Details regarding the NVIDIA Ada Lovelace Gaming GPU which will power the GeForce RTX 40 series graphics cards have been revealed. The new information comes from Kopte7kimi & talks about the block diagram of the next-gen architecture.
NVIDIA GeForce Ada Lovelace GPU SM Block Diagram Detailed: Bigger & Better Than Ever For Gamers!
The NVIDIA Ada Lovelace GPU architecture is no mystery anymore. We have learned the specific configurations that will power the next Gen AD10* series SKUs for GeForce RTX 40 series graphics cards and we have also seen leaked specifications of the lineup. Now, it's time to talk purely about the next-generation graphics chip itself.
NVIDIA AD102 'Ada Lovelace' Gaming GPU 'SM' Block Diagram (Image Credits: Kopite7kimi):
NVIDIA GA102 'Ampere' Gaming GPU 'SM' Block Diagram:
Starting with the GPU configuration, Kopite7kimi compares the top AD102 GPU to various other GPUs from the green team. These include the gaming-focused Ampere GA102 and Turing TU102 while there's also the HPC-Focused Hopper GH100 and Ampere GA100 added to the list. I'll only compare the AD102 to its gaming predecessors since the HPC-focused designs are vastly different than consumer-centric offerings.
The NVIDIA Ada Lovelace AD102 GPU will feature up to 12 GPC (Graphics Processing Clusters). This is an increase of 70% versus GA102 which features only 7 GPCs. Each GPU will consist of 6 TPCs and 2 SMs which is the same configuration as the existing chip. Each SM (Streaming Multiprocessor) will house four sub-cores which is also the same as the GA102 GPU. What's changed is the FP32 & the INT32 core configuration. Each sub-core will include 128 FP32 units but combined FP32+INT32 units will go up to 192. This is because the FP32 units don't share the same sub-core as the IN32 units. The 128 FP32 cores are separate from the 64 INT32 cores.
So in total, each sub-core will consist of 128 FP32 plus 64 INT32 units for a total of 192 units. Each SM will have a total of 512 FP32 units plus 256 INT32 units for a total of 768 units. And since there are a total of 24 SM units (2 per GPC), we are looking at 12,288 FP32 Units and 6,144 INT32 units for a total of 18,432 cores. Each SM will also include two Wrap Schedules (32 thread/CLK) for 64 wraps per SM. This is a 50% increase on the cores (FP32+INT32) and a 33% increase in Wraps/Threads vs the GA102 GPU.
NVIDIA Ada Lovelace GPU Specs 'Preliminary':
| GPU Name | AD102 | GA102 | TU102 | GA100 | GH100 |
|---|---|---|---|---|---|
| GPC | 12 (Per GPU) | 1.7x | 2x | 1.5x | 1.5x |
| TPC | 6 (Per GPC) | Same | Same | 0.75x | 0.67x |
| SM | 2 (Per TPC) | Same | Same | Same | Same |
| Sub-Core | 4 (Per SM) | Same | Same | Same | Same |
| FP32 | 128 (Per SM) | Same | 2x | 2x | Same |
| FP32+INT32 | 192 (Per SM) | 1.5x | 1.5x | 1.5x | Same |
| Warps | 64 (Per SM) | 1.33x | 2x | Same | Same |
| Threads | 2048 (Per SM) | 1.33x | 2x | Same | Same |
| L1 Cache | 192 KB (Per SM) | 1.5x | 2x | Same | 0.75x |
| L2 Cache | 96 MB (Per GPU) | 16x | 16x | 2.4x | 1.6x |
| ROPs | 32 (Per GPC) | 2x | 2x | 2x | 2x |
Moving over to the cache, this is another segment where NVIDIA has given a big boost over the existing Ampere GPUs. The Ada Lovelace GPUs will pack 192 KB of L1 cache per SM, an increase of 50% over Ampere. That's a total of 4.5 MB of L1 cache on the top AD102 GPU. The L2 cache will be increased to 96 MB as mentioned in the leaks. This is a 16x increase over the Ampere GPU that hosts just 6 MB of L2 cache. The cache will be shared across the GPU.
Finally, we have the ROPs which are also increased to 32 per GPC, an increase of 2x over Ampere. You are looking at up to 384 ROPs on the next-gen flagship versus just 112 on the fastest Ampere GPU, the RTX 3090 Ti. There are also going to be the latest 4th Generation Tensor and 3rd Generation RT (Raytracing) cores infused on the Ada Lovelace GPUs which will help boost DLSS & Raytracing performance to the next level. Overall, the Ada Lovelace AD102 GPU will offer:
- 2x GPCs (Versus Ampere)
- 50% More Cores (Versus Ampere)
- 50% More L1 Cache (Versus Ampere)
- 16x More L2 Cache (Versus Ampere)
- Double The ROPs (Versus Ampere)
- 4th Gen Tensor & 3rd Gen RT Cores
Do note that clock speeds, which are said to be between the 2-3 GHz range, aren't taken into the equation so they will also play a major role in improving the per-core performance versus Ampere. The NVIDIA GeForce RTX 40 series graphics cards featuring the next-gen Ada Lovelace gaming GPUs are expected to launch in the second half of 2022 & are said to utilize the same TSMC 4N process node as the Hopper H100 GPU.
NVIDIA CUDA GPU (RUMORED) Preliminary:
| GPU | TU102 | GA102 | AD102 |
|---|---|---|---|
| Flagship SKU | RTX 2080 Ti | RTX 3090 Ti | RTX 4090? |
| Architecture | Turing | Ampere | Ada Lovelace |
| Process | TSMC 12nm NFF | Samsung 8nm | TSMC 4N? |
| Die Size | 754mm2 | 628mm2 | ~600mm2 |
| Graphics Processing Clusters (GPC) | 6 | 7 | 12 |
| Texture Processing Clusters (TPC) | 36 | 42 | 72 |
| Streaming Multiprocessors (SM) | 72 | 84 | 144 |
| CUDA Cores | 4608 | 10752 | 18432 |
| L2 Cache | 6 MB | 6 MB | 96 MB |
| Theoretical TFLOPs | 16 TFLOPs | 40 TFLOPs | ~90 TFLOPs? |
| Memory Type | GDDR6 | GDDR6X | GDDR6X |
| Memory Capacity | 11 GB (2080 Ti) | 24 GB (3090 Ti) | 24 GB (4090?) |
| Memory Speed | 14 Gbps | 21 Gbps | 24 Gbps? |
| Memory Bandwidth | 616 GB/s | 1.008 GB/s | 1152 GB/s? |
| Memory Bus | 384-bit | 384-bit | 384-bit |
| PCIe Interface | PCIe Gen 3.0 | PCIe Gen 4.0 | PCIe Gen 4.0 |
| TGP | 250W | 350W | 600W? |
| Release | Sep. 2018 | Sept. 20 | 2H 2022 (TBC) |
The post NVIDIA Ada Lovelace ‘GeForce RTX 40’ Gaming GPU Detailed: Double The ROPs, Huge L2 Cache & 50% More FP32 Units Than Ampere, 4th Gen Tensor & 3rd Gen RT Cores by Hassan Mujtaba appeared first on Wccftech.