The AMD Radeon VII Review: An Unexpected Shot At The High-End
by Nate Oh on February 7, 2019 9:00 AM ESTRadeon VII & Radeon RX Vega 64 Clock-for-Clock Performance
With the variety of changes from the Vega 10 powered RX Vega 64 to the new Radeon VII and its Vega 20 GPU, we wanted to take a look at performance and compute while controlling for clockspeeds. In this way, we can peek at any substantial improvements or differences in pseudo-IPC. There's a couple caveats here; obviously, because the RX Vega 64 has 64 CUs while the Radeon VII has only 60 CUs, the comparison is already not exact. The other thing is that "IPC" is not the exact metric measured here, but more so how much graphics/compute work is done per clock cycle and how that might translate to performance. Isoclock GPU comparisons tend to be less useful when comparing across generations and architectures, as like in Vega designers often design to add pipeline stages to enable higher clockspeeds, but at the cost of reducing work done per cycle and usually also increasing latency.
For our purposes, the incremental nature of 2nd generation Vega allays some of those concerns, though unfortunately, Wattman was unable to downclock memory at this time, so we couldn't get a set of datapoints for when both cards are configured for comparable memory bandwidth. While the Vega GPU boost mechanics means there's not a static pinned clockspeed, both cards were set to 1500MHz, and both fluctuated from 1490 to 1500MHZ depending on workload. All combined, this means that these results should be taken as approximations and lacking granularity, but are useful in spotting significant increases or decreases. This also means that interpreting the results is trickier, but at a high level, if the Radeon VII outperforms the RX Vega 64 at a given non-memory bound workload, then we can assume meaningful 'work per cycle' enhancements relatively decoupled from CU count.
As mentioned above, we were not able to control for the doubled memory bandwidth. But in terms of gaming, the only unexpected result is with GTA V. As an outlier, it's less likely to be an indication of increased gaming 'work per cycle,' and more likely to be related to driver optimization and memory bandwidth increases. GTA V has historically been a title where AMD hardware don't reach the expected level of performance, so regardless there's been room for driver improvement.
SPECviewperf is a slightly different story, though.
289 Comments
View All Comments
Manch - Monday, February 11, 2019 - link
My bad on Wolf. I thought it was. It's on XB1 which is DX12 and DX12 supt was confirmed by a few places so I didn't check further.As for Vulkan Games, off the top of my head(whats in my library), TWS:ToB, TW: Warhammer II (should have been in my table..oops), Warhamer 40K DoW III, Serious SAM VR games, x-plane. I'm sure there are others. Easy to look up.
IMO FPS should not be the definitive test for all API's. Variety is always nice.
Cherry pick my mistakes but my point stands. I get the test bed needs to be locked down so consistent results can be achieved. Anandtech needs to be able to give specific measurable and repeatable results and they do that. I'm just merely expressed my desire to see a more balanced test suite in regards to APIs & games that are design for NVidia or AMD GPU's.
eddman - Tuesday, February 12, 2019 - link
Are you basing that on personal experience or simply getting the info from vulkan's wikipedia page, without checking the platform column?TWS:ToB, TW: Warhammer II and Warhamer 40K DoW III use vulkan only on linux.
Despite the vulkan addition, Serious sam games are old, non-demanding and not suitable for benchmarking.
X-plane does not support vulkan yet; it's a work-in-progress. Still, even if it does add it eventually, it too is not suitable for benchmarking.
Ryan Smith - Thursday, February 7, 2019 - link
The benchmark suite gets updated on a roughly yearly basis. It was last updated for the Turing launch, so we're only about 5 months into it. As part of ensuring we cover a reasonable selection of genres, these were the best games available in the fall of 2018.The next time we update it will presumably be for AMD's Navi launch, assuming that still happens in 2019. Though it's never too early to suggest what games you'd like to see.
eva02langley - Thursday, February 7, 2019 - link
Devil May Cry, Resident Evil, Anthem, metro Exodus, The Division 2, Rage 2, Mortal Kombat 11krazyfrog - Sunday, February 10, 2019 - link
Half-Life 3SeannyB - Thursday, February 7, 2019 - link
I would like to see a title from each of the general purpose engines, namely UE4 and Unity.Korguz - Thursday, February 7, 2019 - link
maybe i am the only one here.. but the games AT tests... i dont play ANY of them :-)Ryan Smith - Thursday, February 7, 2019 - link
Out of curiosity, what do you play?Korguz - Friday, February 8, 2019 - link
WOW, Starcraft 2 Diablo 3 and some older games... games that dont really " need " a card like this.. my current asus strix 1060, plays these just fine at almost max eye candy... the only game i can think of that i have, and play that might need this card.. is Supreme commander, but im not sure if that game needs a strong cpu, or gpu, maybe a bit of both...Holliday75 - Friday, February 8, 2019 - link
Love me some Supreme Commander. Solid followup to Total Annihilation. As to its performance I think its more CPU based and quite frankly the engine is not optimized for modern hardware.