AMD's Vega 10 Based Dual GPU Graphics Card Inbound - Expected Single Precision Compute Performance To Be Greater Than 18 TFLOPs

AMD'due south Vega lineup will feature immensely powerful cards with HBM2 memory in 2022. Original Image by FallenZeraphine

Information technology looks like the hardware leak season is in total swing because a report by Fudzilla reveals that AMD is working on a dual flake Vega 10 graphics menu for the professional marketplace. This comes a few weeks after Videocardz confirmed the existence of the same – and so y'all know have 2 independent confirmations for dual GPU Vega 10. This is something that isn't actually all that surprising and probably fairly expected in terms of lineup progression simply nevertheless goes on to testify that AMD volition exist leveraging its arsenal of GPUs to their total potential to maximize its return on investment.

Vega x Dual GPU expected to state in Q2 2022 - targeted at the professional person market

Non much details are known about the Vega 10 based dual GPU at this point but since AMD has always included full dies in any dual GPU configuration we tin simply multiply the known configuration of Vega10 by 2. It's really every bit simple as that.
Since Vega 10 has 64 compute units, the dual GPU will have 128 CUs for a full of 8192 stream processors (assuming the SP to CU ratio remains the aforementioned in Vega compages as Polaris). Yous are as well looking at a total of 32 GB of HBM2 on the board with an aggregate bandwidth of one TB/s (512 GB/southward x2). The TBP (Total Board Ability) is expected to be around 300W so nosotros can expect it to exist clocked lower than its single GPU siblings.

The Vega x GPU has roughly 24 TeraFLOPs of 16-bit compute. xvi bit compute is, of grade, half-precision work and since Vega has native 16-fleck compute support, we tin detect out the single precision performance by simply cut the number in half: which gives the states exactly12 TeraFLOPs of single precision compute. The required clock rates for this amount of performance tin can also be easily derived using the function [Stream Processors * Clock Rate * 2 Instructions per Clock] which are somewhere in the vicinity of 1465 Mhz.

Naturally, a dual flake solution is never clocked every bit high every bit the single chip solution so assuming clock rates of 1100 Mhz to 1200 Mhz y'all are looking at en constructive computational power of 18 TeraFLOPs to 19.6 TeraFLOPs hands – and that's single precision figures. If Nvidia does not respond with a dual GPU solution besides (and dual GPUs aren't usually their style in the professional person market place) then this graphics card volition hands be able to give contest to Nvidia'due south P100 and GP102 based offerings.

Both sources have indicated a time span of Q2 2022 for the arrival of this monster but information technology is very much possible that we are going to be seeing a demo of the same by the finish of the year; around the same time the unmarried scrap based version would first shipping. We take as well previously posted virtually all of AMD'south upcoming offerings which include the 7nm based Vega 20 GPU besides as the Navi series. The process node for this particular graphics card volition be the 14nm FinFET process from GlobalFoundries/Samsung.

A Vega 10 dual GPU, at both clock rates (1200 Mhz or 1100 Mhz) hands destroys the P100 on paper – which has a single precision performance of nine.iii TeraFLOPs for PCI-e based cards. It goes without saying that since we are comparison across 2 completely different architectures hither, what is on paper tin be different in real life. This besides means that even if Nvidia responds with a dual GPU solution of its ain, the Vega10 based offer would remain very competitive. The ball is now undoubtedly in Nvidia's courtroom and we shall see what its high-end looks like in the coming months.

AMD Adjacent Generation Vega 10, eleven, xx and Dual GPU Graphics Card Rumored Lineup:

WCCFTech	Polaris 10	Vega 11	Vega 10	Vega ten Dual GPU	Vega 20
Year	2016	2017	2017	2017	2018
Process	14nm FinFET	14nm FinFET	14nm FinFET	14nm FinFET	7nm FinFET
Transistors In Billions	5.7	TBA	TBA	TBA	TBA
Stream Processors	2304	2304+ (est.)	4096	8192	4096
Clock Speed	1266 Mhz	TBA	1526 Mhz	1100 Mhz+ (est.)	1800 Mhz+ (est.)
Operation	5.8 TFLOPS	TBA	12.5 TFLOPS	19 TFLOPS - 24 TFLOPs (est.)	fifteen TFLOPS+
TDP	150W	TBA	225W	300W	150-300W
Memory	8GB GDDR5	TBA	8GB/16GB HBM2	16-32GB HBM2	16-32GB HBM2
Memory Bus	256bit	TBA	2048-chip (2 Stacks)	4096-bit (2048-bit x2)	4096-chip (iv Stacks)
PCI Express	3.0	TBA	3.0	3.0	four.0
Bandwidth	256 GB/south	TBA	512 GB/s	1 TB/s	1 TB/s

Source: https://wccftech.com/vega-10-dual-gpu-graphics-card/

Posted by: craneacursent.blogspot.com

Share this post