AMD Fury X Postmortem: What Went Wrong?

[tential]

LOL, I think I said that back in the 7950 days.

Probably did. He recommended me to buy the 7950 I have. Best decision I made. I would have gotten the 680 or 7970 and not gotten remotely the value out of it the way I did the 7950. I mean I think of starting my own blog or whatever but I mean that guy needs it. I swear a lot of my gpu knowledge is reading his posts over the years.

 

[xthetenth]

Why isn't Russian sensation a writer for this? That dude needs his own blog/review site seriously.

Honestly if he did his own benching, he'd be the best video card review site on the web.

 

[Jaydip]

Wrong. A lot of us were recommending GTX470/480 after after-market versions came out that fixed the noise and temperature issues. I even noted how later revisions of GTX480 used less power. Furthermore, I brought up after-market 480 cards and their overclocking many many times over HD5870.

Just because you are totally oblivious that this actually happened during the Fermi generation, doesn't mean GTX470/480 cards weren't recommended. The main recommendation was to skip reference versions of those cards.

Also, GTX480 used 2X the power of an HD5870 so the comparison between Fury X and 980Ti doesn't even make sense in that case.



That's not the point. 7970Ghz was like GTX680-> 770. Had AMD called it HD7980, what would you be your argument? All you are doing is just arguing semantics. HD7970Ghz beat 680 and HD7970 OC beat 680 OC at 1440P, had 50% more VRAM, and still ran cool and quiet. I don't remember you recommending HD7950/7970/7970Ghz cards that generation though.



No one cares about LN2 results since none of us runs their gaming rigs with LN2 24/7. Talk about making a point about nothing.



So wrong, it's amazing you are trying to pass this off as fact. AC Unity, Wolfenstein NWO, Dead Rising 3, Mortal Kombat X, Shadow of Mordor, Skyrim modded, HD7970Ghz crushes a 680 2GB. Ever bother checking HD7990 vs. 690? It's not close in modern games.

You honestly need me to waste my time proving how wrong your statement is that HD7970Ghz or HD7970Ghz CF cannot take advantage of 3GB of VRAM vs. 680 2GB / 680 2GB SLI (aka 690)?



Sure someone can buy a 980Ti or Fury X for 1080P but it would be largely a total waste of money for all but the few DSR junkies who use tiny 1080P monitors. Are you honestly suggesting 980Ti is actually worth the extra $ over 980 for 1080P? Not sure if serious. In your attempt to discredit Fury X as a 1080P card you failed to recognize that 980Ti OC is a giant failure of a card at $680 against a $470 980 1.5Ghz OC for this non-GPU demanding resolution.



LOL, 2 days after launch and you are already implying that Fury X may never get voltage control via MSI AB. You have short memory as many HD7950/7970 cards didn't have voltage control at launch and some of them needed to be hacked to enable voltage control. Once that happened HD7950/7970 became overclocking monsters.

BTW, in no way am I defending lack of voltage control. Chuck another instance of people putting words into my mouth. :sneaky:



Yawn, as tviceman or most other people on this forum will confirm, if I know AMD/NV GPU launches will be close, I always recommend waiting. So why would I recommend a $550 925mhz 7970 on launch when we all knew 680 was about to come out shortly? Once we knew all the performance at stock vs. stock and OC vs. OC between 7970 and 680, one could make a better informed decision. Also, I am pretty sure you bought a reference HD7970 didn't you? So ya, no offense bud but you messed up twice there by paying early adopter tax on 7970 and getting the POS reference blower 925mhz 7970. Anyone objective in my shoes would have told you to wait for after-market 7970 vs. after-market 680 showdown.

Of course your entire rebuttal doesn't address my point on how this forum viewed HD7970Ghz (that were only for sale in after-market form) against GTX680. You discussing 925mhz reference 7970 that you purchased had literally 0 to do with my post.



It is depending on the review you check. If some reviews use more GPU demanding games and more MSAA at 1440P, 980 gets destroyed.

23% faster at Sweclockers

19% at normal quality and 20% at high quality at Computerbase
http://www.computerbase.de/2015-06/amd-radeon-r9-fury-x-test/6/

While I really enjoy TPU's reviews, you gotta account for majorly CPU limited benches they have like WoW or gaming benchmarks that make 0 sense against all other sites like their Wolfenstein benchmark which is flat out wrong it seems.



