AMD Radeon RX Vega 64 and 56 Reviews [*UPDATED* Aug 28]

[Paul98]

Did any of the reviews test the Vega 56 and 64 clocked the same?

 

[swilli89]

8% faster using 20-25% more watts?

Certainly not a clear win. I honestly can't speak for anyone but myself, but given my cheap electricity and my power plan (free energy from 9pm to 5am.. peak gaming hours) I will always be a pure perf/$ shopper. Also while not at all scientific.. I think we can all honestly admit that the Vega driver will improve over the next while whereas Pascal is pretty much completely mature at this point. So this performance delta may widen.

 

[linkgoron]

Probably because it's hard to say otherwise with reviews and hard data now out.

It is. I just don't get what "outcry" is Tential expecting? We've had the FE for a few weeks now. This is essentially as expected.

Up until Vega FE and the deck from SIGGRAPH, I think no one was actually expecting it to be this bad.

A 486mm^2 GPU not beating a 314mm^2 GPU more than a year later? Of course not.

 

[swilli89]

Based on the articles I have read it just seems like RTG spent a whole bunch of transistors to increase clockspeed at the heavy expense of IPC (in directX11/12 and Vulkan). Identical hardware setup to Fiji yet while 60% higher frequency they only average ~25% faster?

Sorry but if I was RTG and given the fact that they have barely any ecosystem for GPGPU compute.. I would have worked to strip every ounce of compute out of this thing and brought to market a lean chip that could actually compete in the present, multi-billion dollar gaming market.

 

[Arachnotronic]

It is. I just don't get what "outcry" is Tential expecting? We've had the FE for a few weeks now. This is essentially as expected.

There were some who continued to tell us that we should "wait for RX Vega" because Vega FE wasn't the real deal

A 486mm^2 GPU not beating a 314mm^2 GPU more than a year later? Of course not.

There is more to chip architecture -- even GPU architecture -- than just laying down more cores in a given area.

 

[Elixer]

According to an AMD employee, consumer Vega has no SR-IOV. Bummer.

Just 2-4 VFs to help the gaming+virtualization crowd would have been AMAZING, and have a very niche but major and distintictive feature against the GeForces. Now it just simply appear as a competitively priced power hog...

Well, there goes the last reason I would have actually bought a Vega.

Very disappointing to say the least.

 

[moonbogg]

So, bottom line is that performance itself is fine, right?
What is not fine:

Came a year too late
Power hog
Priced too high

What am I missing?

 

[IntelUser2000]

Identical hardware setup to Fiji yet while 60% higher frequency they only average ~25% faster?

Come on.

http://www.gamersnexus.net/guides/2977-vega-fe-vs-fury-x-at-same-clocks-ipc

It's 5-10% slower at the same clock, not 25%+ as you are suggesting. And that 5% difference could be made up with RX Vega.

Vega 56 tests on GN also shows overclocking HBM2 by 20% improves performance only by 3.6%. So its not bandwidth either.

 

[Elixer]

It's not just AMD's dishonesty. It's long term forum members here who perpetuated that dishonesty even though they knew it was just marketing with no facts.
...
This release was EXTREMELY dishonest. It's why I said I'm tired of the AntiNvidia rhetoric as it seems Nvidia gets a lot of hate on here despite giving GREAT performance. Nvidia hasn't been price gouging. They've been offering stellar unmatched performance.

When AMD's marketing arm puts out:

Yet, all the "review" sites REFUSE to call them out on their shady tactics, one wonders, was it just because they were all bribed off with "free" hardware (polaris / vega / ryzen / threadripper), and those fancy boxes & trinkets & dead CPUs & GPUs?

They just brushed it off, without mentioning it.
Then again, they hardly said one word when Nvidia did the whole 970 3.5GB + .5GB fiasco either.

@Ryan Smith review says

..."but the truth of the matter is that while AMD PR puts on their best face, there are signs that behind the scenes things are more chaotic than anyone would care for.

Best face = being dishonest? I can't find a lexicon that agrees with that.

