AMD Vega (FE and RX) Benchmarks [Updated Aug 10 - RX Vega 64 Unboxing]

[nathanddrews]

6/29 @ 2:30pm PT / 5:30pm ET
Radeon Vega Frontier Edition Live Benchmarking
https://www.pcper.com/live/

Thanks, I came to this thread to update this very thing! Can't wait to watch the stream! Starts in 5 minutes!

 

[Rifter]

You're the same style of poster who is like "We have no clue how fast the 1080Ti is!!!!" After the Titan card releases.

Are we seriously just allowed to pretend like we don't understand what the underlying silicon is?
Come on man, lets have a serious discussion based on the facts we know now. It's not fun posting when we do this game. Please.

Lets stick to the Facts as we have them now.

This is the pro version, not optimized or intended for gamers, this is a fact.

The RX version will have better drivers and game optimizations, by how much we have no idea, this is a fact.

This card sits just behind the P5000 in professional workload performance at almost half the price, this is a fact.

The conclusions one can draw from the currently available facts are that in its intended field it is beating the nvidia quadro. We also know it is power hungry, and likely facing a clockspeed wall due to the process, but we dont know how much this is going to effect the RX line as we do not know how much performance the RX is going to pickup over the PRO. .

We know the gaming performance will improve, we dont know by how much, it could be 5% it could be 20% it could be 40%, no one knows.

Like i said originally if we want RX performance numbers we are going to need to wait, i know it sucks but we dont have anything concrete to go on at this time.

Its not looking good, but that doesnt mean its looking dismal either. Time will tell.

 

[Bacon1]

You're right, the Titan line has always been a terrible value proposition. But it was always the undisputed best at something when it came out. ALWAYS. What is a Vega FE the best at? Nothing. Nothing at all. On top of that it's throttling with the stock cooler and consuming more (considerably more) power than any other card in production. And AMD wants someone to pay a premium price for something that isn't premium at anything. And you're defending this?!

It blows my mind more to watch AMD fans defend a dumpster fire than it is to see AMD stumble.

Uhhh you do realize that it is better at the Titan Xp for production related tasks right? The same Titan that also throttles due to a blower design? I'd love to see exactly how much wattage it's pulling, can you link me to someone testing it?

 

[nathanddrews]

"Expect no performance delta between Gaming and Pro mode. Shipping driver is current, but is not comparable to RX Vega driver."
PCPer Ryan quoting AMD

 

[Bacon1]

6/29 @ 2:30pm PT / 5:30pm ET
Radeon Vega Frontier Edition Live Benchmarking

https://www.pcper.com/live/

Starting now

5960 @ stock? 16GB of VRAM

Avg clockspeed of 1440 out of the box from his testing so far. Logged with GPU Z

 

[nathanddrews]

Yeah, they are currently running EVERYTHING at stock, out of the box settings.

 

[Krteq]

I find the claim that Vega was running Fury drivers difficult to believe. If Vega is truly a new architecture with a new instruction set as I believe AMD has claimed, then there's no way Fury drivers could function. All the hardware the driver interfaces would be different.

Well, nope. Vega maintains GCN ISA. AMD calls it GCN5 in some materials, but it's still far away previous GCN iterations.

 

[nathanddrews]

Dirt 4
2560x1440
Titan Xp: 150fps
Vega FE: 100fps

 

[Snarf Snarf]

"Expect no performance delta between Gaming and Pro mode. Shipping driver is current, but is not comparable to RX Vega driver."
PCPer Ryan quoting AMD

Annnnd there we have it. It's a current driver but it isn't the real Vega launch driver. Can we stop all this talk about AMD regressing in IPC and being completely incompetent? Like I said earlier in my post, this is obviously just a paper launch early adopter card to satisfy the promise Lisa made to investors about 1H launch.

 

[OatisCampbell]

Dirt 4
2560x1440
Titan Xp: 150fps
Vega FE: 100fps

Link?

