Monday, October 5th 2020
NVIDIA Unveils RTX A6000 "Ampere" Professional Graphics Card and A40 vGPU
NVIDIA today unveiled its RTX A6000 professional graphics card, the first professional visualization-segment product based on its "Ampere" graphics architecture. With this, the company appears to be deviating from the Quadro brand for the graphics card, while several software-side features retain the brand. The card is based on the same 8 nm "GA102" silicon as the GeForce RTX 3080, but configured differently. For starters, it gets a mammoth 48 GB of GDDR6 memory across the chip's 384-bit wide memory interface, along with ECC support.
The company did not reveal the GPU's CUDA core count, but mentioned that the card's typical board power is 300 W. The card also gets NVLink support, letting you pair up to two A6000 cards for explicit multi-GPU. It also supports GPU virtualization, including NVIDIA GRID, NVIDIA Quadro Virtual Data Center Workstation, and NVIDIA Virtual Compute Server. The card features a conventional lateral blower-type cooling solution, and its most fascinating aspect is its power input configuration, with just the one 8-pin EPS power input. We will update this story with more information as it trickles out.Update 13:37 UTC: The company also unveiled the A40, a headless professional-visualization graphics card dedicated for virtual-GPU/cloud-GPU applications (deployments at scale in data-centers). The card has similar specs to the RTX A6000.
Update 13:42 UTC: NVIDIA website says that both the A40 and RTX A6000 a 4+4 pin EPS connector (and not 8-pin PCIe) for power input. An 8-pin EPS connector is capable of delivering up to 336 W (4x 7 A @ 12 V).
The company did not reveal the GPU's CUDA core count, but mentioned that the card's typical board power is 300 W. The card also gets NVLink support, letting you pair up to two A6000 cards for explicit multi-GPU. It also supports GPU virtualization, including NVIDIA GRID, NVIDIA Quadro Virtual Data Center Workstation, and NVIDIA Virtual Compute Server. The card features a conventional lateral blower-type cooling solution, and its most fascinating aspect is its power input configuration, with just the one 8-pin EPS power input. We will update this story with more information as it trickles out.Update 13:37 UTC: The company also unveiled the A40, a headless professional-visualization graphics card dedicated for virtual-GPU/cloud-GPU applications (deployments at scale in data-centers). The card has similar specs to the RTX A6000.
Update 13:42 UTC: NVIDIA website says that both the A40 and RTX A6000 a 4+4 pin EPS connector (and not 8-pin PCIe) for power input. An 8-pin EPS connector is capable of delivering up to 336 W (4x 7 A @ 12 V).
25 Comments on NVIDIA Unveils RTX A6000 "Ampere" Professional Graphics Card and A40 vGPU
The 3090 is a workstation card that's been marketed as a gaming card. And that's why we rushed out to buy one. NO ONE rushes out to buy Quadro cards because there's far less demand for workstation cards than thereis for gaming.
My question is whether / when a card will get the RTX Titan 3000 badge.
www.nvidia.com/en-us/data-center/a40/
Retails cards have drivers optimized for gaming.
If you buy a Quadro cards for gaming, you'll have bad gaming performance since the drivers will not be designed for gaming. That's why no one rushes out to buy Quadro cards for gaming.
Dam that cooler looks pretty dam solid, Nv Might be getting more and more to move away from this plastic shroud crap.
TIL 8-pin EPS is capable of 336 W (7A * 12V * 4), and that any mainstream desktop motherboard with >1 EPS input is a scam.
Still waiting for 3080 i ordered 20 days ago, nvidia said production in full swing and there would be waves of cards every week but i haven't seen any waves coming in.
They said there was good stock at release BUT won't say how much, what an obvious lie, the stock where non existent.
In Sweden it was calculated among all big retailers that there where about 700-800 cards available at release date in the entire country while demand was at least 20x more.
Edit: @btarunr beat me to it above. Thx for providing the power rating btw, my two seconds of googling didn't provide it, so I gave up :P
Without actual stock numbers, being "sold out" means nothing.
What is a fact is that Nvidia announced there won't be wide availability of their cards until sometime in 2021. Launching a product and then being out of stock for months on end is bad business. People stop caring after a point.
You don't need GPUs to be physically in stock in order to move them , all resellers have implemented preordering/queuing systems so as soon as they actually receive some physical stock it disapears since this stock is being used to fulfill those preorders . In other words GPUs move so fast that retailers don't even have time to add them to their inventory ( commonly called stock ) .... now i don't know how good your sense of business is but thats the definition of 3rd world problems !
If anything else Huang also declared in the same statement that Nvidia is expecting a very big Q4 season which when translated means , business couldn't be any better for them ........
"Yeah their GPUs being sold out everywhere really shows how disastrous their marketing is ............ :kookoo: "
Given that we have zero data on what exactly constitutes "sold out", it could be anywhere from 0 to 999,999.
I'm not saying sales are good or bad, I'm saying that if you make the argument that sold out = good, you should provide data to back it up.
" all resellers have implemented preordering/queuing systems so as soon as they actually receive some physical stock it disapears since this stock is being used to fulfill those preorders"
It should also be mentioned that pre-ordering and queuing are not the same. Many retailers that accept pre-orders do not queue. Of course it makes sense that they do not, a queue is only useful if you expect an extended period of very low stock. I've pre-ordered plenty of products from Newegg and not once have I seen a queue. I also queued for the index near release and that system was fundamentally different. You'd apply to the queue, be given a time estimate of your product would be ready, and then when it's ready pay for it. If you don't pay for it within 48 hours, you loose your spot. That is the exact same format of queue that EVGA is using. A system like that requires a decent amount of devs hours to create and cannot be done overnight. It makes sense for EVGA to implement a system like that given that Nvidia graphics cards are likely 50% of their business. For others though it may not make sense. I wish it were more common as it can help greatly when new PC parts release.