HD7970Ghz was never available with a blower, only after-market versions, available on day 1 with 3 games bundle, many of which sold for $469-489 in July 2012. Regardless even when GTX780 was $500, people on here bashing Fury X still didn't recommend a $400 R9 290. Heck, it was cheaper to get a reference R9 290 and slap on an after-market cooler/AIO CLC than to buy a 780. In the last 9 months 780 is getting owned by a reference 290, nevermind an after-market one.

1. 480 used twice the power sure but it faster, not the same case with Fury X.You always talk about how NV fans skipped 5870/50 and 6970/50 cards, what was it about then?

2. Lol no, 7970 GHz was a oced 7970 so it is only fair to compare it against a oced 680.I rarely recommend people what to buy but if you are that curious sure you can check that.Look at the charts you yourself linked, 680 is still beating the 7970 and 770 is damn close to 280X.

3. You talk about about max oc vs max oc a lot, so here you go.

4. The point is 680 and 7970 is very slow to play new games @1440p with max settings now a days( as you yourself said 1080p doesn't matter). It is irrevalent how much more ram they got.

5. That was not your point, first you tried to show that [H] is biased by saying that Fury-X is a 980 competitor but after seeing the AT benches you are making a complete u-turn.

 

[B-Riz]

Honestly if he did his own benching, he'd be the best video card review site on the web.

Kickstarter? :hmm:

 

[B-Riz]

1. 480 used twice the power sure but it faster, not the same case with Fury X.You always talk about how NV fans skipped 5870/50 and 6970/50 cards, what was it about then?

2. Lol no, 7970 GHz was a oced 7970 so it is only fair to compare it against a oced 680.I rarely recommend people what to buy but if you are that curious sure you can check that.Look at the charts you yourself linked, 680 is still beating the 7970 and 770 is damn close to 280X.

3. You talk about about max oc vs max oc a lot, so here you go.

4. The point is 680 and 7970 is very slow to play new games @1440p with max settings now a days( as you yourself said 1080p doesn't matter). It is irrevalent how much more ram they got.

5. That was not your point, first you tried to show that [H] is biased by saying that Fury-X is a 980 competitor but after seeing the AT benches you are making a complete u-turn.

I want to meet ppl IRL that play *all* games at max settings on single cards 1440p and greater...

To argue #4

While I have good hardware (OC 4790K 4.6Ghz, Nepton 140XL cooler, Asus 290 DCU2, 16GB RAM, 480 ARC 100 SSD, 144Hz 1080p Asus) I am shooting for that 120+ FPS.

Even if I had a 1440p panel, I would still go for FPS.

If that means BF4 is on low / med quality, so be it...

 

[blastingcap]

Sigh, just read the Techreport review. Maybe frame lag is not the correct term? What do you call this?

And their conclusion is also damning. An excerpt:

Thanks for posting this. They and HardOCP report similar results but TR apparently goes one step further.

This is why experiential testing matters. HardOCP was one of the few voices that said anything about microstuttering in SLI/Crossfire because they actually played games through instead of running canned benchmarks. They also played multiplayer games instead of running in single-player mode; and in RPGs they try to play that instead of just camping out in a town.

With the advent of FCAT, we had better tools to quantify microstuttering. AMD could no longer avoid the issue. Their fix was kludgier than NV's, at least at first, but it did mostly solve the issue with DX11 games. TR is reporting that maybe some of those old bugbears have come back, though.

Anyway the point that I hope is not getting lost is: MINIMUM framerates matter, forget average. An experience full of major hitches (low minimums) is worse than an experience with few hitches, even if the former has a higher average framerate.

And frametimes matter for the same reason; you don't want huge variations between frames or for runt frames to count as frames per second.

The good news is that AMD can probably fix these issues in drivers, and that we're on the verge of Windows 10 which may also address these issues as well.

Says TR: http://techreport.com/review/28513/amd-radeon-r9-fury-x-graphics-card-reviewed/14

 

[Jaydip]

I want to meet ppl IRL that play *all* games at max settings on single cards 1440p and greater...

