NVIDIA RTX A6000 48GB GPU NVIDIA Drops Quadro Designation

By
Cliff Robinson
-
8
NVIDIA RTX 6000 Cover
NVIDIA RTX 6000 Cover

The NVIDIA RTX A6000 GPU is particularly interesting due to a key omission. This was originally set to be a “Quadro” brand product. In a last-minute change, the “Quadro” designation was dropped, even though it is a card typically found in that segment. We see this as perhaps an update to the NVIDIA Quadro RTX 6000 if not maybe the NVIDIA Quadro RTX 8000 we reviewed (also with two in NVLink.) We confirmed with NVIDIA that the company is phasing out the Quadro branding as they did with the “Tesla” brand in 2019. We were told the “Tesla” brand was phased out to reduce confusion with the car brand, but we are less certain what the Quadro brand would conflict with.

NVIDIA RTX A6000 48GB GPU

With the NVIDIA RTX A6000 we get a 48GB GDDR6 GPU. We typically see GDDR memory in this space instead of HBM2 as this is an actively cooled card. TDP on the card is raising to 300W continuing a trend towards higher-power GPUs.

Architecture Ampere
Foundry Samsung
Process Size 8nm
Transistors 28.3 billion
Die Size 628.4 mm2
CUDA Parallel Processing cores 10,752
NVIDIA Tensor Cores 336
NVIDIA RT Cores 84
GPU Memory 48 GB GDDR6 with ECC
Memory Interface 384-bit
Memory Bandwidth 768 GB/s
Max Power Consumption 300W
Graphics Bus PCI Express 4.0 x16
Display Connectors DP 1.4 (4) *
Form Factor 4.4” H x 10.5” L Dual Slot
Product Weight 1.179 kg
Thermal Solution Active
vGPU Software Support NVIDIA ® GRID®, NVIDIA Quadro® Virtual Data Center Workstation, NVIDIA Virtual Compute
Server*
vGPU Profiles Supported 1 GB, 2 GB, 3 GB, 4 GB, 6 GB, 8 GB, 12 GB, 16
GB, 24 GB, 48 GB
NVIDIA® 3D Vision® and 3D Vision Pro Support via 3 pin mini DIN
Frame lock Compatible (with Quadro Sync II)
NVLink 2-way low profile (2-slot and 3-slot bridges)
Connect 2x RTX A6000
NVLink Interconnect 112.5 GB/s (bidirectional)
Power Connector 1x 8-pin CPU
NVENC | NVDEC 1x | 2x (+AV1 decode)

Something we thought was interesting is that NVIDIA is listing GDDR6 memory, not GDDR6X memory as we saw with the higher-end NVIDIA GeForce RTX 3000 series GPUs. We would have expected to see GDDR6X on the higher-end GPUs.

Final Words

Power and cooling for a 300W GPU, or perhaps, more importantly, multiple 300W GPUs is harder than many of the older generations. Not only are we seeing TDP creep in the data center on CPUs and GPUs, but we are also going to see more in higher-end workstations. Although many of the specs are similar between this and the NVIDIA A40, the memory bandwidth his higher on the RTX A6000 with 768GB/s versus 696GB/s on the passively cooled GPU.

RELATED ARTICLESMORE FROM AUTHOR

8 COMMENTS

  1. Phasing out Tesla reduces confusion with the car brand, maybe phasing out Quadro reduces confusion with the Audi 4WD brand /s

  2. Is the power connector spec correct? It seems odd that the card would require a “1x 8-pin CPU” connector, instead of the standard 8-pin PCIe connector.

  4. Yes Michael, while this is the first time a workstation variant has has used EPS power the server compute cards have used them since the k80. Depending on the wire gauge the EPS connector can support 288-384w, which in addition to slot power is more than enough to hit the standard 300w tdp of server compute cards, this is however a decent increase over the previous workstation cards.

  5. Thanks Patrick and Bryan, this is good to know. I assume the 8-pin CPU power connector will impact whether A6000(s) can be powered in existing workstations & servers.

  8. Audi uses “quattro” (four) moniker as a reference to 4WD, “quadro” is the italian word for “painting” and was probably used to emphasize visual and color quality.

    The reason to “trash” a very highly recognized product name is a mistery to me …

    ps: the K600 quadro graphics card in my pc shares my opinion

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.