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

[nurturedhate]

I wrote that they were both hit and miss, not that they were equal.

But arguing that AMD usually benefits more from undervolting than NVidia, is like arguing fat people are better at losing weight than skinny people. NVidia is very power efficient to start with, so of course the gains on that front are going to be smaller.

You've been implying for pages that Nvidia undervolts just as well otherwise there is no need to bring Nvidia into a conversation regarding AMD and undervolting. You get called on it and you backtrack, now stating "of course they aren't equal". The vast majority of your posts have been in vega threads and all display a pattern of showing vega in the worst light possible (something it doesn't need your help doing).

 

[Hitman928]

Another quick update, HBCC seems to have a good effect on Crysis 2 (sorry, never bought 3) at 4k as well. Doom seems to run the same either way. I'll try and get actual data up over the weekend, just been busy lately.

 

[crashtech]

It's not really competitive on Perf/Watt either.

Even a distant second is still second. Perhaps that's not what you call competitive, but it's all we have, and the only thing preventing Nvidia from having a total monopoly in dGPU space.

 

[krumme]

The top voltage states p6 and p7 it set similary on all vega variants from rx 56 with top boost 1471 to rx 64 liquid with top boost 1677.
I think perhaps those states is set to make sure 1677MHz can be reached on all full die bins with 1.2V. Rx 56 get same treatment.
Unlike eg TR vs regular r7 there is no binning.
Aparently there is to few sold for it to make sense and then the aib can do whatever they want.
Its the only explanation that i can see for all those regular 56 and 64 to reach such absurd low voltage; The 14lpp makes a frequency wall and their voltage is set up the steep hill so even a 100 MHz or 200 MHz lower freq means voltage can drop a freaking lot. Like typically from 1200 to 1050.
We need more data from the liquid cooled but the one we have in this forum cant undervolt and its aparently running 1750. Makes sense so far.
Whats your take on it?

 

[JoeRambo]

On average you will gain far less from undervolting Nvidia (performance vs power saved) compared to AMD. Furthermore, you have to know this. It's pretty well known.That only leaves your post and its intentions.

Is it so? On this very forum there is a guy who tested 1080TI undervolting. He is running @ like 150w and still getting great performance*. There are plenty of ppl who are running 0.85-0.9V ABOVE so called stock boost clocks. And plenty of ppl who can hit 2Ghz @ 1V and 1900 @ 0.9. If that is not great undervolting, then I don't know what is.

To put 150w on 1080TI in context, realise that AIB cards have up to 330W power limits, so absolute savings are also massive here.

 

[nurturedhate]

Is it so? On this very forum there is a guy who tested 1080TI undervolting. He is running @ like 150w and still getting great performance*. There are plenty of ppl who are running 0.85-0.9V ABOVE so called stock boost clocks. And plenty of ppl who can hit 2Ghz @ 1V and 1900 @ 0.9. If that is not great undervolting, then I don't know what is.

To put 150w on 1080TI in context, realise that AIB cards have up to 330W power limits, so absolute savings are also massive here.

There's always outliers for about anything and shouldn't weighted too heavily otherwise we would have to start taking the PUBG Vega stuff more seriously. However, that does not change the overall average and that is Nvida does a better job at setting proper voltage for their chips than AMD leading to AMD cards gaining more from undervolting.

 

[Muhammed]

However, that does not change the overall average and that is Nvida does a better job at setting proper voltage for their chips than AMD leading to AMD cards gaining more from undervolting.

Without a proper statistical analysis this is just hogwash.

There's always outliers for about anything and shouldn't weighted too heavily otherwise we would have to start taking the PUBG Vega stuff more seriously

Well, fantastic lets also use all the cases where Vega 64 is merely 1070 performance, I am sure that will paint Vega in a more positive light aright!

 

[littleg]

To put 150w on 1080TI in context, realise that AIB cards have up to 330W power limits, so absolute savings are also massive here.