To argue #4

While I have good hardware (OC 4790K 4.6Ghz, Nepton 140XL cooler, Asus 290 DCU2, 16GB RAM, 480 ARC 100 SSD, 144Hz 1080p Asus) I am shooting for that 120+ FPS.

Even if I had a 1440p panel, I would still go for FPS.

If that means BF4 is on low / med quality, so be it...

1. That is the sole point of these cards, some people don't want to compromise on visual quality.

2. That is a valid option too, as I said there are many folks(I believe 1080P is still the most popular resolution) who want to play with the highest visual settings.

 

[tential]

I want to meet ppl IRL that play *all* games at max settings on single cards 1440p and greater...

To argue #4

While I have good hardware (OC 4790K 4.6Ghz, Nepton 140XL cooler, Asus 290 DCU2, 16GB RAM, 480 ARC 100 SSD, 144Hz 1080p Asus) I am shooting for that 120+ FPS.

Even if I had a 1440p panel, I would still go for FPS.

If that means BF4 is on low / med quality, so be it...

Ya I'm an eye candy gamer now but if I was gaming like how I used to then ya, it'd turn down the settings because a) kills fps, b) distracting you wanna turn settings off to make enemies easier to see c) it's all about winning anyway who needs eye candy?

But know that I'm eye candy and don't do twitch games, I only need 30+ fps and I'm OK.

The 680 wasn't a good buy looking back and I recognised that pretty quickly into my research. I didn't think the 980ti was a good buy when fury was still unknown. Now I can't recommend fury on any real metric. The only thing that can salvage the actual value for the card is driver improvements which with amd you know it's coming. It's just exactly like I said leading up to this launch, no one wants to pay 650 for a finished product. Amd needs the software to be ready for this launch. No surprise, it wasn't ready....

 

[TechyGeek]

Honestly if he did his own benching, he'd be the best video card review site on the web.

That +1

 

[krumme]

Thanks for posting this. They and HardOCP report similar results but TR apparently goes one step further.

This is why experiential testing matters. HardOCP was one of the few voices that said anything about microstuttering in SLI/Crossfire because they actually played games through instead of running canned benchmarks. They also played multiplayer games instead of running in single-player mode; and in RPGs they try to play that instead of just camping out in a town.

With the advent of FCAT, we had better tools to quantify microstuttering. AMD could no longer avoid the issue. Their fix was kludgier than NV's, at least at first, but it did mostly solve the issue with DX11 games. TR is reporting that maybe some of those old bugbears have come back, though.

Anyway the point that I hope is not getting lost is: MINIMUM framerates matter, forget average. An experience full of major hitches (low minimums) is worse than an experience with few hitches, even if the former has a higher average framerate.

And frametimes matter for the same reason; you don't want huge variations between frames or for runt frames to count as frames per second.

The good news is that AMD can probably fix these issues in drivers, and that we're on the verge of Windows 10 which may also address these issues as well.

Says TR: http://techreport.com/review/28513/amd-radeon-r9-fury-x-graphics-card-reviewed/14

Playing games thats the experience i get too. Average means zip vs the problems that arises from dips and frame time variance. Clearly amd have work to do here for fury but they have made it before as you say.

I remember when mantle hit bf4 and was working. Within seconds it was noticiable. A different gaming experience (i dont play bf today so dont know if they messed it up). And look at the smooth frametime variance graphs it got vs dx11.
Really hoping dx12 will come soon. Its a pleassure for all.

 

[YBS1]

Sure someone can buy a 980Ti or Fury X for 1080P but it would be largely a total waste of money for all but the few DSR junkies who use tiny 1080P monitors. Are you honestly suggesting 980Ti is actually worth the extra $ over 980 for 1080P? Not sure if serious. In your attempt to discredit Fury X as a 1080P card you failed to recognize that 980Ti OC is a giant failure of a card at $680 against a $470 980 1.5Ghz OC for this non-GPU demanding resolution.

Your problem is you think you understand the mentality of the people who actually buy $600-1000 dollar cards, sometimes multiples of them. I believe you think they are satisfied with a "playable" experience. When most likely they are buying them to get ridiculous frame rates at ultra settings, not having to compromise and turn things down, trying to push the minimums up to meet 120-144Hz and honestly sometimes just for benchmarking fun.

