Review: ASUS Radeon ROG RX Vega 64 STRIX 8GB

I would go for 56 over 1070 easy. 64 and 1080 I would lean towards 1080 as of now, even tho 64 can be good with undervolting.

 

Oh ASUS... bet you guys are kinda bummed after that oops stock BIOS.

 

Well, at least we will have two result sets to compare. Old BIOS versus ASUS BIOS:

 

Well, it will be interesting to see what kind of performance increase you can get transferring to liquid (as well as any power increases vs performance gained), but I doubt it's gonna be much - probably in the realm of not worth the hassle I reckon. I mean if it's fun to slap on some watercooling & you like to see the lower temperatures & less noise then why not.

 

I certainly would like to see what the performance is like on Vega when all it's features are being implemented/used but the trouble is, AMD are now banking on future use than it's current performance.

Now obviously most people prefer to see how it performs right now and so far it's shaping up to match the 1080 but the idea of it's future performance does remind me of the 2900XT and how it took several generations for Terascale to become a successful architecture (notably 4000 and 5000 series), on the otherhand, GCN has proven to be a long lasting architecture so who knows, perhaps Vega will indeed be worthwhile in the long run, just we aren't able to see that speculative performance right now and that's the problem.

I do hope this isn't another 2900XT lemon but at the very least AMD have something to compete with the 1080 albeit with high power usage but I guess that's the trade off for having the superior compute functionality (lets be honest, it is a compute monster).

 

Nice QA from their side.

 

What I find especially interesting about this is that still suggests their cooler isn't a whole lot better than the reference cooler. In fact it's arguably worse, since the fans seem to consume more power. Think of it in this way: if the BIOS was the same as the reference and the overall performance, thermals, and noise levels remained roughly the same, that suggests their heatsink is no better [for Vega] than the reference. That means whatever improvements they make in the BIOS could be applied to any other card, including the reference. So, why wouldn't anyone just go for the reference card (or any other 3rd party OEM) and just tweak clocks and voltages themselves?

No wonder Asus wanted this article taken down. Either way, that's nice they're sending a replacement. I personally would've been fine with just updating the BIOS but a new card is nice.

Hilbert:
If you're allowed to keep the original review sample, you should try replacing its BIOS with the one from the replacement card. From there, you could do Crossfire tests!

 

Vega64 isn't in the 1080Ti price range, it is in the 1080 price range.
Its doing pretty well against the 1080, in some Games, and loses in others.

FFS, I'm not pro AMD, i'm more Anti-Nvidia, due to them buying PhysX, privatizing it, and then shafting hybrid PhysX owners and Ageia owners.
Nvidia is Pro-Nvidia's wallet. They are in it for the Money only. They have shown time and time again, that oif the can F*ck the consumer, they will.
AMD is more Pro-Consumer. So I'd rather give them my money. Thats it.
edit : removed "stop calling names" as it wasn't you who started this.

 

I am already hitting the "power wall" with the standard BIOS.
An extra 50W on the TDP might help it go over 1700MHz, I hope. I don't care about power consumption, its a gaming PC first and foremost, FPS for the win.
The watercooling is specifically to remove temperature from the list of possible problems. Its not really quieter, since all the fans are at 12V.

With the TDP thse cards have, pushing them is going to require watercooling. All of the add-in manufacturers are going to do multi-fan heatsinks, which will dump all that heat into the inside of the case.
At least the turbine cards dumped the heat outside of the case ASAP. I actually prefer them to the aftermarket GPU air coolers.

 

For Freesync I'd go to V56 and undervolt maybe, but even isn't the V56 more power hungry compared to 1070? Problem is: I have freesync disabled because my motherboard freaks out. Also, with enhanced sync it's not like I need freesync anymore.

It would be ok if it didn't sip much more power than the 1080Ti just to be at 1080 performance.
Also, If I wanted to buy a V56/64 at MSRP I wouldn't be able to.

 

My point being, VEGA failed.

 

You say they're both noisy, so that's not a win there unfortunately, all that remains is any potential benefit in performance. How much extra performance have you got from watercooling vs your card when it was on air? I suppose it's best to compare overclocked on air vs overclocked on water, and Timespy Graphics Score would be a good comparison.