Not really. 150W for 8 hours a day at a very expensive $0.2/kWh works out at about $87 a year.

 

[Unreal123]

Some of the reviews did not even compare current price and performance like they did with Nvidia Pascal series.
They are only comparing performance of nvidia third fastest gaming GPU with AMDs flagship AIO water cooler ,which cost $700. Every single benchmark of VEGA 64 needs to compared with GTX 1080 Ti not GTX 1080 and RX VEGA 56 needs to compared with GTX 1080. VEGA is a total commercial failures in terms of marketing, reviews and performance. People are only miss guided by the VEGA reviews nothing more.

Some of the reviews and reddit are people consist with double standard and some of the seriously think that they people owe AMD and they support VEGA due to sympathy but not on merit bases.

I renumber the Rebellion hate AMD campaign and it had a motive to protest against higher priced GPUs. I do not see shills tweets anymore and those people need to tweets against RX VEGA. That is why i do not like AMD because there user base is very aggressive and single minded.

 

[krumme]

Seems more rx 64 and rx 56 at msrp - meaning below 1080/1070 prices is in for scandinavia via komplett (dk/no/se). Batch of 100+ xfr 64 comming the 6 sep, and just plenty of rx 56 - but they now sold out now untill the 15 sep - but plenty to grab now. Nice low prices. While the situation with polaris is still very high prices and next to no availability - so a rx 580 is like 15% less than a rx 56 ref.

The guys at Reddit have started to make statistics about undervolt results and settings here
https://docs.google.com/spreadsheet...beRiOTMsGeIkCDQUhXSBA/htmlview#gid=1964178281

 

[Tup3x]

There's always outliers for about anything and shouldn't weighted too heavily otherwise we would have to start taking the PUBG Vega stuff more seriously. However, that does not change the overall average and that is Nvida does a better job at setting proper voltage for their chips than AMD leading to AMD cards gaining more from undervolting.

Not really... It just means that NVIDIA doesn't have to squeeze every possible marketing MHz & % because they are ahead. If that would change then obviously NVIDIA would start squeezing everything out of their hardware at the cost of power efficiency.

 

[tential]

Remember how there was a large contingent on here insistent that AMD helps lower prices?
Thanks to the inflated Vega prices, all of a sudden Nvidia now has "issues" and has to raise prices.
In reality, AMD GPU shortages during launches are so bad that retailers price the GPUs really high, and then they realize that people will still actually buy them at those levels and next year, we get higher priced GPUs. Really, you can make a case for AMD helping to push up GPU prices a ton by showing the prices people will pay for subpar performance.

After the last 3 high end launches, there really isn't anything positive that happens at the initial launch of a high end AMD product.

 

[stahlhart]

This thread is for discussing Vega reviews. Back on topic, or if you're done discussing them it's getting locked.
-- stahlhart

 

[Unreal123]

This a simple conclusion of AMD RX VEGA. It was even worth to release. Competition are beating RX VEGA at everything expect mining with much lower prices.

 

[jackstar7]

If mining keeps them going (or even lifts them), then I'm okay with that. But sitting as I am on a 980ti, I haven't seen a compelling reason to upgrade and I'll just keep waiting.

I will lean AMD if they have something compelling and available when the time comes, as I Freesync and it's lack of significant price bump appeals to me. It's usually about features for me and less performance, just because few new games are at all compelling to me these days. PUBG and the discounted NMS should keep me for a long while and they just don't require that much grunt.

 

[maddie]

There goes my chance of a Vega 56 in the next few months. If accurate, AMD will sell every one made.

https://www.hardocp.com/news/2017/0...s_polaris_in_ethereum_mining_efficiency_by_2x

Redditor S1L3N7_D3A7H:
"So, I just managed to tweak my Vega to 1000 MHz Core and 1100 MHz HBM with a power target of -24%.

Getting 43.5 MH/s in ethereum. Pretty sure I can get similar with Vega 56. Things can only get better from here.

Edit. Here is an album with my screenshots and a picture of my desktop: https://imgur.com/a/1GgvR

2nd Edit. To clear up any confusion, 130w is total card power, not core power. Core power is around 104w."

 

[itsmydamnation]

My Vega 56 sits at 1050mv @ 1580-1600mhz and I haven't started pushing it yet. But wattman is very nice

 

[darkswordsman17]

Some of the reviews did not even compare current price and performance like they did with Nvidia Pascal series.
They are only comparing performance of nvidia third fastest gaming GPU with AMDs flagship AIO water cooler ,which cost $700. Every single benchmark of VEGA 64 needs to compared with GTX 1080 Ti not GTX 1080 and RX VEGA 56 needs to compared with GTX 1080. VEGA is a total commercial failures in terms of marketing, reviews and performance. People are only miss guided by the VEGA reviews nothing more.

Some of the reviews and reddit are people consist with double standard and some of the seriously think that they people owe AMD and they support VEGA due to sympathy but not on merit bases.

I renumber the Rebellion hate AMD campaign and it had a motive to protest against higher priced GPUs. I do not see shills tweets anymore and those people need to tweets against RX VEGA. That is why i do not like AMD because there user base is very aggressive and single minded.

This is not uncommon and this is not something exclusive to Vega. Generally at launch most reviews base it on MSRP (as that's all they have to go on, as typically the reviews are posted at the same moment cards go on sale because they were actually written up before; you know that, right, that these companies send out review samples that get tested by sites prior to the actual launches so there isn't actual real world pricing info for them to even base things on yet), then after things have settled down will mention real prices when doing buying guides or individual card reviews.

Its...interesting...how we went from weird, what seems to now be, conspiracy theory, about AMD doing some weird price manipulation, to now them being condemned for retailers jacking up prices (which is not at all a new thing).

You really must be new to video card market if you're castigating AMD alone for shills and aggressive single minded supporters. Many of the people you're complaining about became that way because of things Nvidia did (for years and years) where they intentionally had people troll forums and spread FUD about ATi in exchanged for free video cards. Also baffling is how you don't notice there's people exactly like you describe that are like that about Nvidia (sadly, you can find that about just about every company these days for some bizarre reason).

Especially odd is that, I haven't seen a lot of blind AMD support (at least on here, I don't know maybe if you go to some fanboy places, which I don't know why you'd bother doing that). By that I mean, the people that are not joining the "fire Raja, fire everybody, AMD needs to burn!" crowd have pretty consistently said that prices are stupid now (and it might continue due to mining), and that even at MSRP the cards are at best just competitive in gaming performance with the 1070 and 1080 right now. So, even at MSRP they're aren't great buys. I don't know what more people want to be said?

If mining keeps them going (or even lifts them), then I'm okay with that. But sitting as I am on a 980ti, I haven't seen a compelling reason to upgrade and I'll just keep waiting.

I will lean AMD if they have something compelling and available when the time comes, as I Freesync and it's lack of significant price bump appeals to me. It's usually about features for me and less performance, just because few new games are at all compelling to me these days. PUBG and the discounted NMS should keep me for a long while and they just don't require that much grunt.

I've never understood why people have been so angry at AMD over mining. They weren't hurting their AIBs or retailers (in fact it helped the retailers and very likely AIBs). I don't even think you can argue that it hurt gamers, as they had alternatives, and mining was boom and bust, leading to fairly cheap and abundant cards for gaming (290s at $200 in 2015). We'll see what it does on Polaris and Vega (and Nvidia's offerings even) this time and moving forward (if Nvidia and AMD cards are both good enough that the alternatives dry out).

Absolutely. Unless you just had tons of money to burn, I don't know why you'd even be considering an upgrade yet. If I bought a top of the line card, unless there was some major upheaval, I'd be expecting to go 3 years before even looking at changing. Not only that, but with how display stuff is changing (HDR, adaptive sync, changing inputs like Thunderbolt, USB, DP, etc; even in VR where prices are seeing shakeups), I'd wait for more stability before I'd drop a lot of money on monitors. Hopefully we'll see HDR standards, we'll get adaptive sync standards, and prices will drop.