If true, the only way that is made up with drivers is if they unlock 1/2 the chip not in use.

 

[Bacon1]

Dirt 4
2560x1440
Titan Xp: 150fps
Vega FE: 100fps

1080 gets ~100 fps as well:

 

[nathanddrews]

Link?

If true, the only way that is made up with drivers is if they unlock 1/2 the chip not in use.

The link is the PCPer stream in the first post.

 

[.vodka]

pcper's live stream. They're live right now. It's linked a few posts above.

I suppose they're gonna publish the results later or upload the stream

GTA V built in benchmark right now, 2560p max settings 4xAA, it's around a 40-70FPS range, running at around 1400MHz. It's capped at 83-84°C with no annoying fan noise, the water cooled version won't have this issue... as usual.. or if they crank up the fan so it runs at 1600MHz all the time.

Witcher 3 next, nothing special

ROTR next

 

[Bacon1]

GTA V built in benchmark right now, 2560p max settings 4xAA, it's around a 40-70FPS range, running at around 1400MHz.

Hmm thats basically the exact same results as the Fury X

https://www.pcper.com/image/view/79311?return=node/67279

 

[Armsdealer]

GTA V built in benchmark right now, 2560p max settings 4xAA,

Pretty sure it was 1440p.... Could be wrong though

The results are below 980ti

 

[.vodka]

1440p, yeah. 2560x1440, brain fart there. Yeah performance seems similar to a Fury X all around. It's VERY weird.

So this is running in Fury X mode as it was all the way back in January, and the final RX Vega driver somehow unlocks all of Vega's architectural improvements and brings performance up at least to what a hypothetical 1400-1600MHz Fury X should do?

Power testing up next with the witcher 3, he's seen some weird power draw curves with no clock speed throttling

but it's around the 280-300w mark for their standard test run

ROTR power next, similar stuff

Metro 4k power next, not exceeding 300w

 

[stahlhart]

Everyone participating in this discussion needs keep their rhetoric under control. Agree, disagree, agree to disagree, but stop deliberately provoking anger with hyperbolic language.
-- stahlhart

 

[Maverick177]

Weird, the VEGA FE as of now is exactly at FuryX at VEGA clock.

 

[Kenmitch]

Very boring to watch. His actions or lack of make me believe he knows something isn't right about the results.

 

[Elfear]

1440p, yeah. 2560x1440, brain fart there. Yeah performance seems similar to a Fury X all around. It's VERY weird.

So this is running in Fury X mode as it was all the way back in January, and the final RX Vega driver somehow unlocks all of Vega's architectural improvements and brings performance up at least to what a hypothetical 1400-1600MHz Fury X should do?

Power testing up next with the witcher 3, he's seen some weird power draw curves with no clock speed throttling

but it's around the 280-300w mark for their standard test run

ROTR power next, similar stuff

Metro 4k power next, not exceeding 300w

Sorry, don't have access to the stream right now but is that system power or strictly GPU?

 

[Malogeek]

Sorry, don't have access to the stream right now but is that system power or strictly GPU?

GPU of course under full load. It's a 300w GPU as per AMD's specs.

 

[.vodka]

Sorry, don't have access to the stream right now but is that system power or strictly GPU?

Strictly GPU. The power meters are hooked up to the GPU. About 20-25w are drawn from the PCIe slot, the rest from the PSU.

 

[USER8000]

Fallout 4 performance.

The major divisions on the chart look to be probably 50FPS differences. So that would place the Vega FE at GTX1080 level performance in that game.

 

[tential]

Pcpers:

game mode driver has optimizations, he expects +8% gain in best scenarios using rx vega drivers in a month

Source:https://www.reddit.com/r/Amd/comments/6katp6/radeon_vega_frontier_edition_live_benchmarking/djkrrz7/

 

[Karnak]

Yeah, gaming optimizations. But we don't know if AMD is still using a Fiji driver like months ago. That would explain why we're seeing only Fury X performance just with higher clocks.