 

[Carfax83]

AMD is stuck with these cards for years now. This looks really bad for everyone. Nvidia is getting used to having no competition at all now. Guess that next Ti is going to be at least $50-$100 more next time. Probably just start charging $1,000 for the Ti models now actually. Screw it, might as well. What can we do? Complain and buy it anyway? Yep.

Doubt that's going to happen. NVidia needs to sell cards to make money, and if the price is too high for most gamers and enthusiasts, they won't buy them. I think their current pricing strategy works really well for generating profits, so if it ain't broke.....

 

[formulav8]

Well at least their CPU division hit a home run for once. It's been their GPU division bringing in the dough for awhile now. Now the CPU division needs to help the GPU. Looks like AMD cared way to much about trying to have a dual purpose card. But given their budget, they may not have had a choice. But I don't know for sure, or really care I guess.

 

[Stuka87]

When AMD's marketing arm puts out:

Yet, all the "review" sites REFUSE to call them out on their shady tactics, one wonders, was it just because they were all bribed off with "free" hardware (polaris / vega / ryzen / threadripper), and those fancy boxes & trinkets & dead CPUs & GPUs?

They just brushed it off, without mentioning it.
Then again, they hardly said one word when Nvidia did the whole 970 3.5GB + .5GB fiasco either.

@Ryan Smith review says
Best face = being dishonest? I can't find a lexicon that agrees with that.

AMD isn't being dishonest in what you posted above. Vega 64 is the fastest GPU in pure compute under $500. And it will get over 60FPS in pretty much every game out. Now saying they are fastest based on compute alone is kind of cherry picking, but getting over 60fps in those games is not dishonest. Sure it may only get 90fps to a 1080Ti's 130fps, but those extra frames mean nothing to the vast majority of people with 60Hz displays.

 

[Carfax83]

These results really suck. I was hoping Vega would get AMD back in the game, but I guess not. The only hope now is that there is a significant amount of untapped performance that drivers and special optimizations like FP16 can expose over the coming months. Vega needs to be within striking distance of the 1080 Ti, and by that I mean within about 15%. Otherwise it's a complete unmitigated disaster.

 

[Konan]

When AMD's marketing arm puts out:

Yet, all the "review" sites REFUSE to call them out on their shady tactics, one wonders, was it just because they were all bribed off with "free" hardware (polaris / vega / ryzen / threadripper), and those fancy boxes & trinkets & dead CPUs & GPUs?.

I agree with you.

AMD isn't being dishonest in what you posted above. Vega 64 is the fastest GPU in pure compute under $500. And it will get over 60FPS in pretty much every game out. Now saying they are fastest based on compute alone is kind of cherry picking, but getting over 60fps in those games is not dishonest. Sure it may only get 90fps to a 1080Ti's 130fps, but those extra frames mean nothing to the vast majority of people with 60Hz displays.

Where are the 100+ games running 4k @ 60+ FPS? I can find 2/3...

 

[nathanddrews]

Linux benches @ Phoronix

http://www.phoronix.com/scan.php?page=article&item=rx-vega-linux1&num=1

 

[Ranulf]

AMD is stuck with these cards for years now. This looks really bad for everyone. Nvidia is getting used to having no competition at all now. Guess that next Ti is going to be at least $50-$100 more next time. Probably just start charging $1,000 for the Ti models now actually. Screw it, might as well. What can we do? Complain and buy it anyway? Yep.

You can just not buy them. There are days I regret buying into rx480 (and then not selling them earlier this year at the height of mining) but at least by staying at a lower res I don't worry about outlaying $500+ for a decent video card. Of course, as you say, we're looking at another $50-100 rise at each performance tier again when the mining isn't sucking up supply anyway.

 

[Mopetar]

AMD is stuck with these cards for years now. This looks really bad for everyone.

I don't know if that's necessarily true. I think that while Vega is seriously underwhelming, that AMD have probably put a lot of different technology in it that they intend to use over the next several years. It's just that a lot of it is half-baked right now.

Take the 7970 which was the first high-end GCN GPU from AMD. It certainly faired relatively well for its time compared to its contemporaries, but it is a card that did age particularly well because AMD had put a lot things into their architecture that took a while to be used, but there are still people using that card today and getting reasonable performance out of it.

I don't think Vega will be quite the same long-term success story as Tahiti was, but I don't think it's going to be as bad as some think either.

