily be called the GeForce RTX 40 series. This name is now “timed” by various indiscretions. Nvidia is likely to adopt the GeForce RTX 50 series name for its generation ada lovelace. kopite7kim please mention these solutions as **90/**80 etc., in case they are not called RTX 40 but RTX 50.

GeForce RTX Ada LoveLace

The new video memory speculation evokes not 24 Gbps but 21 Gbps. We would suddenly find ourselves with a profile identical to the one proposed by the new GeForce RTX 3090 Ti. The thermal envelope must reach 600 Watts. This figure is plausible with respect to the PCIe 5.0 connector installed on the current 3090 Ti.

GeForce RTX “Ada lovelace”.

GPU GA102 AD102
Architecture Amp ada lovelace
gravure 8nm 5nm
Graphics Processing Clusters (GPC) 7 12
Texture Processing Clusters (TPC) 42 72
Streaming Multiprocessors (SM) 84 144
CUDA kernels 10,752 18,432
L2 cache 6 months 96 months
TFLOPS of power 37.6 90?
memory type GDDR6X GDDR6X
memory bus 384 bit 384 bit
memory capacity 24 Go (3090) 24GO (4090 or 5090?)
TGP 350W 600W
Release date September 20 2H 2022?

based on rumors

the GPUs ada lovelace AD102 would consist of 144 SMs, a 71% increase over the current GA102 GPU! At the heart of its mechanics would be in action no less than 18,432 CUDA cores and even a 384-bit bus, enough to ensure a bandwidth of 1004 Gb/s.

GeForce RTX 5090 or RTX 4090, a monster equipped with AD102 GPU and 24 GB of memory

Radeon RX RDNA 3 (Radeon Rx 7950 XT?)

On the AMD side murmur that the upcoming Navi 31, the flagship GPU of the RDNA 3 family (Radeon RX 7950 XT?) will be equipped with 7 chipsets. Not everything is very clear on this topic. Typically, AMD is supposed to use a multi-chip module MCM contraction approach with 2 GCDs (5nm) and 4 MCDs (6nm), all accompanied by an IOD.

This hypothetical Radeon RX 7950 XT Navi 31 GPU should offer 15,360 stream processors powered at 2.5 GHz. We do not know if this frequency represents the maximum possible value or the base. Meanwhile, this is an 11% improvement over the Radeon RX 6900 XT, enough to ensure, with other changes, a theoretical power output of 38.4 TFLOPs at FP32, a nice +67% over the Radeon RX6900XT. The card should be accompanied by an imposing TDP of 500 Watts, again a 67% increase compared to its predecessor.

On the equipment side, it is not impossible that the RAM offers 32 GB, which is double that of the current AMD flagship. The operation of this GDDR6 at 21 Gbps would be ensured by a 256-bit bus. Similarly, the Infinity Cache should also be doubled (512 MB).

Here is a first signed report based on rumors VideoCardz.

GeForce RTX 4090 Radeon RX7950XT
engraving fineness

TSMC-N5

TSMC N5/N6
Architecture ada lovelace RDNA3
Conception Monolithic Multi-chip module (MCM)
Arrays one 2 GCD + 4 MCD + 1 IOD
MEGA clusters 12 GPC (Graphics Processing Group) 2×3 shader engines
SUPER clusters 72 TPC (Texture Processing Groups) 2 × 30 WGP (RDNA)
GPU clusters 144 SMS
FP32 cores 18,432 CUDA cores 15,360 stream processors
GPU frequency ~2.2GHz ~2.5GHz
living memory 24 Go GDDR6X ? Go GDDR6
Memory frequency and memory bus 21Gbit/s – 384bit ? Gbit/s – 256 bits
Cache 96 MB of L2 cache 256 or 512 MB of Infinity Cache
Consumption 600W (TDP) 500W (TBP)
Release date Q3/Q4 2022 Q3/Q4 2022
performances 2GA102

2.5 x NAVI 21

Leave a Comment