Also, you keep trying to rewrite history. By the time the 7970 was actually sorted out well enough to "destroy" the 680, most of those users who actually purchase these high dollar cards had already moved onto the 780 which again trumped the 7970 easily. I don't recall ever seeing my 680 Lightnings being beaten down by 7970's while I actually had them in any benchmark. That's AMD's biggest problem. Their hardware has honestly been pretty damn spectacular over the recent years, they just stumble so badly out of the gate they can't recover. I don't want to buy the cards that will be fastest a year from now, I want what will be fastest until (roughly) the next big thing.

 

[SolMiester]

LOL, I think I said that back in the 7950 days.

He already has a job!

 

[boozzer]

What's with the hilariously editorialized title? Who died again? Do we have to agree with the premise that something went wrong?

They certainly didn't hit it out of the park but performance that is I'd assume is indistinguishable from a 980ti outside of benchmarks isn't anything to scoff at.

one person props it up to the clouds before release, another slams it down after :biggrin::twisted::awe: you would think they are working together :sneaky::awe:

 

[5150Joker]

You know I feel sorry for AMD because they are in a tough spot. On one hand you have their "fans" saying stuff like, "this card should sell for $450-$550" and their rebrands should be practically given away vs NVIDIA premiums but on the other, they feign sadness at the thought of AMD going bankrupt and cry about GameWorks.

If you really want AMD to stick around, pay the premium they are asking instead of crying about the price/performance vs 980 Ti. Given the fact that AMD is cash strapped and they have a declining R&D, what choice do they have but to at least ask $650 for this thing? They might be forced to drop the price but in the long run, that will do them more harm than good.

On most boards I see a lot of AMD fans with old cards like 7970 who haven't spent a dime on AMD in years and they are part of the reason Fury X is viewed as a disaster because they also hyped it up like the next coming of Jesus and then at release declared it wasn't worth the money and will opt to wait. Meanwhile, NVIDIA fans put their money where their mouth is and buy the products they claim to love. This is why you see AMD failing, they have penny pinching cheerleaders who do them no good.

 

[JDG1980]

one person props it up to the clouds before release, another slams it down after :biggrin::twisted::awe: you would think they are working together :sneaky::awe:

Seriously? It does AMD no favors to treat all criticism as some sort of conspiracy to bring the company down.

For my part, I want to see a competitive AMD, but they haven't been doing a particularly good job of that recently. This is why my original post didn't just focus on what AMD was doing wrong, but on what they need to do to fix it. I would be very happy to see a competitive Arctic Islands release, preferably as early as possible, but AMD seems to drop the ball on releases much too often.

There were several major mistakes in the Fury X release that could and should have been avoided. The release price should have been $549 given its relative performance against GTX 980 Ti. Overvolting tools should have been available on day 1, and if this wasn't feasible, then AMD's representatives should have refrained from falsely representing Fury X as an "overclocking monster". Up-to-date WHQL drivers should have been available on day 1, and should have been provided to all reviewers. No one should be in the position of saying "wait and things will get better". That is NOT how you launch a premium product.

AMD needs to learn some old sayings. Like "you never get a second chance to make a first impression". And "better to underpromise and overdeliver than overpromise and underdeliver".

I honestly believe that there are several dozen people on this board who could do a better job running AMD's marketing than the people currently in there. Even without full control over the operational and R&D side, most of us would have avoided some of the worst blunders that have been seen in this release and the prior releases of Hawaii, Bulldozer, etc. AMD's marketing is the worst by a substantial margin of any company of its size.

 

[DarkKnightDude]

https://www.reddit.com/r/pcmasterra...y_x_possibly_reviewed_with_incorrect_drivers/

Apparently AMD provided the wrong drivers for the reviews?

 

[Silverforce11]

2. Lol no, 7970 GHz was a oced 7970 so it is only fair to compare it against a oced 680.I rarely recommend people what to buy but if you are that curious sure you can check that.Look at the charts you yourself linked, 680 is still beating the 7970 and 770 is damn close to 280X.

7970Ghz is actually faster than 280X (which is faster than 770, which again is faster than 680) if you didn't know your history.

Again the current problem with Fury X is poor value against custom 980Ti, it's not a flaw that is hardware related. If AMD drops the price, problem solved. Unfortunately it looks like they won't, cos its sold out everywhere.

