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

Page 42 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136

Good summary. It would be much better if AMD just launched something like NVidia FE models at a $499 V56/$599 V64 and said AIB could sell up to $100 lower prices later. Much like the way NVidia launched.

I laughed out loud when he said "The intent is either malicious or it's incompetent, and incompetence here would be so bad, that I would actually hope it's malicious". Yikes!

Though I disagree with paying $100 more for better cooling would help that much. The problems is fundamentally the huge power draw. Coolers don't fix that. They just fix noise.
 

Qwertilot

Golden Member
Nov 28, 2013
1,604
257
126
Well presumably they help a little with the power draw as keeping things cooler does help a bit. Of course doing that just ups the overall power draw, as witnessed by the liquid cooled version.....
 

Crumpet

Senior member
Jan 15, 2017
745
539
96
Isn't this an illogical statement?

"...............Vega 64 is 2 years too late, too hot, too thirsty, and he can't recommend it over MSRP (499), nor does he think it will ever come down to MSRP again, and his personal expectation is that custom board 64's are going to be firmly in 1080ti pricing, making them a very bad sell................."

Vega 64 is a bad sell above $499 and yet it will never drop to that price again. I feel like such a fool in being unable to follow that reasoning. Is no one buying them? The unstated assumption is that AMD does not want to sell any Vega 64. Correct?

There is always someone willing to make a very poor purchase decision.
 
Reactions: Phynaz

Jackie60

Member
Aug 11, 2006
118
46
101
AMD Vega has an optimization in "14 nm +", according to road map from Financial analyst day. Could do wonders like NVIDIA GeForce 400 to 500 series, or be almost rebrand-like as in AMD Radeon RX 400 to 500 series. Optimized Vega may answer NVIDIA GeForce GTX 1080 Ti.
I really hope it does but I'm doubtful I'll be buying another AMD GPU in the foreseeable future. I was ready to go on VEGA crossfire until AMD launched an abortion that I really can't buy. Reminds me of the 2900/3870 days.
 

fleshconsumed

Diamond Member
Feb 21, 2002
6,485
2,362
136
I'm skeptical of undervolting based on my personal experience. I remember reading about 50-100mV undervolt on 290/Fury/Polaris for forever. I've never had much luck undervolting AMD GPUs. My latest 480 undervolt attempt I ended up undervolting by just 24mV which doesn't save me any power. When I went for 39mV, I got a video driver crash less than 24 hours later. I think undervolting reports are like bad reviews, you tend to see extreme results on the net and think they're the norm when they're not. I think Balanced Profile from AMD is best optimization you're going to get, anything else is a lottery.
 

jackstar7

Lifer
Jun 26, 2009
11,679
1,944
126
I really hope it does but I'm doubtful I'll be buying another AMD GPU in the foreseeable future. I was ready to go on VEGA crossfire until AMD launched an abortion that I really can't buy. Reminds me of the 2900/3870 days.
Oh man that makes me hope it is akin to those 2900/3xxx days because they really sorted things out and had a few solid generations after that with the 4-series (not bad) and the 5-series (great) especially.

And based on reviews so far I feel comfortable waiting potentially a generation or two with my 980ti just continuing to chug along. Then I'll be able to get a solid upgrade and jump to a x-sync monitor of whichever makes the most sense.

Ahh, the old waiting game. Enemy of my itch to always upgrade.
 

Magic Hate Ball

Senior member
Feb 2, 2017
290
250
96
I'm skeptical of undervolting based on my personal experience. I remember reading about 50-100mV undervolt on 290/Fury/Polaris for forever. I've never had much luck undervolting AMD GPUs. My latest 480 undervolt attempt I ended up undervolting by just 24mV which doesn't save me any power. When I went for 39mV, I got a video driver crash less than 24 hours later. I think undervolting reports are like bad reviews, you tend to see extreme results on the net and think they're the norm when they're not. I think Balanced Profile from AMD is best optimization you're going to get, anything else is a lottery.

Eh, I could undervolt my Fury by 60mv without issue for months and months (that I owned it before selling it to the mining craze) and my RX470 I can undervolt by 75mv no problem at stock clocks, and I play on it all the time.
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,362
5,029
136
The unstated assumption is that AMD does not want to sell any Vega 64. Correct?

