AMD Radeon RX 7900 XT Flagship RDNA 3 Graphics Card Could Hit Nearly 100 TFLOPs, Navi 31 GPU Frequency Hits Over 3 GHz
AMD Radeon RX 7900 XT turns out to be a completely different ‘Navi 31’ RDNA 3 GPU-powered beast in latest rumors Greymon55. The leaker reported that based on new information, AMD has created a much more powerful design than expected.
AMD Radeon RX 7900 XT Flagship RDNA 3 ‘Navi 31’ GPU-Powered Graphics Card Could Offer Clock Speeds Above 3 GHz and Nearly 100 TFLOPs of Performance
Over the past couple of days, there have been several sightings of AMD RDNA 3 (GFX11) GPUs starting to appear in the LLVM project in the Linux operating system drivers and developer tools. It looks like AMD is currently working on at least four new GPUs which have been confirmed by their leaked ID which is as follows (Credits: @Kepler_L2):
- GFX (1100) – Navi 31 GPU possible
- GFX (1101) – Navi 32 GPU possible
- GFX (1102) – Navi 33 GPU possible
- (GFX 1103) – Possible Phoenix APU
Now, while these IDs don’t reveal much information such as specs or configurations, they tell us that preliminary support is being added and that soon these GPUs will start testing on the respective platforms for delivery. optimized performance. But this topic is not about leaked IDs, this is the latest rumor about the flagship RDNA 3 GPU, the Navi 31, which will power the Radeon RX 7900 series cards.
According to the latest rumor from Greymon55, it looks like clock speeds have seen a major increase from 2.5 GHz to now hovering around and even above the 3 GHz frequency mark. This means that the flagship chip with 15360 stream processors will be able to deliver nearly 100.0 TFLOPs of FP32 compute performance (92 TFLOPs to be precise). This is without adding the OC limits which will certainly help him to approach the barrier already close to 100 TFLOPs.
— Greymon55 (@greymon55) April 30, 2022
— Greymon55 (@greymon55) April 30, 2022
But it seems that recent reports that NVIDIA is making a major change to its Ada Lovelace GPU architecture for the GeForce RTX 40 series have also been taken into consideration and leaks expect that Ada Lovelace flagships will not be not just the first to violate the 100 TFLOP GPU. Compute barrier, but also offers clock frequencies similar to AMD’s RDNA 3, which could be due to the move from Samsung’s 8nm node to TSMC’s cutting-edge 4N node, which is an optimized variant based on the TSMC N5 process (5 nm).
AMD RDNA 3 Navi 31 GPUs for Radeon RX 7900 Series
The AMD Navi 31 GPU, the flagship RDNA 3 chip, is said to power next-gen enthusiast cards such as the Radeon RX 7900 XT graphics card. We’ve heard that AMD will be ditching CUs (computing units) in favor of WGPs (workgroup processors) on its next-gen RDNA 3 GPUs. Each WGP will host two CUs (computing units) but with twice the SIMD32 clusters instead of only 2 on each CU in RDNA 2.
According to the Linux drivers, it seems that the compute units still exist and the WGPs are still DualCUs. The big change is back to 4x SIMD per CU like GCN, but now with RDNA’s 32 ALU design. pic.twitter.com/hAzGs9M9E6
— Kepler (@Kepler_L2) April 30, 2022
The Navi 31 GPU configuration shown here includes two GCDs (Graphics Core Die) and a single MCD (Multi-Cache Die). Each GCD has 3 Shader Engines (6 total) and each Shader Engine has 2 Shader Arrays (2 per SE / 6 per GCD / 12 total). Each Shader Array consists of 5 WGPs (10 per SE / 30 per GCD / 60 total) and each WGP has 8 SIMD32 units with 32 ALUs (40 SIMD32 per SA / 80 per SE / 240 per GCD / 480 total). These SIMD32 units combine to make up 7,680 cores per GCD and 15,360 cores in total.
The MCD Navi 31 (RDNA 3) will be linked to the dual GCDs via a next-generation Infinity Fabric interconnect and will have 256-512 MB of Infinity cache. Each GPU must also have 4 memory connection links (32-bit). That’s a total of 8 32-bit memory controllers for a 256-bit bus interface.
There are also rumors of 3D stacking on the flagship Navi 31 GPU and the launch is currently slated for late Q4 2022, although it’s likely we could see a paper launch with actual availability from of the first quarter of 2023.
Upcoming AMD, Intel, and NVIDIA GPU Specs (Preliminary)
|GPU name||AD102||Navi 31||Xe2-HPG|
|Code name||Ada Lovelace||RDNA3||battle mage|
|Flagship SKU||GeForce RTX 4090 Series||Radeon RX 7900 series||Arc B900 Series|
|GPU process||TSMC4N||TSMC 5nm+ TSMC 6nm||5nm TSCM?|
|GPU package||Monolithic||MCD (Multi-Chiplet Die)||MCM (multi-chip module)|
|GPU dies||Mono x 1||2x GCD + 4x MCD + 1x IOD||Quad-Tile (tGPU)|
|Mega GPU clusters||12 GPCs (Graphics Processing Clusters)||6 shader engines||10 render slices|
|Super GPU clusters||72 TPC (Texture Processing Poles)||30 WGP (by MCD)
60 WGP (total)
|40 Xe Cores (per tile)
160 Xe cores (total)
|GPU clusters||144 stream multiprocessors (SM)||120 compute units (CU)
240 compute units (total)
|1280 Xe VE (per tile)
5120 Xe VE (total)
|Cores (per matrix)||18432 CUDA cores||7680 SP (by GCD)
15360 SP (Total)
|20480 ALUs (total)|
|Rush Clock||~2.85GHz||~3.0GHz||To be determined|
|Calculation FP32||~105 TFLOPs||~92 TFLOPs||To be determined|
|Memory capacity||24 GB||32 GB||To be determined|
|Memory bus||384 bit||256 bit||To be determined|
|Memory speeds||~21 Gbps||~18 Gbps||To be determined|
|Cache subsystems||96 MB L2 cache||512 MB (Infinity Cache)||To be determined|
|PAD||~600W||~500W||To be determined|
|To throw||Q4 2022||Q4 2022||2023|
News Source: Videocardz
#AMD #Radeon #Flagship #RDNA #Graphics #Card #Hit #TFLOPs #Navi #GPU #Frequency #Hits #GHz