 

[Subyman]

https://www.reddit.com/r/pcmasterra...y_x_possibly_reviewed_with_incorrect_drivers/

Apparently AMD provided the wrong drivers for the reviews?

I saw that, but it seems the drivers didn't make much difference according to other discussions.

 

[boozzer]

Seriously? It does AMD no favors to treat all criticism as some sort of conspiracy to bring the company down.

For my part, I want to see a competitive AMD, but they haven't been doing a particularly good job of that recently. This is why my original post didn't just focus on what AMD was doing wrong, but on what they need to do to fix it. I would be very happy to see a competitive Arctic Islands release, preferably as early as possible, but AMD seems to drop the ball on releases much too often.

There were several major mistakes in the Fury X release that could and should have been avoided. The release price should have been $549 given its relative performance against GTX 980 Ti. Overvolting tools should have been available on day 1, and if this wasn't feasible, then AMD's representatives should have refrained from falsely representing Fury X as an "overclocking monster". Up-to-date WHQL drivers should have been available on day 1, and should have been provided to all reviewers. No one should be in the position of saying "wait and things will get better". That is NOT how you launch a premium product.

AMD needs to learn some old sayings. Like "you never get a second chance to make a first impression". And "better to underpromise and overdeliver than overpromise and underdeliver".

I honestly believe that there are several dozen people on this board who could do a better job running AMD's marketing than the people currently in there. Even without full control over the operational and R&D side, most of us would have avoided some of the worst blunders that have been seen in this release and the prior releases of Hawaii, Bulldozer, etc. AMD's marketing is the worst by a substantial margin of any company of its size.

the very fact that you think I am on amd's side is what makes your entire post super funny :biggrin: and very telling :awe:

 

[RampantAndroid]

I saw that, but it seems the drivers didn't make much difference according to other discussions.

I'd be curious to see if a new driver is released that bolsters performance. Wouldn't AMD, if this is true, be asking reviewers to redo tests?

 

[IEC]

I saw that, but it seems the drivers didn't make much difference according to other discussions.

I think there's likely quite a bit of performance to be gained through better drivers, especially on Gameworks titles, but until I see that become a reality I really don't have a great incentive to upgrade over my current setup. Barely better than a 390X in some games isn't exactly making me reach for my wallet.

That could of course change with the release of aftermarket Fury cards (possibly better value) and better drivers (happened with 7950). I'm content to wait and see before I upgrade in the fall. If Fury X reaches parity with 980 Ti it would be a better deal. If not, I go back to nVidia for a generation

 

[RampantAndroid]

I think there's likely quite a bit of performance to be gained through better drivers, especially on Gameworks titles, but until I see that become a reality I really don't have a great incentive to upgrade over my current setup. Barely better than a 390X in some games isn't exactly making me reach for my wallet.

That could of course change with the release of aftermarket Fury cards (possibly better value) and better drivers (happened with 7950). I'm content to wait and see before I upgrade in the fall. If Fury X reaches parity with 980 Ti it would be a better deal. If not, I go back to nVidia for a generation

I plan to give things some time before buying (it's not like I can find 980 ti cards in stock anyway) - so maybe things will change by then...

 

[96Firebird]

Also, you keep trying to rewrite history.

This times a bajillion. And people eat it up...

 

[SolMiester]

7970Ghz is actually faster than 280X (which is faster than 770, which again is faster than 680) if you didn't know your history.

Again the current problem with Fury X is poor value against custom 980Ti, it's not a flaw that is hardware related. If AMD drops the price, problem solved. Unfortunately it looks like they won't, cos its sold out everywhere.

Actually, if you use relative 1080 perf the 770 is faster than 280x!

 

[erunion]

AMD doesn't have unlimited leeway in pricing of the Fury X.
Its got a built in radiator and using an interposer. These are cost adders. They can't cut its price simply because Nvidia performs better. Its price still needs to reflect its manufacturing cost.

AMD has tried pretty much every pricing strategy conceivable with your CPUs, from slashing the FX8xxx prices as low as possible to selling that same die for nearly $1000 as FX9xxxx. But no pricing strategy changes the fact that its a more expensive to manufacture but worse performing product.

AMD has that same dilemma with Fury X. They can chase margins or volume, but not both. While Nvidia has been growing ASP and gaining market share at the same time.