The NVIDIA Ampere family of graphics cards ushered in a huge expansion, with up to six new models in a single breath, but not gaming cards, but professional graphics workstation cards, that is, the previous Quadro series (this brand has been abandoned), for artists, designers, engineers and other crowds.
Visit the purchase page:
For desktop computers are RTX A5000, RTX A4000And the former main partner of RTX A6000 constitutes a full line of mid-to-high-end products.
RTX A5000 originally assumed it would be downgraded to use GA104 core, but did not expect it to be the same as RTX A6000Large GA102 core, CUDA core reduced to 8192(There is no corresponding game card), there are also 256 tensioner cores and 64 light chasing cores, which are roughly equal to three-quarters of RTX A6000 and between RTX 3080 and RTX 3090.
Performance is almost 30% lower, FP32 floating point performance is 27.8TFlops (2.78 trillion calculations per second), optical tracking performance is 54.2TFlops, tensioner performance is 222.2TFlops, and the power consumption of the whole card is from 300W to 230W
Video memory is full384-bit GDDR6 ECC, The bandwidth is also 768GB / s, that is, the capacity is halved to 24GB, and dual cards can be connected in parallel to get 48GB.
It is also compatibleNVIDIA RTX vWS SoftwareSupports multiple instances of high-performance virtual workstations, remote users can share one card.
RTX A4000 is GA104 core, CUDA 6144 core, Slightly higher than RTX 3070,Yield is equivalent to approximately 70% of RTX A5000The knife is accurate and the power consumption is only 140W.
Video memory is 256-bit GDDR6 ECC 16GB and bandwidth is 448GB / s.
For notebooks they are RTX A5000, RTX A4000, RTX A3000, RTX A2000, Both supportThe third generation of Max-Q technology, Includes dynamic acceleration 2.0, silent mode 2.0, adjustable BAR acceleration, DLSS smoothing.
RTX A5000 G102 core, 6144 CUDA cores, Same as the mobile version of RTX 3080, but with higher video memory with 256-bit 16GB GDDR6, 80-165W power consumption, dual-slot size, equipped with four DP 1.4 output interfaces and dual-channel NVLink compatible , vGPU virtualization.
RTX A4000 and RTX A3000 are GA104, 5120 and 4096 CUDA cores respectively, The memory is 8GB 256-bit GDDR6, 6GB 192-bit GDDR6, 80-140W power consumption, 60-130W, single slot size, no longer support vGPU.
RTX A2000 is a small core GA106, 2560 CUDA cores, 4GB 128-bit GDDR6 video memory, 35-95W power consumption.
Oh yeah, NVIDIA also released it at the same timeTuring architecture T1200, T600, Also for workstation notebooks, but no specific specifications are available at the moment.
The RTX A series desktop products will begin shipping later this month, and the portable products will be available in the second quarter.