I'm sure they are selling every single Vega GPU they make, though I wouldn't be surprised if the bulk of production is going towards Instinct GPUs/higher margin products... and Apple.

Consumer Vega appears to be the rejects/scraps that didn't make the cut, shoehorned into the gaming role.
 
Reactions: beginner99

cfenton

Senior member
Jul 27, 2015
277
99
101
Isn't this an illogical statement?

"...............Vega 64 is 2 years too late, too hot, too thirsty, and he can't recommend it over MSRP (499), nor does he think it will ever come down to MSRP again, and his personal expectation is that custom board 64's are going to be firmly in 1080ti pricing, making them a very bad sell................."

Vega 64 is a bad sell above $499 and yet it will never drop to that price again. I feel like such a fool in being unable to follow that reasoning. Is no one buying them? The unstated assumption is that AMD does not want to sell any Vega 64. Correct?

I think there are two ways to make sense of that statement.

1. Vega 64 is a bad deal for gamers (this guy's target audience), but a good deal for some other segment of the market (compute?). If the supply is limited, or Vega 64 is an especially good value in these other market segments, then they might never sell for MSRP.

2. Some people will make bad purchasing decisions. Some people really hate Nvidia, or really love AMD. I think Vega 64 is close enough to a 1080 that many people can justify the higher power usage to support AMD, even if it costs them more money.
 
Reactions: Crumpet

Rifter

Lifer
Oct 9, 1999
11,522
751
126
I'm skeptical of undervolting based on my personal experience. I remember reading about 50-100mV undervolt on 290/Fury/Polaris for forever. I've never had much luck undervolting AMD GPUs. My latest 480 undervolt attempt I ended up undervolting by just 24mV which doesn't save me any power. When I went for 39mV, I got a video driver crash less than 24 hours later. I think undervolting reports are like bad reviews, you tend to see extreme results on the net and think they're the norm when they're not. I think Balanced Profile from AMD is best optimization you're going to get, anything else is a lottery.

Fully agree, my RX480 will not undervolt at all, which is fine it OC's alright and runs cool, but undervolting is NOT a grantee any more than OCing is.

But i suspect not all Vegas will undervolt well, as the chips they chose to use are binned so poorly to require the stock voltage in alot of cases. Otherwise why set stock voltage so high? Maybe this situation will improve in a few months after launch as they build up a chip supply and are not forced to use as many marginal chips, but by then they will have a bigger problem with Volta.

Any way you try and spin this Nvidia already has 2 models firmly above vega(3 if you use MSAA as 1080 blows the doors off of Vega 64 using MSAA), and when volta launches in 6 months the GPU reviews are going to look ridicules with Nvidia having 4-5 cards unanswered at the top end.

Hopefully AMD can get a respin of Vega out by Volta launch to at least try and compete with it, otherwise Nvidia is going to head back to 90%+ market share again over next 2 years.
 
Mar 11, 2004
23,180
5,641
146
With regards to the rebate or whatever between retailers and AMD, could this be a misunderstanding by the retailers? Because how that looks to me isn't necessarily AMD upping the MSRP, but rather them trying to make sure that at least some cards actually do sell at MSRP (and in this case the lowest price), and not marked up like always happens at launch of new graphics cards. They offered the rebate for the retailers because they know they would lose out on money by not marking up the cards (so they're trying to let customers get the lowest price, but also trying to keep the retailers happy by not totally screwing them and in fact somewhat rewarding them for actually selling at MSRP).

Coupled with AMD setting bundles (which normally would be a way for retailers to up revenue by adding other stuff to get people to buy generally high profit margin stuff, and obfuscate their price shenanigans; but with AMD setting official bundle options it limits what they'd be able to do there), and retailers aren't happy as it limits their options, so they try to paint AMD in bad light to try to stop this from becoming a normal thing.

There absolutely does exist proof of what is happening. A deal like this would have a paper trail, either in contract (pretty sure rebates require legal terms, and if I was a company I definitely would not agree to any price and/or sale stipulations I wasn't in control of without legal terms to keep from being screwed), or at minimum financial evidence (with some manner of communication about pricing setup). It really is as simple as the retailers providing this proof if they want it to be accepted. They don't have to post it publicly, but the way they're claiming, I'd be talking to regulatory bodies about the legality of it.

