Oct 06, 2020 12:13:00

NVIDIA announces new professional GPUs 'NVIDIA RTX A6000' and 'NVIDIA A40'

On October 5, 2020, chip maker NVIDIA announced the new professional GPU ' NVIDIA RTX A6000 ' and its data center version ' NVIDIA A40 '. All adopt the

Ampere architecture , and it is said that high-end 3D rendering, graphics, simulation, 4K / 8K video editing, etc. will be significantly faster than the previous generation.

NVIDIA Ampere GPUs for Professional Designers Bring Unparalleled Performance to Desktop, Data Center and Cloud | NVIDIA Blog
https://blogs.nvidia.com/blog/2020/10/05/nvidia-ampere-pro-graphics/

Quadro No More? NVIDIA Announces Ampere-based RTX A6000 & A40 Video Cards For Pro Visualization
https://www.anandtech.com/show/16137/nvidia-announces-ampere-rtx-a6000-a40-cards-for-pro-viz

'The new coronavirus epidemic has changed the way content is created and consumed,' said NVIDIA, which requires large amounts of computing resources while distributed teams work remotely with high-definition content. Pointed out that it is being pressed. So NVIDIA says it can help you work from desktop to data center with the NVIDIA RTX A6000 and NVIDIA A40 built on the Ampere architecture.

The two GPUs are essentially successors to the 'Quadro RTX 6000' and 'Quadro RTX 8000' that have adopted the Turing architecture, but the brand name 'Quadro' has been removed. The table below summarizes the specifications of the newly announced 'RTX A6000' and 'A40' and the previous generation 'RTX 8000' and 'GV100'.

	RTX A6000	A40	RTX 8000	GV100
architecture	Ampere	Ampere	Turing	Volta
GPU	GA102	GA102	TU102	GV100
Number of CUDA cores	10752	10752	4608	5120
Tensor core count	336	336	576	640
Memory clock	16Gbps / GDDR6	14.5Gbps / GDDR6	14Gbps / GDDR6	1.7Gbps / HBM2
Boost clock	unknown	unknown	1770MHz	1450MHz
Memory bus width	384 bit	384 bit	384 bit	4096 bit
VRAM	48GB	48GB	48GB	32GB
Thermal design power (TDP)	300W	300W	295W	250W
Cooling method	Active	passive	Active	Active
Manufacturing process	Samsung 8nm	Samsung 8nm	TSMC 12nm FFN	TSMC 12nm FFN
Release price	unknown	unknown	10,000 dollars (about 1.06 million yen)	9000 dollars (about 950,000 yen)
Release date	December 2020	First quarter of 2021	Q4 2018	March 2018

The 'RTX A6000' and 'A40' use the same GA102 chip used in the 'GeForce RTX 3080/3090', with a memory of 48GB and a manufacturing process of Samsung's 8nm. Early adopters of construction, manufacturing, media, entertainment companies, etc. are already using the 'RTX A6000'.

The main difference between the 'RTX A6000' and the 'A40' is the cooling method, and the 'A40' designed for passive cooling is intended to be incorporated into large-scale servers installed in data centers and the like.

The 'RTX A6000' will be available from mid-December 2020 through channels such as PNY , Leadtek , Ingram Micro, and Ryoyo Electro , and the official NVIDIA website, and the 'A40' is expected to be available from the beginning of 2021. I am.