 

[USER8000]

Digital Foundry and Joker reviews.

Vega 56 performance actually looks better than I expected against the GTX1070. I thought it would be slower.

 

[Elixer]

... but getting over 60fps in those games is not dishonest.

I think there is a problem here, I define dishonest as "intended to mislead", so, if any one of those games, @4K don't hit 60+ or is greater than 60.0 in "over 100+ games", then, that is misleading is it not?
Not having some of the latest games out there listed, is misleading right?

AotS:E hits 59.1 looks like it is under 60.
.9 frames you say... well, AMD said 60+, not ~60. AFAIK, most of those games shown are also really old, and nobody that has a Vega actually tested them besides a few.

We got the "poor volta" rhetoric, "drivers have disabled features", " and on it goes, if you want specific examples, check the other Vega thread that was locked, I posted lots of them.
"GPU King?" I think the emperor has no clothes.
https://youtu.be/V9yTlRlxSVc?t=1047 "missing features, not enabled yet, lots of surprises."
But, this is beside the main point here I was trying to make.
If company A tries to mislead, then it is a fair reviewer's job to keep them in line.
When reviewers fail (like the did with the 970), it took the average joe to bring these kind of shady tactics to light.

If AMD does this crap, they should be nailed on it.
If Nvidia does this crap, they should be nailed on it.
If Intel does this crap, they should be nailed on it.

That is all I am trying to say. don't let them get away with it.

 

[USER8000]

I have a 980ti and probably wont be upgrading for a while (Nvidia or AMD). But I have to say, sometimes these comment sections seem like they are just a competition to see who can write the most elegant and embellished post on how bad a product is. It cracks me up sometimes.

Yep,I am on a GTX1080 here,and some of the nerd rage is almost like AMD went and killed someones beloved Cat,for it to repeated again next year on the next release cycle.

Its even more hilarious,when you have a quick peek on Steam,most people are on slower cards than even a GTX1070,and the most represented "performance" card is a GTX1070 which has more owners than all the GTX1080,GTX1080TI,Pascal X and Pascal Xp cards!! Even the GTX1070 has less owners than the GTX750TI....the GTX750TI. Yep boys and girls you read that right.

 

[Zstream]

I think there is a problem here, I define dishonest as "intended to mislead", so, if any one of those games, @4K don't hit 60+ or is greater than 60.0 in "over 100+ games", then, that is misleading is it not?
Not having some of the latest games out there listed, is misleading right?

AotS:E hits 59.1 looks like it is under 60.
.9 frames you say... well, AMD said 60+, not ~60. AFAIK, most of those games shown are also really old, and nobody that has a Vega actually tested them besides a few.

We got the "poor volta" rhetoric, "drivers have disabled features", " and on it goes, if you want specific examples, check the other Vega thread that was locked, I posted lots of them.
"GPU King?" I think the emperor has no clothes.

But, this is beside the main point here I was trying to make.
If company A tries to mislead, then it is a fair reviewer's job to keep them in line.
When reviewers fail (like the did with the 970), it took the average joe to bring these kind of shady tactics to light.

If AMD does this crap, they should be nailed on it.
If Nvidia does this crap, they should be nailed on it.
If Intel does this crap, they should be nailed on it.

That is all I am trying to say. don't let them get away with it.

Than you need to turn off all ads, commercials etc.. stop being a dolt.

 

[renz20003]

What the reviewers aren't telling is you need to pair vega with a ryzen chip. It quadoubles the performer with all dem moar cores an threadsm.

 

[USER8000]

Than you need to turn off all ads, commercials etc.. stop being a dolt.

Shock and horror as the marketing department of a company does marketing stuff. I tell you Apple never would try ridiculing their competition.Oh wait.

 

[urvile]

Well the vega 64 is now available here. The AIO version costs more than gtx1080 AIO version. It should sell well.

EDIT: Well I stand corrected because they sold out and so did the vega 64 air.

 

[dark zero]

Seems that the only good card here is Vega 56. Vega 64 and Liquid are a dissaster.

But the real star will be Vega Nano. When is launch, very few cards of the same kind of Nano will stand a chance against it.