Right now, this seems more like retailers just weren't happy that AMD was doing things to limit their ability to markup cards. Which that's fair for them to complain about to AMD, but they're trying to act like AMD is screwing consumers over, when I think its feasible that they're actually trying to keep retailers from screwing customers with ridiculous markups.

I'm skeptical of undervolting based on my personal experience. I remember reading about 50-100mV undervolt on 290/Fury/Polaris for forever. I've never had much luck undervolting AMD GPUs. My latest 480 undervolt attempt I ended up undervolting by just 24mV which doesn't save me any power. When I went for 39mV, I got a video driver crash less than 24 hours later. I think undervolting reports are like bad reviews, you tend to see extreme results on the net and think they're the norm when they're not. I think Balanced Profile from AMD is best optimization you're going to get, anything else is a lottery.

I don't think its all that rare (I see quite a lot of people able to easily get noticeable improvements and not even seeing how far they can push), but I think you need to tune more than just the voltage though to be able to get the full benefit (and possibly stability). Which I think the only way to undervolt 290 and Fury was with BIOS tweaking or special 3rd party tools, was it MSI Afterburner that was popular for that? Which the BIOS modding seemed to work the best, but that was also on AIB versions that tended to have more robust power delivery setup. Wattman didn't show up until Polaris, and in my experience you need to do more than just go and throw some manual voltage settings. It also seems to be buggy still (hence you need to tweak multiple parameters, and not always in the way you think, seems most people push the power limit up all the way for instance; from what I recall on Polaris you could generally change just the power limit down by about 20% and see a minimal performance drop but decent improvement in power use too without messing with other manual adjustments, which means you will lose a bit of performance).
 
Reactions: dlerious

Elixer

Lifer
May 7, 2002
10,376
762
126
This epoxy fill height difference seems like it could be a real debacle. I mean EK just launched a new full cover block for Vega but will it make proper contact on both? which package did they design around?
The TL;DR version is, it makes no difference at all for thermals. Just use a good thermal paste, and your good to go.
 
Mar 11, 2004
23,180
5,641
146
Fully agree, my RX480 will not undervolt at all, which is fine it OC's alright and runs cool, but undervolting is NOT a grantee any more than OCing is.

But i suspect not all Vegas will undervolt well, as the chips they chose to use are binned so poorly to require the stock voltage in alot of cases. Otherwise why set stock voltage so high? Maybe this situation will improve in a few months after launch as they build up a chip supply and are not forced to use as many marginal chips, but by then they will have a bigger problem with Volta.

Any way you try and spin this Nvidia already has 2 models firmly above vega(3 if you use MSAA as 1080 blows the doors off of Vega 64 using MSAA), and when volta launches in 6 months the GPU reviews are going to look ridicules with Nvidia having 4-5 cards unanswered at the top end.

Hopefully AMD can get a respin of Vega out by Volta launch to at least try and compete with it, otherwise Nvidia is going to head back to 90%+ market share again over next 2 years.

It is true it isn't a guarantee. What do you mean it won't undervolt at all? I personally haven't seen ones that can't undervolt at all. Don't think you'll necessarily be able to both undervolt and overclock, but that isn't the same as just being able to undervolt. If I'm not mistaken, some people have seen no performance loss by undervolting and underclocking (not a lot, but like 1250-1200), since it removed throttling. Plus with the PCIe socket issue on Polaris, there's other things that can be potential causes for instability.

I don't believe it has anything to do with binning. From what I can tell, it doesn't seem that AMD is binning or even putting in work to see what tweaking they can do, they're just taking the chips they get from the foundry and throwing voltage they know works and doesn't degrade the silicon too quickly (I really truly think that is the only "binning" that AMD is doing, and I believe that would even be more the foundry doing that level of testing), and then slapping them on cards, while they try to get the hardware power features working (which they have thus far not actually accomplished). I think this is the result of lack of resources, AMD just does not have the resources to test the chips themselves. And AIBs aren't going to do much of that as it costs time and money and they don't make enough from AMD's products to be worth it for the most part, so they just rely on their improved power setup and cooling to push for some overclock.

