NVIDIA RTX A6000 was the only Ampere workstation graphics card you can get, but that is about to change because the company is preparing the RTX A5000 and RTX A4000 as the most affordable professional options for content creators and heavy users.
NVIDIA RTX A5000 and RTX A4000 Workstation Ampere Graphics cards spotted, laptop variants are also coming
The NVIDIA RTX A5000 and RTX A4000 graphics cards are based on the GPU Ampere architecture and are sister to the flagship RTX A6000. Both graphics cards were submitted to Khronos Group’s OpenCL 1.2 validation. Currently, NVIDIA RTX A6000 uses the full GA102 GPU, so it can be assumed that the RTX A5000 and RTX A4000 will also come with the complete configurations of NVIDIA’s Ampere GA104 and GA106 GPUs, respectively.
This would give NVIDIA RTX A5000 a total of 6144 CUDA cores and considering that we will be looking at a 256-bit bus, we can expect 16 or 32 GB of GDDR6 memory. The NVIDIA RTX A4000 would obtain the full GA106 GPU with 3840 CUDA cores and could end up with up to 12 GB of GDDR6 memory running on a 192-bit bus interface.
NVIDIA amp workstation graphics cards
| Video card | NVIDIA RTX A4000 | NVIDIA RTX A5000 | NVIDIA RTX A6000 |
|---|---|---|---|
| GPU | Ampere GA106 GPU? | Ampere GA104 GPU? | Ampere GA102 GPU |
| GPU Process | Samsung 8nm | Samsung 8nm | Samsung 8nm |
| Matrix size | 276 mm² | 392 mm² | 628mm² |
| GPU cores | 3840? | 6144? | 10752 |
| Tensioning cores | 120 | 192 | 336 |
| Boost Clock | TBC | TBC | 1860 MHz |
| Single Accuracy | TBC | TBC | 40 TFLOPs |
| VRAM | 12 GB GDDR6 | 16-32 GB GDDR6 | 48 GB GDDR6 |
| NVLINK VRAM | TBC | TBC | 96 GB GDDR6 |
| Memory Bus | 192 bits | 256 bits | 284 bits |
| Memory bandwidth | TBC | TBC | 768 GB / s |
| TDP | TBC | TBC | 300W |
| Launch price | TBC | TBC | $ 4650 US |
| Release date of | TBC | TBC | December 2020 |
In addition to shipping, the NVIDIA RTX A5000 workstation graphics card was located on a workstation laptop made up of an Intel Core i7-11600H CPU. The laptop was tested in the Ashes of The Singularity benchmark and, although we haven’t received any new details about the GPU itself, the CPU tested here is surprising, as it is the first time since the 8th generation parts that a Core i7 SKU is getting 6 cores and 12 threads. Sure, this is a Tiger Lake-H laptop CPU, but it looks like it would have a base clock of 2.90 GHz and 9 MB of L3 cache.
No other details of the laptop are confirmed and the score should be discarded, as it is slower than the RTX 3060 due to non-optimized drivers. All we can say is that NVIDIA is in fact preparing video cards for more affordable workstations based on its GPU Ampere architecture and we can expect them to be released in the next few months in discrete versions of desktops and laptops.