EDIT: in the Vega64 Guru3d review in the overclocking section they say of their results that in the "first segment of our test would run nicely at 1.7 GHz, but the second it hits a limiter it goes down fast to 1500 and 1600 MHz ranges". I suppose if you manage to maintain a stable/constant 1700Mhz with watercooling (don't know if you see that with yours) then this is 'only' 6% above the 1600Mhz seen in the air cooling - performance in games doesn't scale linearly with clockspeed, so I'm guessing you would see in the region of 4% improvement in gaming performance through watercooling. Is that far off?

 

i know folks have assembled into their pro/con camps. but a little plain talking fact will help (as much as it can...lol).
Vega is an unqualified success for AMD, a wtf moment for gamers.
these are not mutually exclusive statements.
AMD may or may not have aimed for miners, but more than miners use openCL.
and the real PC Master Race is workstation users anyway.

Nvidia is literally up against a wall (not from AMD) due to die size and chip yields for Volta. atm, Volta is not ready for mass market and it is unlikely it ever will be. what we will probably end up with is a new smaller die, instead of less perfect Volta chips with disabled functions.

AMD has already dealt with the issue of massive chip size and die yield - its called Ryzen (in all its flavours). Infinity Fabric is the answer and both Nvidia and Intel are spending tens of millions of R&D dollars to emulate it. that is not an opinion. anyone can read the trade blogs.

 

Chill a cycle, dude. It wasn't a trolling attempt in the first place, it was hyperbole joke. Nobody here is at the moment expecting AMD to go under and consequently there's no grounds for such trolling. Not that it would have even been the point of the post. The point was that AMD with its meager resources seems to be pushing the development in a far riskier way than Nvidia who could have spent the same money as a side project.

Don't get stuck on individual words. You might turn into a political journalist!

 

When I read the initial review I thought wft and scoffed.

 

In the Superposition Test I can see the GPU MHz in realtime.
At the best, I hit 1675MHz, and dipped to 1666MHz. At 44°C Max temp at the end of the run.
I did manage to get it above that, but it failed in one of the later tests.
Lowering the voltage to 1170mV/+5%OC GPU HBM@1100MHz, 1000mV was the sweet spot for me.
Any higher, and the clock would go down, any lower and it would fail. i.e. Lock up the PC.
I have rebooted god knows how many times doing the voltage/OC tests.

edit : The HBM made more of a difference, in a way, where the GPU clock is more pure FPS.

 

Ok, so you're not gonna show the difference between your air cooled performance & your now watercooled performance. From what I've read water cooling doesn't really significantly increase performance of Vega64, I'd be surprised if it's more than 4% performance difference between air cooled overclocked vs watercooled overclocked in game fps.
EDIT: although you do say you're hovering at least at 1666Mhz, and the Guru3D overclocking part of the Vega64 review showed overclocked on air dropping to about 1600Mhz, so that's only about a 4% increase in boost clock, so gaming performance increase of water cooling would be less than 4%. (Just some suppositions based on what you've told me so far).

EDIT 2: if you're not getting much performance increase from watercooling, then why don't you go in the opposite direction & instead let the temperatures climb a bit higher, and instead turn down your fan speeds, at least that way you end up with one of the benefits - quieter than a stock Vega64 air cooled card. You may as well get at least one practical advantage!

 

^^ This
I would've expected the cooler to work way better than it did on a Vanilla 64 bios compared to the ref blower card, I would've expected to at least maintain boosts for longer yielding noticeable better performance overall.
Either that AMD ref design blower cooler is the best one ever made or something went terribly wrong with the ASUS cooler!

 

I can't show you aircooled performance, because it was never aircooled.
I don't have a PC setup where I can really stuff an aircooled card in it, and I had the block ready, so i just went with it.
I set the overclock to 5%, only had it running a week, and have seen that people can set a target clock.
Not tried that yet, but I dont expect it to do better than what I have tried so far.
The Liquid BIOS has a higher TDP allowed, which might allow 1700+ stable, we'll see.
edit: 3DMark had my GPU at 1697MHz (1630MHz in brackets)
edit2: Noise isnt a problem, 5m long DP cable, USB cable and PS2 cable Other side of door. Out of mind and the speakers.

 

Ok, I understand, that's why you couldn't show me the difference in performance. Well it's good that you can keep your PC in a separate room! Ha, but joking aside, one good thing about your card, I imagine it will stand the test of time better than a GTX 1080 - not by much, but a little I predict. I'm still not a convert to the idea that it's worth watercooling these Vega's from a performance perspective, but if you can reduce the god-awful noise of the reference Vega64 then that's a pluspoint, but given the cost & hassle it might be better to get an aircooled 3rd party card instead - although this review that got taken down wasn't that promising!