 

[darkswordsman17]

My Vega 56 sits at 1050mv @ 1580-1600mhz and I haven't started pushing it yet. But wattman is very nice

Nice.

Definitely would like more data. Seems like an opportunity for AIBs as well, where they can do some testing and figure out more reasonable voltage settings,maybe even bin chips and offer variations (i.e. some people might value efficiency more, and testing more chips would likely help them find better overclockers too; either way they could potentially get more money or just differentiate themselves from other brands).

I'd also love for more reviews to go into this and push for these companies to do something about it. I've never understood why this specific issue wasn't brought up to or about AMD before (I've seen some sites do it, I'm talking about as an industry), as its been a consistent problem across their GPUs and CPUs for years. We've been openly discussing overclocking for years, which uses the same tools and methods, so there's no reason this can't be addressed similarly.

 

[Krteq]

Some info from Sebbi

Vega has some advantages, but none of them are seen in current games or current software. It's hard to predict how much these new features will matter in the close future (when Vega is relevant).

For gaming, the biggest change in Vega is full DirectX feature level 12_1 support. Both Intel and Nvidia had already released two GPU generations with these features. These features allow implementing more efficient algorithms for many rendering problems. Now that AMD also supports these features, we can expect more game devs to start adding support to their games. AMD has slightly higher DX12 feature level support than Nvidia regarding to some aspects. Resource heap tier 2 (vs 1), conservative raster tier 3 (vs 2) and support for stencil ref from PS. My prediction is however that none of these slight advantages will have an actual impact during the Vega life time. But the common 12_1 features (that are supported by all IHVs) will certainly be used in some games, and that's when Vega will prove to be highly superior compared to previous AMD cards. There are also some other new gaming related features (such as primitive shaders) that are not yet exposed to developers. Whether these features will be used during Vega life time (on PC) remain to be seen. Evolution of PC graphics APIs hasn't been that fast lately. I don't expect big changes that only benefit one IHV. Consoles could of course be a good proving ground, but it will take some time until Vega base GPUs reach consoles.

For compute, Vega has double rate fp16 support. This is handy for some compute problems, such as deep learning. Nvidia's consumer cards don't have this feature (it's limited to professional P100 and V100 + mobile). Vega also has a brand new CPU<->GPU memory paging system. The GPU on-demand loads data from DDR4 to HBM2 at fine granularity, when a page is accessed. This allows 8 GB Vega to behave like a GPU with much larger memory pool, assuming that the memory accesses are sparse. This is a common assumption in games and also in many professional software. Usually games only access a tiny fraction of their total GPU memory per frame and the accessed data changes slowly (smooth camera movement and smooth animation). This could be a big game changer if games start to require more than 8 GB of memory soon. If a 8 GB Vega behaves like a 16 GB card with this tech, it could greatly extend the life time of the card. The same is of course true for many compute problems. However Nvidia Pascal also has a similar GPU virtual memory system for CUDA applications, but their tech doesn't work for games. CUDA has an additional page hint API to improve the efficiency of the automated paging system. Developer can tell the system to preload data that is going to be accessed soon. Whether or not the automated paging system will be relevant during Vega life time depends on software and games you are running. Xbox One X console with 12 GB memory (and a 24 GB devkit) is coming out very soon. Maybe we need more than 8 GB on PC soon. Let's get back to this topic then.

Currently the appeal of Vega is mostly forward looking. There's lots of new tech. AMD was lagging behind the competition before, but now they have a GPU that has lots of new goodies, but no software is using them yet. It remains to be seen whether AMDs low PC market share is a roadblock for developers to adapt these techniques, or whether AMDs presence in the consoles helps AMD to get high enough adoption for their new features.