One reason I say this is, look at how Microsoft talked up their tweaking for Scorpio (which I think Microsoft is hyping, the reality is that Microsoft is just actually binning chips somewhat, something AMD has not been doing, so it is something actually new being done in consoles).

The talk of them binning for Apple I don't think is true either. I would almost be willing to bet that Apple is just having their engineers set reasonable voltages (either in software or possibly in BIOS). Maybe they're having AMD do it, but I wouldn't be surprised if its all Apple doing the tweaking there.

I could see improvements, although I'm kinda skeptical that we'll see a respin of current Vega. (Did they even call the 580 a Polaris respin? I know they talked up process improvements but I don't recall if they called it a respin as well.) Don't forget there's a small Vega set to replace Polaris coming.
 

Rifter

Lifer
Oct 9, 1999
11,522
751
126
It is true it isn't a guarantee. What do you mean it won't undervolt at all? I personally haven't seen ones that can't undervolt at all. Don't think you'll necessarily be able to both undervolt and overclock, but that isn't the same as just being able to undervolt. If I'm not mistaken, some people have seen no performance loss by undervolting and underclocking (not a lot, but like 1250-1200), since it removed throttling. Plus with the PCIe socket issue on Polaris, there's other things that can be potential causes for instability.

I don't believe it has anything to do with binning. From what I can tell, it doesn't seem that AMD is binning or even putting in work to see what tweaking they can do, they're just taking the chips they get from the foundry and throwing voltage they know works and doesn't degrade the silicon too quickly (I really truly think that is the only "binning" that AMD is doing, and I believe that would even be more the foundry doing that level of testing), and then slapping them on cards, while they try to get the hardware power features working (which they have thus far not actually accomplished). I think this is the result of lack of resources, AMD just does not have the resources to test the chips themselves. And AIBs aren't going to do much of that as it costs time and money and they don't make enough from AMD's products to be worth it for the most part, so they just rely on their improved power setup and cooling to push for some overclock.

One reason I say this is, look at how Microsoft talked up their tweaking for Scorpio (which I think Microsoft is hyping, the reality is that Microsoft is just actually binning chips somewhat, something AMD has not been doing, so it is something actually new being done in consoles).

The talk of them binning for Apple I don't think is true either. I would almost be willing to bet that Apple is just having their engineers set reasonable voltages (either in software or possibly in BIOS). Maybe they're having AMD do it, but I wouldn't be surprised if its all Apple doing the tweaking there.

I could see improvements, although I'm kinda skeptical that we'll see a respin of current Vega. (Did they even call the 580 a Polaris respin? I know they talked up process improvements but I don't recall if they called it a respin as well.) Don't forget there's a small Vega set to replace Polaris coming.

If they arnt at least binning enough to remove the crap bottom barrel 10-15% of the chips thats just sad, and AMD is in worse shape than i thought. And this explains the insane voltage, if you are using every single chip you are obviously going to need to run it way way overvoltage to accommodate the crap chips, but this would mean the top 80% of the chips that do not need this voltage are being overvolted for nothing. Why would AMD do this? it would result in exactly what we see in reviews, insane power consumption.
 

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
Yeah under volting is just as hit and miss as over clocking. My RX480 won't OC that much, only about 75Mhz. But I can undervolt it by 100mV which results in a HUGE drop in power consumption and heat.
 

krumme

Diamond Member
Oct 9, 2009
5,956
1,595
136
Yeah under volting is just as hit and miss as over clocking. My RX480 won't OC that much, only about 75Mhz. But I can undervolt it by 100mV which results in a HUGE drop in power consumption and heat.
Guys get into gear! This is a new situation

For the undervolting test we talk gpu here where lower voltage coupled with lower freq to get stable at that lower voltage means higher gaming performance (goes for the rx64 - the rx56 shows a disproportionate increase in perf vs freq)
Yes more testing needed.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,687
6,243
136
Easy: top bin parts are going to Apple for their thin MacPro design.
Entire AMD SW team developing drivers for Mac ... most likely.
Also Linux drivers changes seems to have slowed down to a trickle.

