The 82A issue and so on seems to suggest an issue with OCP kicking in
The 82A issue and so on seems to suggest an issue with OCP kicking in
There is something strange, ie possibly unknown, still going on though since certain reference cards are not having problems, even some oc'ed cards.Quote:
Originally Posted by zerazax
nope it's configured via a resistor, no software way to control or disable ovpQuote:
Originally Posted by Tetedeiench
Well, there are still a few problems :Quote:
Originally Posted by SNiiPE_DoGG
- Take an old game, play it with your brand new card. Not alot of geometry. Simple shaders. Wow, black screen ? That could happen. It's worth a try. Any idea of a game or bench ? I'd say 3dMark2003 or 06... or any game of that era that uses simple shaders. Can't think of one at the moment.
- GPGPU applications. They may reach this value also. They don't have to wait for geometry there. ATI will have to lower the values also. Even if i don't think they'll reach this value, my limited experience in the field cannot answer that question for sure.
- You buy a card that boast it can do X, yet when you have it, it can't. I'd say it's a problem.
Now, right now, what ARE the implications of this problem ?
- Limited overclocking margin. You'll reach it quickly.
- You cannot run any 3d app you want on your card. The list is not that huge, i'd agree, it's not that bad. But morally, ethicaly, this comes into question. Especially since this list can grow in the future.
I have checked the logs, AMD downloaded OCCT yesterday. I guess they must not be taing the problem lightly, or at least, they're taking a look at it ;)
It's true it is not a bug that will make everyone ditch their card, or make ATI/AMD recall their card, IMHO. But still, it's a flaw, a quirk in their design, and worth notice.
Heh just go for GTX 285 for $300 dell deal. Thats what Im thinking of doing after I play around with this 4870x2 and sell it.Quote:
Originally Posted by Bo_Fox
I understand the concern many may have, but as shaders get more complex the problem will be less and less relevant, no? - I have played games like star wars jedi academy @ 1600x1200 with all the settings on high no problem (a great game) its old and has very simple geometry but my card has no problem with it as it barely loads the GPU to make it run great....
I'm not so sure GPGPU apps would do the same.Quote:
Originally Posted by Tetedeiench
Sure it will stress the shaders but not every part of the chip to the max, i.e. TMUs/ROPs.
The overclocking margin is only limited if you are soley going by OCCT. Most people are going to base their overclocks on their favorite games and benches, since that is where the extra performance might be needed.
What other 3d app can you not run based on the card failing OCCT?
Thanks for the precision :) So it's hard-wired... wow.Quote:
Originally Posted by W1zzard
I think one of the senior members here should assist Tetedeiench in contacting Macci. In this way, someone close to the community and who works for AMD can help colaborate with these findings.
Just me 2 cents..................
AMD clearly already knows about the situation...Quote:
Originally Posted by mongoled
one question i keep asking and have no response on is, even though this is a potential problem, how does the performance compare to previous ATI cards, and the nvidia competitor cards. to generate that much heat and power, id expect this thing to be demolishing all others.
i think in the future they will go less with shaders, and more in other places, since these things are never being used to their max, they can probably better optimize their cores.
Largon concurred with the results I was seeing from other forum members.Quote:
Originally Posted by Manicdan
Underclocked 4890(850/850, ~83FPS @ 1680x1050 on setting 3, is quite a bit faster than even a GTX285, ~53FPS @ 1680x1050 on setting 3.
Don't know how much weight the CPU has on the score but the Nvidia card has the CPU advantage as well...
i don't see the issue here, it's a single app out of thousands and the first one able to show this "problem"
i think amd did this on purpose to prevent the card form burning in case someone loads the card with so simple shaders that each ALU is fully uitilized which is impossible under normal circumstandes
i never EVER saw my card pulling more than 70A :confused:
Oh yeah, the volterra ones.. but why such a limitation on the volterra vrms? I think the volterra ones are more expensive only because they offer more of a digital support (tweaking via RT/bios, etc..) and the 55nm NV cards do just fine with analog VRMs. 55nm GT200 cards have a higher reliability rate than 65nm cards that are more prone to failure, and a GTX285 can eat just as much power as a GTX280. I remember reading somewhere recently that the analog VRM's are actually better than the digital ones.Quote:
Originally Posted by BababooeyHTJ
W1zzard, have you yet done a Vmod on disabling the OVP for 4870/4890 cards? Anybody? I'm curious if those reference cards can handle 100A or so without any problems. It just gives some peace of mind, since there are so many older games out there--literally thousands of them--that we never know which one could actually push the card to the same black-screen crashing scenario (of which we would probably just discard as "buggy" piece of software when it's really the hardware). Also, this issue is interesting, nonetheless.
Strange how RT reports it as 40A when GPU-Z reports it as 80A... which one do you think is true?
Someone should grab a fireblanket and run this on a HD2900XT! That card has some craaazy vDDC current capabilities (6× VT1195SF).
=P
edit:
That's what I was wondering too.Quote:
Originally Posted by Bo_Fox
That 80A figure everyone is seeing does not make sense.
Quote:
Originally Posted by largon
I run a myriad of new and old games. No game comes close to stressing the card as much as this OCCT.Quote:
Originally Posted by Tetedeiench
I run Folding@Home on my 4870 all the time that I'm not playing games, ever since I got the card pretty much. GPU-Z reports only 45-ish amps or so, for 96-98% GPU load.Quote:
- GPGPU applications. They may reach this value also. They don't have to wait for geometry there. ATI will have to lower the values also. Even if i don't think they'll reach this value, my limited experience in the field cannot answer that question for sure.
I don't know about that - so far, everything the card boasts it can do, it has done :shrug:Quote:
- You buy a card that boast it can do X, yet when you have it, it can't. I'd say it's a problem.
I'm maxed out on my overclock that CCC will allow for. But we all know that overclocking is a gamble, so even if you couldn't OC the card more than 30MHz well you still got what you paid for.Quote:
Now, right now, what ARE the implications of this problem ?
- Limited overclocking margin. You'll reach it quickly.
- You cannot run any 3d app you want on your card. The list is not that huge, i'd agree, it's not that bad. But morally, ethicaly, this comes into question. Especially since this list can grow in the future.
Any 3D app does run. Only OCCT stress test doesn't, but OK...
And if it is a real issue then I'm sure they'll take a look at it. Seems they already are if they downloaded OCCT, which is a good thing.Quote:
I have checked the logs, AMD downloaded OCCT yesterday. I guess they must not be taing the problem lightly, or at least, they're taking a look at it ;)
It's true it is not a bug that will make everyone ditch their card, or make ATI/AMD recall their card, IMHO. But still, it's a flaw, a quirk in their design, and worth notice.
So whats the final verdict here, bad hardware or bad software????
I wouldn't call it "bad" just something to take note of. I'm thinking of it like the redline of a car's engine, sure you can run up to that, but you can't go over it. Keep it at redline too long you could break it. For whatever reason, OCCT is trying to push it over the redline, and that's a no go.
Doesn't really bother me too much.
Reassembles quite the fuzz about furmark back then, there where people that blew there VRMs with Furmark, yet there have to be a cases where this happens with games or GPGPU apps.Quote:
Originally Posted by SparkyJJO
Quote:
Originally Posted by LordEC911
My GTX280 pulls just short of 80A with the same settings with my core at 648mhz. This is much more stressful than furmark on my card. I was seeing under 60fps, btw. With a Q9650 at 4ghz.
I'm not sure what point you are trying to prove.
I wasn't trying to prove a point, I was simply answering a question.Quote:
Originally Posted by BababooeyHTJ
I can see where he might have been headed by asking that question though.
4890 is what ~10-15% behind a GTX285 with both at stock on average in "normal" games and apps?
Yet with OCCT, the 4890 is ~56% faster, using the 83FPS vs 53FPS.
However this app is programmed it stresses every part of the chip to the max, or at least quite a bit more than other "normal" apps/games.
Also none of the numbers, i.e. FPP, seem to add up.
4890@850mhz= 1.36Tflops
GTX285@1476mhz= 1.06Tflops(MADD+MUL), .708Tflops(MADD)
1.36/1.06= 1.28x greater (1/2 the FPS difference)
1.36/.708= 1.92x greater (amusing since it doesn't mean anything but = largons power draw increase)
Simply using max theorectical FPP is not an accurate way to estimate performance but in this case it seems to be related. Since this app has been said to use simple shaders to completely load the ALUs, you could come to the conclusion that the MUL is only being used ~45% of the time.
Basically, the way this app is programmed it is able to use the 4890's architecture to the max and seems to not fully load Nvidia cards, persay.
Edit- Anyone know what the stock volts for a GTX280 is under load? 1.3-1.4v?
Nvidia may have already put a safety in to prevent reaching peak AmperageQuote:
Originally Posted by BababooeyHTJ
Instead of shutting down it restricts the the frames per second
Which would make the GPU stress test meaningless
I think it's 1.19V under load at stock volts, 1.11V at idle.Quote:
Originally Posted by LordEC911
Link Here
Good point but AMD should have put safety as well ;)Quote:
Originally Posted by AMDDeathstar
Well, that will be useful to tell to my friend who had to RMA 3 Wolfdales before he realized that he was damaging them with Orthos Prime. Granted, he was overclocking them, (and then using OP to test for errors...), the test would not have caused enough stress under stock settings. In any case, he can keep his 4.2 ghz overclock under normal operation (games, movies, etc..), while after an hour of OP, it will start BSODing until underclocked to 4.1. And so on.Quote:
Originally Posted by Shintai
Damage was not caused by OrthosQuote:
Originally Posted by iddqd
Damage was caused by overvolting, lack of proper cooling and pebkac :rofl:
I've run 52 hour prime95 25.6 64bit with my q9650 at 4.42Ghz using 1-32v-1.42v multiple times and its still doing great !
I'll second that, orthos doesn't kill chips - voltage kills chips.Quote:
Originally Posted by Greg83
Well they do have a safety. Nvidia's is limit fps to keep amps down, ATI's is an OCP. Both protect the card, just different methods :shrug:Quote:
Originally Posted by kiwi
If you were an enthusiast, you'd certainly call it a good thing. Install a turbocharger, and you're trying to push it. Install NO2, and you're doing the same. When you race around the track, you're always keeping it at redline. When you go drag-racing, you want to push it as far as you can even at the risk of blowing the engine or tranny.Quote:
Originally Posted by SparkyJJO
Guess what? This guy here just gave us a new way of testing engines and trannies to see if they can be pushed without breaking. And we can find out which engines are capable of handling 10000rpm instead, and which ones have a built-in limiter of 8000rpm, so you do not have to waste your time buying the engines and racing them yourselves. All you gotta do is look up the online database, thanks to this program!
That is what being an Xtreme enthusiast is all about, and for that reason, I will not tolerate pansies or threadcrappers here.
Edit: If you guys still want to debate this issue, then it's just the same as saying that LN2 suicide runs are pointless, just like drag-racing is pointless. A mod would probably ban a troller if he didnt stop voicing out against the purpose of enthusiast thrill.
When they capped furmark, look what shill nVidia sites like Expreview wrote.Quote:
Originally Posted by kiwi
Agreed, I ran my Wolfdale C0 steping at 70c+ in IBT for some pretty long runs(hours) and hours of prime at 1.37v under load (more when I picked up a TRUE) and never had a hint of voltage degredation. My q9650 has seen the same sort of abuse and again not a hint of voltage degredation. Whenever I hear stuff like that I think of BenchZowner's VTT thread.Quote:
Originally Posted by Greg83
Guys were talking about a chip with 800 shader cores and it's performance (4870) isn't too far off and in some cases better than a GTX280 in games. I'm not surprised that it stresses a 4870/4890 more and gets such a higher framerate. I doubt that Nvidia is limiting framerates in this benchmark. Trust me it stresses my card hard.
Chill :rolleyes:Quote:
Originally Posted by Bo_Fox
Your comparison isn't quite valid. When you install a turbo, or NO2, etc, you don't do it on a bone stock bottom end. You have to upgrade the bottom end to handle the increased output.
And I am hardly trolling. Seriously dude, just because someone doesn't meet your definition of "enthusiast" doesn't mean they are threadcrapping or trolling or anything like that.
LOL, all right.. but I think mine was more valid than yours. :D:p:Quote:
Originally Posted by SparkyJJO
LOL @ more car analogies. They never end.
I have never seen where ATI has boasted or made claims as to there abilities of running stress tests, they made no such promises to anyone. The only claims I have seen are what the cards will do with games and so far they live up to it.
Dont get me wrong, I look at this as a problem now myself. I just dont feel that ATI has failed to deliver on what they sold me.
Oh, and BTW Sparky. Been running a 150hp shot on bone stock bottom end on 88 mustang for years now ;)
Ahh, made mine sound better, thank you Goldie! By the way, it's dirt-cheap to replace the exhaust.Quote:
Originally Posted by G0ldBr1ck
Let me correct you there, voltage doesn't kill chips, current kills chips through electro-migration.Quote:
Originally Posted by halo112358
If you run silicon cool enough you can pile on the voltage until you start getting too much leakage through the gate oxide for stable operation. Because the resistance is proportional to temperature, current is lower and the rate of electro-migration is reduced.
Even if you have an ATi GPU that can pass this test you may be drastically reducing it's lifespan by running it. It is possible that the non-reference cards that pass also contain higher binned GPUs which have a higher current threshold or have the limiting mechanism disabled. Maybe someone with a 25x16 or 2160p screen could find the limit if it's there.
I've tested the OCCT GPU stress test and agree that the R770 is limited to ~80A as indicated by the upper limit in GPU-Z. With my clocks (775/1100) the test will not run at 1280x1024 on complexity 3, I get a grey screen with a white square where the mouse was left (CCC 9.4).
I am going to stick with furmark as my GPU stress test for two reasons. One is that this test causes an over-current condition that is potentially damaging to the GPU the second is that given the difference in FPS at the same settings on different architechtures with otherwise equivalent performance it is not a fair test as it clearly stresses ATi GPUs far more than nVidia GPUs. However, it may also mean that the quotes FLOPs performance figures are not achievable due to this apparent limitation. I also don't like that crossfire does not kick in it windowed mode and that windowed mode is limited to 1280x1024.
That's not to say it is not a worthwhile finding as it may explain why an overclocked ATi GPU is stable in some games bust crashes in others and armed with this knowledge ATi may be able to introduce, through driver updates, an over-current mechanism that does not require a reboot.
<mutter mutter>Quote:
Originally Posted by G0ldBr1ck
All right but you get my drift! :p:
You have to remember the OP has said multiple times that this program stresses different GPUs differently with the same settings. IE. Setting 3 may be the hardest on ATIs, while nVidias might get loaded down the most with Setting 5
Id realy like to hear what AMD has to say about this, they have been asked many times and even banned my account when I asked in there public forums! If there was nothing to it, they wouldnt try to hide it so hard. Come on AMD, I dont think I was screwed on my cards but I would like to know what is going on and why!!!!
I run 3 3870's and all 3 pass this test, as well as the 3 together in crossfire. Temps get high! Up to 55C with EK full cover blocks. they NEVER break 40C while gaming. The VRM's get crazy hot, I measured 90C using an IR gun.
I don't know if that matters to anyone but thought it was worth a mention. I use OCCT alot for benching. The gpu test and the power supply test are my favorites, the power supply test in particular really pushes your entire system to its limits.
In an attempt to contribute something more than speculation and admonishment to this thread I thought I'd try to dig up some OCP mods for the 4870/4890 in case some brave soul wanted to test the 82A theory.
The problem therin is that XS is usually my first choice for Vmod info and since largon, initialised and other worthies have posted here and haven't linked to the relevent guide... but WTF eh? Search is said to be my friend.
Well, I searched and wouldnt you know I ended up back here at XS in the Xtreme Graphx Vmods forum, but no love on the OCP fix.
Although I failed in my quest it seems others are looking for this as well.
Quote:
Originally Posted by SebfromSaxony
hmmm this sounds familiar(same thread).Quote:
Originally Posted by SebfromSaxony
There were many other similar posts.Quote:
Originally Posted by Søndergård
It would seem that this problem may well manifest itself in games on overvolted/overclocked cards. Not merely while testing OCCT.
Hey, I downloaded the OCCT 3.1.0 app and ran the GPU test on my HIS 4870 1GB (the one with the reference cooler). Is it supposed to black as soon as possible? Mine didn't crash. Oh and I know you said that it has nothing to do with heat but I do have a Thermalright HR-03 GT on it w/ a Noctua fan.
It does. Check if your PCB is the reference one (it does not have to do with temperature, but with VRM). Check also if you forced V-sync in the driver, or Antialiasing, or Anistropic, or anything like that.Quote:
Originally Posted by P4rD0nM3
Also make sure you followed the configuration stated in the first post (shader complexity 3, fullscreen, high resolution, no error check).
OK guys, some news, as we had alot happening yesterday :
Two french websites had the problem happening in their testing labs. One of them has a huge reputation, and cannot be questioned about it. He actually came to the same conclusions as i did : we did hit a limit in the design of the boards. Phew, i'm not a complete moron ;)
He tried a non-reference design board : passed. Reference board : doesn't pass. He reproduced my test : very same conclusion. My test is not at stake, the cards are. Truly, he IS trustable. I'd say he is a demi-god of hardware in France. He is from x86-secret and memtest86.org . Just so you know who he is.
Here is the link to the first news :
http://www.canardpc.com/news-36049-g...et_4890__.html
His conclusion is interesting. If i translate it quickly (i'll let you use google to get an idea) : he says that it is unacceptable that a card cannot handle any code that can be produced with an API it is supposed to support (OCCT fits into that category, as it is pure DirectX9). AMD's answer, according to him, is almost already known : limit OCCT's execution speed, just as furmark. But is this acceptable ?
The interesting part is this : according to him, this defect is sure to have been detected in AMD's quality check process, but have been ignored on purpose. And he goes on by saying that the OCP mecanism is badly implemented, and that this is a true problem on those cards... unless the VRMs aren't at stake here, which is his supposition in the news, but the OCP protection mecanism (which i tend to believe myself).
And the 2nd one, which was reproduced in Yahoo News :
http://www.pcinpact.com/actu/news_multi/51011.htm
I'm actually waiting for AMD's official response now. I'm truly wondering what's going on.
I had an PM on my board waiting for me from AMD, asking me for contact information. I answered yesterday, still waiting for input from them. I'll keep you guys updated.
I can say that right now, we're going out of the "hypothesis" to the "there's a problem for sure", we just don't know if the problem is the VRM themselves, or the OCP, but we do know that the problem are the cards, that the problem occur :
- With OCCT GPU at stock speed
- With Furmark with a slight bump in vGPU/Frequencies (as it is a tad less efficient than OCCT)
And we're starting to get professionnal view on that.
OK, the sources are unknown for you, it's non anandtech, xbitlabs, or alike...
We're starting to get alot of cards touched, alot of testing done, and some professionnal reviews that prove our testing.
I updated the first post to include the first "professionnal" review we had. I'm sure people will question them because they don't know them. Well, i can't do a thing about that :(
I hope they are completely independent. Otherwise AMD will just tell them to shut up :) We have already seen their attitude in forums. Sorry, but this time AMD you have failed.
OCCT 3.1.0 fails on my reference Diamond 4870 that I bought one month after they were released. I just ran Furmark after changing the furmark.exe filename (to disable any driver tweaks) and it ran perfectly fine. I also run games going back the last 11 years! I've had no problems with any software I've run on my GPU, except OCCT. I'm not saying OCCT is flawed, I'm just saying there haven't been any issues on my end while overclocked. Extreme overclocking is where folks will be having their problems so that doesn't apply to me (yet). The next time I buy an ATI GPU, I'll just be sure to get a non-reference design though.
Any idea which manufacturers make the strongest, most robust non-ref boards so far? How about you make a list of the ones that work the best and overclock the highest while remaining stable?
Until we have a real world code(be it GPGPU related or game related) that fails on reference Radeon cards my opinion will be that this test is unrealistic and borderline power virus.There is not one real world workload that can cause the same effect on reference Radeon HD4xxx.
One more thing:whoever runs this test on their systems with Radeon HD4xxx risks either immediate or long term damage to their previously working Radeon cards.The VRM of reference cards is pushed beyond its specifications so no wonder the test fails(and the card could ultimately fail too).
We're not saying "this will happen in every game". We're saying "the design is flawed". We're saying "there's a possibility this will happen in another program". Why is that ? Because it happened in a legitimate program (OCCT is just a a DirectX9 scene, i remind you), so it can happen again.
When the pentium bug appeared for Intel (remember : http://en.wikipedia.org/wiki/Pentium_FDIV_bug ), it appeared very rarely, and in very specific applications, and in none that were available publicly. Yet Intel did recall the CPUs Am i saying we do have the exact same thing ? No, the P4 bug wasn't stress related. Was the problem described by the one who discovered the Pentium bug a power virus ? No, of course not. Am i saying AMD should do the very same thing, a huge recall ? That's up to them, i am not knowlegdeable enough to be able to know if they should, or not. I'd say no, personally.
I am not fiddling with frequencies, timings, or anything of the sort. I'm linking with DirectX9 libraries, happily loading a donut mesh, applying the shaders, and seeing the results.
So no, OCCT is not a power virus of any sort. You choose to install it, you choose the settings, you choose when you start it, you stop it, and how to configure it. So no, it is not a power virus. A stressfull stability test, yes, indeed. But a power virus ? No. Let's stop this 18-page old assertion that's adding nothing to the current problem please.
I do wonder how come a legitimate DirectX9 scene could push a VRM beyond its specifications on a card, at stock frequencies. They should have been stronger, more robust, don't you think ? They should be able to handle everything. They are supposed be compatible with DirectX... Yet, here's a completly valid DirectX app they can't handle... Interesting, isn't it ?
Instead of posting the whole code, would you consider some psuedocode of the alrogirthm used in the stress test?
What would be the point ?Quote:
Originally Posted by Nullack
So that people would copy my test ? ;)
No sorry, i'd prefer not to.
that's exactly what will happen. a new driver release fixes the issue. the rendering output is unchanged, so isnt that an acceptable solution?Quote:
Originally Posted by Tetedeiench
it's basically a software based OCP that results in exactly the same what the canardpc author suggests "the regulator should warn the pilot that the limits of the components are about to be reached, resulting in a secure and automatic underclocking."
uh what evidence does he have for such an accusation? just wild guesswork? maybe he called fudo and asked for advice how to write a story.Quote:
according to him, this defect is sure to have been detected in AMD's quality check process, but have been ignored on purpose. And he goes on by saying that the OCP mecanism is badly implemented, and that this is a true problem on those cards...
the ocp mechanism implementation is perfectly fine, it's just the ocp limit that is set too low.
i doubt the problem lies in 3 phase vs. 4 phase but ocp set too low vs. ocp set not too low. no data or evidence in the whole article, next please
It could happen again, but it seems unlikely. Programs doing a real workload aren't going to only be using such simple shaders and doing nothing else. Even the next closest stress program (furmark) doesn't come close - and the most stressful game or GPGPU app are even further from the limit then that.Quote:
Originally Posted by Tetedeiench
It is possible that a program could reach the limit doing ordinary operations, but it's hard to imagine what it'd be doing (besides stress testing).
It would be a good move from the PR side of things but a bad move from the business side of things, IMO.Quote:
When the pentium bug appeared for Intel (remember : http://en.wikipedia.org/wiki/Pentium_FDIV_bug ), it appeared very rarely, and in very specific applications, and in none that were available publicly. Yet Intel did recall the CPUs Am i saying we do have the exact same thing ? No, the P4 bug wasn't stress related. Was the problem described by the one who discovered the Pentium bug a power virus ? No, of course not. Am i saying AMD should do the very same thing, a huge recall ? That's up to them, i am not knowlegdeable enough to be able to know if they should, or not. I'd say no, personally.
It's not like the chips themselves have a problem. It's just the card implementation details that are a problem. For the 99% of people that don't use OCCT, the reference board is fine. For the rest of us there are 4 phase boards and easy workarounds for 3 phase cards.
The real problem isn't deciding if a recall is a good idea, it's how to address this issue without it becoming a PR disaster. There are people out there who will use this as a chance to smear AMD, no matter how they handle it.
But wouldn't the 4 phase cards probably have a higher OCP limit as well?Quote:
Originally Posted by W1zzard
physical limit before the card explodes? probably, yes.Quote:
Originally Posted by Solus Corvus
but the ocp limit is an artificial limit designed to avoid exactly that. it is chosen by the person who designs the power circuitry of the card. i am sure that there is plenty of headroom left in the 3 phase design
I'm still confused at how people are missing the fact that some reference cards are not failing...
I fail to see that as a valid mecanism. It is not a valid OCP. Why ? Because of two things. One which is funny, one less :Quote:
Originally Posted by W1zzard
- Let's imagine that in a future i want to turn OCCT GPU:3D test into a benchmark. That is NOT planned, but let's say i want to (after all, this code is pretty efficient, isn't it ? would be interesting to get some scores out of it). As ATI slowed down some cards because of an engineering problem, i wouldn't be able to do so. That would defeat the purpose of the benchmark. That's the very same problem as 3dmark's optimization, but backwards
- Imagine that in Final Fantasy XXV, PC DX9 version, you encounter the Great Furry Donut god, which is supposed to reveal your future as a hero. Poof, black screen. Yes, i'm joking by giving a silly example, but only midly. They ensure the sole app out there right now that could trigger the bug at stock value is castrated. They would have to redo it for every other app out there... to me, that doesn't sound like an OCP that's viable.
I already used the trick "run as another exe" to defeat limitations in the catalysts (i guess i learned from the furmark story already). I have already decided that if i ever see any limit put in the drivers on OCCT GPU:3D, i'll be working on releasing, very quickly, another version that doesn't have this limit. Just for impartiality purposes, and because i fail to see the point, ethically speaking, of any restraint put in an app due to a engineering bug. Because the card will still boast its capabilities... which it cannot use fully under all conditions. And that is, to me, a problem.
As for the article, i am not the one who wrote it, but yes, they're assumptions.
Yes it's unlikely it'll happen again, just as the Pentium bug ;) I already stated that.Quote:
Originally Posted by Solus Corvus
I don't think a recall is a good idea, personally. I fail to see it a good solution. But people had to know that the design had a problem, that my app wasn't responsible for the black screen bug, and that the overclocking capabilities, which is a marketing argument, is limited on those cards.
I really see the recall being highly unlikely.
That is due to the fact that, for whatever reason, those cards are not drawing greater than 82A. There are a multitude of possible reasons for this outlined throughout the thread (not running the app with correct settings, having AA/AF forced in control panel, etc.) but the bottom line is that under the correct circumstances these cards can draw more than 82A and that all evidence indicates that once this happens the system will lock up instantly.Quote:
Originally Posted by LordEC911
I guess the cards without bugs fell into the following categories :Quote:
Originally Posted by LordEC911
- A card that was not a reference design
- A setting that was forced in the driver that lowered the load, and thus made the test unable to reach the limit (Anisotropic 16x, FSAA, etc)
- Bad configuration of the test
it is exactly that, if you purely look at the definition. "overcurrent protection". your app brings the card into an overcurrent state. the driver does not allow that. -> it implements overcurrent protection.Quote:
Originally Posted by Tetedeiench
final fantasy devs will notice that issue, talk to ati devrel, they will tell them to change their rendering method/shaders, problem fixed. if the dev isnt willing to do that ati will just add app/scene detection to fix it.
good luck playing cat and mouse with ati.
what do you expect ati to do? recall all cards? send all 4870 owners a 4890 ? leave the bug unfixed ? there are not really many options. they could give you a chunk of money to recode your application :)
another possible idea could be tolerances in components employed. it's not uncommon to see 5% for resistors.Quote:
Originally Posted by Tetedeiench
That makes sense. But I imagine that the designer of the 4 phase board would feel comfortable picking a higher limit without running into the artificial safety margin.Quote:
Originally Posted by W1zzard
It wouldn't be a very good benchmark though. It would only be testing how fast any card could process that very specific simple shader. It would be able to test raw shader power but not much else (which is why the R770 whips the G200 in your test).Quote:
Originally Posted by Tetedeiench
I guess you could make a whole series of tests with each one running a different simple shader. It would make a good benchmark, though extremely synthetic.
No other scene geometry, AA/AF, vsync, or user interface? Seems a bit unrealistic and easy to workaround by just turning up the settings (force AF or whatever).Quote:
- Imagine that in Final Fantasy XXV, PC DX9 version, you encounter the Great Furry Donut god, which is supposed to reveal your future as a hero. Poof, black screen. Yes, i'm joking by giving a silly example, but only midly. They ensure the sole app out there right now that could trigger the bug at stock value is castrated. They would have to redo it for every other app out there... to me, that doesn't sound like an OCP that's viable.
I personally wouldn't expect ATI to recall cards, but make newer revisions with a higher OCP or a more robust VRM. Judging by how hot the VRM was getting on some of these cards, I'd question the capability of it.
Well, the goal of my app is to put the card into the highest load possible, just as furmark. That's the problem. I worked hard to get to this point. Should i get castrated without reacting ? I don't know.Quote:
Originally Posted by W1zzard
Playing cat and mouse doesn't appeal me at all. I have to say i don't know what to do.
If anything they are going to limit the load in the driver while running OCCT. It's not a big deal, the end user can rename the exe if they want the full effect.
Then maybe you should ask Nvidia why the MUL isn't 100% efficient or maybe why your code doesn't fully load Nvidia cards?Quote:
Originally Posted by Tetedeiench
Anyone try running a renamed .exe on Nvidia cards?
Tetedeiench here are some questions for you.
- What would AMD have to do, to satisfy you? Admit to you privately there is a problem? Recall every card affected? Issue a press release that they have sold millions of GPU's with a "fatal" flaw?
- Why does your test not load down the Nvidia cards with the same severity as the ATI hardware? Have you made any effort to recode your application to address this?
I don't know about him, but I'd be happy with: don't make the same mistake on the R800 reference board. :p:Quote:
Originally Posted by eleeter
Please shelve the ignorantaccusations. I thought we'd settled this already.Quote:
Originally Posted by informal
If teh 82A cuttoff is appropriate to the card design then I agree that running OCCT 3.1.0 GPU test may risk damaging the 4870/4890 series cards. However that being the case, AMD is in for a bit of fun seeing as one could easily break VGA cards without invalidating warranty. Furthermore if 82A is a realistic limit to ensure continued function what does this portend for non-reference designs that allow greater current.Quote:
Originally Posted by informal
You can't have it both ways. Someone has screwed up a card design here. Either AMD/ATI or their AIB partners. Time will tell.
What is really the issue at hand is that some 487/4890 must be downclocked to pass this test.
What AMD should have done was set a lower clock speed on the reference design. Thus any problems experienced while overclocking would be the fault of he who tweaked rather than the marketing department.
Then again if these cards were clocked 100-150MHz lower how would they stack up to their NVidia competitors at stock frequencies?
Downclock the card and reduce performance across the board just to handle a corner case? Horrible idea.Quote:
Originally Posted by Nightstar
Yep, to statisfy the .1% using a program that came out after release...Quote:
Originally Posted by Solus Corvus
Much better to misrepresent the capabilities of your product and trigger the PR nightmare that is beginning right here and now eh?
Tetedeiench, please superimpose a maze over your GPU test with a little mobile furry donut to navigate said maze. Then we can call it a game and be done with the illogical dismissal of this problem.
Uh, it would have been much better to just have a slightly higher OCP limit. AMD isn't misrepresenting anything - these cards run games and gpgpu apps just fine.
as i said, its only orthos that does it. he can run prime all day without any degradation. doing like ~4.2ghz @ 1.4-ish vcore with one of those 120mm heatpipe towers.Quote:
Originally Posted by Greg83
(there is also no degradation during "normal use" that some people over here are sneering at)
on the other hand, he is dumb for running it overclocked as high as possible for 24/7 use. Doesn't even need it - games are all about the gpu.
A higher OCP limit may have been appropriate for these designs. But I'm not an EE and I don't work for AMD designing video cards do you?
Your dismissal of OCCT as a valid application is illogical. I asked Tetedeiench to make his test the background for a simple game in order to demonstrate this(I don't really expect him to comply).
What is misrepresented is that this card is able to function in a stable manner at the specified frequency.
If you are reading these forums indeed 1%< you. If you are posting here your part of an even smaller group. I didn't wan't to be the guy to say it but this is "Xtreme systems", not "Good enough systems" or "Mildly defective systems".
In my opinion any component that is running at factory clocks should be able to deal with anything that you throw at it, if you overclock it and it fails then OK you are overspec and that's what you get but at factory clocks you should be able to throw anything at it.
It's not because you don't have issues in games that it's ok, a component has to be able to run at 100% and not fail. If it fails at 100% load then there is an issue.
OCCT is not running over the red line, it is running exactly at the red line and at original specs, underclocking the card fixes the issue which proves that the crashing cards at stock speed are not 100% stable, since they cannot cope with a 100% load.
It is not acceptable if my CPU runs at 100% load at stock speed and the CPU crashes, why would it be acceptable for a GPU? why do some of the reference cards work fine and others not? this again proves that some cards are good and others faulty, it would be another story again if all crashed.
ATI pushed the limits of the design, found a setting that was preventing most of the cards from burning out and yet ran all games fine in most situations, but yet many cards crash at 100% load. They have made a compromise and they know it, that's why they won't comment about it.
I don't think that the manufacturers using stronger power supply circuits and non reference designs are doing it for the fun, they know why they do it and why they spend more money on it. How much money does one of these components cost? 1$ maybe? imagine ATI saving hundreds of thousand of dollars just because they use one component less and it still works in 99.99% of the situations.
I'm not saying Nvidia is better and maybe the software does not load Nvidia cards as much as ATI, yet it's not excuse for ATI playing with design limitations to save $$$.
W1zz doesn't work for AMD designing cards either but I am willing to take his word for it that there is probably enough overhead for a higher OCP limit. And even if he is wrong, there are lots of cards out there to choose from with a beefier power section.Quote:
Originally Posted by Nightstar
I did no such thing.Quote:
Your dismissal of OCCT as a valid application is illogical.
Feel free to dig up some product literature that says that.Quote:
What is misrepresented is that this card is able to function in a stable manner at the specified frequency.
The 99% represents normal users. I said "the rest of us" in describing extreme users, clearly including myself in the group of OCCT users and overclockers (ie. the 1%). How was that not clear from my posts?Quote:
If you are reading these forums indeed 1%< you. If you are posting here your part of an even smaller group. I didn't wan't to be the guy to say it but this is "Xtreme systems", not "Good enough systems" or "Mildly defective systems".
I'll look into it. The temps are skyrocketing as well as the VRM usage. I do wonder why the FPS is below, but that's about it.Quote:
Originally Posted by LordEC911
I'll try to look at it next week. I promise you it'll be done ASAP. But i probably won't have enough time before the end of the next week, looking at my schedule.
I do wonder if this is not caused by an optimization on a function AMD did, and Nvidia didn't do. I'll do some testing.
Havent been XS in a while and so missed thread. My reference hd 4890 crashes when heavily overclocked. Only OCCT used to do this and I suspected it was due to the 85A load, something nothing else could. Since no game even come anywhere close to this figure, I don't care.
How the nVidia and ATI architectures handle complex/transcendental functions is a bit different. You could try some shaders that do square roots, exponentials, logs, sine/cosine, etc.Quote:
Originally Posted by Tetedeiench
That's what i'm doing, don't worry. I'm looking at the combination i found, and seeing if i can optimize it for both architectures :DQuote:
Originally Posted by Solus Corvus
If it comes from the shader that is. And if that part of the shader is at stake, and the more i look at it, the more i doubt it. i'll test it and let you guys know.
I have an Asus 4870 with Glaciator heatsink. Non-reference design, no problems with OCP but... it has 1,20V GPU stock and it generates errors on absolutely stock settings.
Assuming this is a valid stability testing program - not good at all :p:.
http://img29.imageshack.us/img29/7366/4870stock.jpg
do that and i doubt you will get anywhere near that loadQuote:
Originally Posted by Nightstar
Time to build higher quality pwm circuits, were are moving into 1GHz frequencies and yet have such cheap designs, resulting in ringing or unacceptable temperatures.
i dont know, does anyone design a car that can go 1000 miles on a highway in 1st gear ? technically that would be valid driving too
i'm a car noob, so i might be wrong.
I did some tests:
Used latest dx9 updates set catalyst 9.5 to defaults, set fan to 90% to prevent overheating set gpu and mem to 750 /900 default clocks(temps never got above 60 degrees in any tests performed ).
Used Furmark(OpenGL) : renamed it to something else ,set rez 1920 X1200 set full screen and ran benchmarking and stability test and ran without a problem no artifacts no crashes no vpu restarts and vddc curent max was 80 A using gpu-z 0.3.3
So i failed at making it go above 83 A but let's see if it matters because no games ever make it reach that limit.
Next i used ATT (DX9): this lil app has a very small 3d-render test and in less than 3 min it made my card crash(computer restarted) !!! i was watching gpu-z and vddc curent never got above 57 A so that seemed weird?
I then used OCCT(DX9): gpu test in window mode on low rez 1024 x758 shader complexity default 0 and it crashed(computer restarted) even faster then ATT and there is no way it got to 83 A using those settings.
Why does it matter that att and occt never reached 83A and still crashed ?
Because my card Gainward 4870 512 reference design is unstable in games like crysis unreal 3 mass effect dx 9 (winxp 32) even on stock clocks and games never reach that high vddc .
After reading http://www.geeks3d.com/?p=3246 it made me think that maybe the catalyst drivers are buggy or the last dx9 update isn't getting along with the catalyst drivers(ati have been focusing on dx10 optimizations) or contains some shader bug that afects ati cards in some way ,that would explain why some reference cards don't crash (they don't have the latest dx9 update).
Any other explanation as to why only directx stuff crashes?
ps: i am not a 3dguru so don't get mad if i did something wrong
Thanks for your tests ;) However, this bug can be reproduced with furmark if you bump the vGPU a little, as furmark is a tad less efficient than my test. It bumps the A on the VRM (and temps too), and you have the very same crashes.Quote:
Originally Posted by bluedevil
I'm working on the fps on the Nvidia architecture at the moment. i'll keep you updated on my findings.
I would love a new feature in OCCT. :)
A graphics test that generated a 50% Graphics workload, and a 50% PhysX workload, at the same time.
But that's another thread...
http://www.xtremesystems.org/forums/...44#post3806644
I could't resit having the programmers ear, that made such a handy App. :up:
Carry on with your discussion. (Derail end.)
I've spent about 4 hours on the FPS "problem", and cannot find the cause. Everything i tried that seem to augment the FPS is simply simplifying the shader code once compiled, and thus augmenting the efficiency on both architecture once used. Truly.
I'd think the difference between the ATI FPS score and the Nvidia Score is due to the way they implemented the Shaders. The latter are still under heavy load, mind you. I asked users to do some testing on their Nvidia cards, and here are the results. Overclocked, they could get a GTX280 to 114A :
http://img199.imageshack.us/img199/4052/aroxx.jpg
I'm trying to get a screenshot @ stock values, which is what you are going to ask next. I do wonder if the latest value is @stock. I asked the user to do it.
Unfortunatly, the card i have is a GTX285, without VRMa readings :(
Mind you, i'm not abandoning that fast the matter, i'll still investigate. But the test is already really stressfull on those cards. That's why i'm starting to think the FPS value is "normal".
tet, install nvidia perfhud. very useful for debugging such things
Hmmm... so this test also shows that you are not receiving 100% the performance of an Nvidia chip.Quote:
Originally Posted by Tetedeiench
Why would you purchase a product that you cannot fully utilize?
See, same argument can go both ways...
Please, everybody understood you want to defend ATI/AMD by any means, every arguments you want to find, so let's try to quit the fanboyism for a bit...Quote:
Originally Posted by LordEC911
As the architecture are different, the apps and test will give different results, and, globally, you'll have a ranking. But every test taken one by one will give you a different ranking, or performance rating for the cards.
I'm waiting for the RivaTuner VRMA screenshot @ stock.
More, i said i didn't give up on the issue. I'm following wizz advice right now, as i did hit the limit of what i could do manually. I'm still working on it.
Tetedeiench, I've loved OCCT since version 1 but there is not point in declaring this a design flaw. Every single game on the market runs completely fine on the ATI 4800 series, I've got my ATI 4850 @ 715/2150. (I'm debating between 720/2180 and 715/2150)
Does it fail the test? Yes, in less than 5 minutes. Does it run all my games for 6 hrs straight every single day? Yes. It's stable. What you need to do is put an approximate 90% load on the card, and monitor it.
Not really. It's probably loading the NV cards 100% also. It's just that the ATI cards whip the nVidia cards when it comes to processing this shader code.Quote:
Originally Posted by LordEC911
Tetedeiench, I'm curious about the nature of the instructions you are sending these cards. Is it mostly really simple instructions like FP/INT multiplies, adds, etc. Or are you using the complex functions more like sin, cos, log, etc. Or is it a mixture of simple and complex functions? It is interesting that there is such a gap between how well the two architectures handle this code.
And you seem to be in Nvidia's pocket...Quote:
Originally Posted by Tetedeiench
Get off your high horse and take a look at the numbers, they don't make sense.
You agree they don't make sense and then dismiss there is a problem.
As I stated and showed, it is not loading Nvidia cards 100%, Tet even agreed to this.Quote:
Originally Posted by Solus Corvus
Also why are you asking him about his code? He obviously has something to hide and doesn't want any to know what it is.
He agreed that the NV cards aren't as fast, not that they aren't being loaded 100%.Quote:
Originally Posted by LordEC911
I'm asking him about his code because I am genuinely curious about how the differences in the architectures play out in real code. Instead of grabbing for straws to find any gotcha that I can use to insult Tetedeiench and dismiss the entire issue.
Clearly you are no use to this thread looking at your last 5-6 posts. Please stay out of it. Thanks.Quote:
Originally Posted by LordEC911
Quote:
Originally Posted by W1zzard
Well, Based on Car terminology. It would be like this. In ATi's case.
They designed a very competing and good car. But little does the consumer know that it has a 5 speed transmission. But the 4th gear is missing. This helps cut production costs. Casual people don't know this 4th gear is missing. Only Enthusiast level people know.
:rofl::rofl::rofl::rofl:
Car analogies are the worst, lol.
Exactly :up:Quote:
Originally Posted by nfm
They just keep getting worse and making less and less since :confused:Quote:
Originally Posted by Solus Corvus