Beyond3D forums

 

[krumme]

Some info from SebbiBeyond3D forums

Obviously the scalability of Vega is also meant in relation to consoles. When i think Zlatan is more right than Sebbi about the time perspective for ngg around nov/dec its because he has got insight into the new stuff from both AMD and NV months if not years in advance working with the console stuff. And btw where did he get the info and knowledge about the lacking HBM2 pseudo channel?

Can the reason for some of the delay be that its difficult to get the NGG to work together with tiled based rendering and the Pseudochannel? - and you need to tune the entire avfs to it.
- ofc. the pseudochannel might be broken, it is a possibility but this stuff is technologogy from 2015, so it doesnt strike as such high a probability, and come on the actually meassured peak discard rate is from a beta branch of 17.3.. Yeaa its not that simple but come on

I dont care about all the new dx12 stuff and FP16, advanced HBMC, rapid packed math what not. Yeaa we might get some small immidiate benefit from rendering features here because they were a step behind NV here, but otherwise to h... with it. Its technology for 2 years down the road. Look at the power consumption of the card. They NEED this peak discard rate up and tiled based rendering working together, so these card stop working on stuff nobody sees, and eg a 25% memory utilization uplift to match it. It would suck big time if we had to wait 1½ year for that vital hardware to be enabled with the drivers.

The problem beeing also all cards is sold, and perhaps miners get onto them, and then AMD have far less incentive to force this, but is better concentrating on the pro market. Cant blame them if their vision for the gpu market is meassured as the best cashflow.

 

[majord]

sorry if this was already posted. forum search seems broken tonight

http://radeon.com/_downloads/vega-whitepaper-11.6.17.pdf

has some data on DSBR , amongst a lot of other stuff

 

[Headfoot]

I'm sure the primitive shader and new fast path for geometry isn't out yet because it's really, really, really hard to program a smart driver / compiler to take instruction streams from the older programming model and re-jigger them to not only work in the new fast path, but to have no regression cases and to have a substantial number of cases where there is noticeable speedup. Probably less than 1% of all programmers in the world could even come close to knowing how to write code like that. This is really hardcore computer science and engineering stuff. So no doubt its taking a long time because its really hard and AMD doesn't have a massive high end dev team (but certainly a respectable one).

 

[kawi6rr]

Yes, great idea. People on these forums and elsewhere kept saying "Wait for Vega" for more than a year, and what did it get anybody? Just time spent NOT playing games.

Judge by the here and now, not the fantasies that some people cook up.

This had me laughing pretty hard. So you're insinuating that everyone waiting for Vega were not playing games? Every time I read that I can't help but imagine a 5 year old's voice stomping his/her feet. I'm pretty sure most people probably had a video card to do something with.

 

[Rifter]

Some info from SebbiBeyond3D forums

Sounds to me like even sebbi is doubting the magical drivers/features increasing performance soon.

Hes right about one thing though, if anything saves AMD its going to be the consoles, as they just dont have the PC GPU market share to push the devs to code for it on PC. There only hope is that the consoles push the devs to use the new features, and that has a trickle down effect to the PC ports.

The thing i dont understand is these new consoles arnt even out yet, if we are going to have to wait for games developed for the new consoles to begin to see improvements from new vega tech/features thats going to be to late, it may help the next gen navi but i dont see anything gained from next gen consoles having a meaningful effect for vega, by the time new consoles and games are out it will be volta/navi time.

I like that AMD is generally forward looking, and that there products generally age well. But with vega if sebbi is right and there is literally no current software that use any of this new tech then AMD has blundered and looked to far ahead for there own good. Perhaps they should have released vega as compute card only, and only moved it into RX line after new consoles and new tech have a foothold in software, and there is software out that can actually use all this new tech perhaps when the navi version is out.

Or perhaps this will not take as long as i suspect and everything will be rainbows and unicorns as soon as AMD flips on the features in the drivers.