On one hand on their discussion with PCPer they said the a lot of the Linux Driver changes will follow the Windows changes. So yeah. It looks like Mac --> Mining --> Windows/Gaming --> Linux
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Another update on Vega Pricing.
http://wccftech.com/radeon-rx-vega-64-msrp-pricing-update/
"
Our source has informed us that AMD will be handpicking around 10-20 e-tailers that they will incentivize via rebates to enable the price point of $499 on an ongoing basis. Amazon and Newegg are amongst these and should they choose to accept the rebate then users will be able to buy RX Vega 64 [standalone version] at the promised MSRP of $499. Now these rebates will be provided in batches but on an ongoing basis, which means you might have to wait a week or so before you are able to snag one.
"

The rebates to retailers are a real thing. The reason there are conflicting stories, is because only select retailers got the rebates.

The optics of this, really don't look good for AMD. But par for the course with this messed up launch.
 

zinfamous

No Lifer
Jul 12, 2006
110,810
29,564
146
Another update on Vega Pricing.
http://wccftech.com/radeon-rx-vega-64-msrp-pricing-update/
"
Our source has informed us that AMD will be handpicking around 10-20 e-tailers that they will incentivize via rebates to enable the price point of $499 on an ongoing basis. Amazon and Newegg are amongst these and should they choose to accept the rebate then users will be able to buy RX Vega 64 [standalone version] at the promised MSRP of $499. Now these rebates will be provided in batches but on an ongoing basis, which means you might have to wait a week or so before you are able to snag one.
"

The rebates to retailers are a real thing. The reason there are conflicting stories, is because only select retailers got the rebates.

The optics of this, really don't look good for AMD. But par for the course with this messed up launch.

It flat out looks bad. I think the reality is that the stock is just very, very low and they can't meet demand, so rather than discuss the actual inventory they go with this, which is probably going to backfire in the minds of consumers. But hey, AMD might yet luck out: they continue with these sorts of shenanigans, demand just might plummet and they'll no longer have that inventory issue!
 
Reactions: Kuosimodo

Despoiler

Golden Member
Nov 10, 2007
1,966
770
136
wccftech.....who even visits that site?

It's pretty obvious what is going on. AMD knows they have a higher demand than supply and they know that e-tailers like Newegg will gouge the living heck out of cards in short supply. AMD is literally paying the difference via rebates so that consumers can get a card @ MSRP. Once supply catches up with demand AMD won't have to subsidize predatory e-tailers. The rebates will end.
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
102,425
8,388
126
i'm still trying to wrap my head around exactly why the rebates are horrible. amd can't force retailers to sell the things at $499. that's illegal. it's not even the entity selling to the retailers/distributors; the board partners are. amd also can't force retailers to accept the rebates to sell at $499. again, that's illegal. amd can't force retailers to sell 1 board to 1 person. a) implementing that would be a technical nightmare - shopping cart and account software would need revisions, at the minimum. b) it's also probably illegal.

so, what is amd to do? it can't set prices at the retailer level. it probably can't sell boards directly into this segment under its contracts with its board partners. it also knows that all graphics cards are getting bid up to what miners think they're worth. (case in point: 1080 is better from an FPS/$ standpoint than the 1070 is right now, when is that ever the case?). and end-user rebate for $100? i don't know that that would bring the effective price down for a consumer to where amd wants it, but maybe the optics would be better. although, again, i don't know why the optics of the retailer rebate are so bad, anyway.
 

Elixer

Lifer
May 7, 2002
10,376
762
126
Entire AMD SW team developing drivers for Mac ... most likely.
Also Linux drivers changes seems to have slowed down to a trickle.

On one hand on their discussion with PCPer they said the a lot of the Linux Driver changes will follow the Windows changes. So yeah. It looks like Mac --> Mining --> Windows/Gaming --> Linux
Apple makes their own drivers. AMD just gives them a source dump most likely, along with all the tech manuals, and even access to the hardware guys.
I seriously doubt AMD has anyone actually writing drivers for Apple.
Another update on Vega Pricing.
wccftech...
Seriously, wccftech?
Scraping the bottom of the barrel to find one of the worst reliable rumor sites there is?
 
Status
Not open for further replies.
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |