Discussion in 'Reviews' started by W1zzard, Mar 22, 2011.
and you are still in denial that 6x 6990`s in reviewers hands have either failed or are failing....
I don't think so. Deaths from one side does not justify the cards failing from the other camp. Shouldn't happen at all, as far as I am concerned, and AMD's cards dying has NOTHING to do with the 590GTX deaths, other than these GPUs are made by the same company, on the same process. It's the PCBs that have failed, not the GPUs.
this is the first `real` foray passed the pci-sig limits , and its obvious those limits are there for a good reason - whilst the gpu`s themselves might be up for it , they are failing at the weakest point
and any cards failing from either camp is not good at all.
What I don't understand is why, when these cards are really only wired to provide 375w(slot plus 2x8-pin), people expect them to exceed that power draw. nVidia states 365w for GTX590, so to expect much overclock, if any, seems a bit silly.
However, it is very concerning that these cards can even go past that, nevermind things like OCP and OVP, in my opinion, should be hardware-based, and not software, nor should it be able to be affected by drivers, and in that alone, I consider these cards a failure.
I appreciate the reviewers killing these cards though...that leaves clear lines drawn in the sand as to what people can expect, no matter how far over specification these cards were run. It's been a long time since we've seen VRM failures, yet DR-MOS tech on VGAs is relatively new...and quite cost-effective. I do not beleive cost-effective solutions should be used on "flagship" cards, either, so I'm left looking at nVidia for an explanation.
i know wizzard said about another power plug - but thats way outside pci-sig spec , since the spec only allows for 2 power plugs
either amd and nv should make the cards complient or not - pretending to do it isnt working.
What about the molex adapters on some boards (most notably any quad SLI boards). I believe that is what he was talking about...
the slot can deliver up to 150W I believe, but sits only at 75W...?
i like it better when they are honest and outright not put issues with certain cards in highlight and cover others up.
the same sites that killed the 590 some with over 20% added voltage,have added 0%(besides the switch)in the 6990 in there reviews thats not odd?
Then we have to read a comparo on the two cards from a reviewer and expect him to be non biased?
w/e i think its best to know of facts before someone(joe public) goes off and says the power circuts are too weak for stock clocks/overclocking @1.06v as alot are doing without having a clue..
Harlequin check this out(online google translator should be good enough):
and then this:
making it ~400W compliant(PCIe + 2x8pin) does not translate in to making it utter sh*t.
LoL. Way i see it, GTX590 came with votlage boost guidelines for review, so reviewers were expected to increase volts. Meanwhile, maybe 6990 came with recommendation to not adjust volts?
I mean, you can insinuate all you want, doesn't change the fact that very often reviewers are given specific requests as to what they cover in a review. Many sites covered GTX590, and most adjusted voltage. Many sites covered 6990, and very few adjusted voltage.
So, no, to me, it is not odd, at all.
Well why did they bother to adjust volltage at all?
you did not have to,its also a better guage to what most people will get in overclocking and not have an unobtainable result from an end user.?
now if say a reviewer did get a crasy clock out of the card,would he have said how much voltage he used? or just the max 1.06(in the two bios i'v read) or just the headline "1000 on just stock volts"?
i do agree this is not the time and place for the lack of reporting on the 6990 cards dying,its just hard for me to look at certain card reviews anymore.
Reak of bias has no room in a honest review site
where did you hear that bs that they had to use volage with the 590??because thats what it is.
I didn't say they HAD to use voltage. I did not review this card, so I am not privy to any info given to reviewers. I am just jumping to conclusions as you are.
However, OC testing is standard. 6990 features AUSUM OC bios, no need to adjust volts, IMHO, as PCIe spec is broken by AUSUM bios.
GTX590 offers voltage adjustment, and specifically, ASUS card have "Voltage Tweak" stuff. Seems pretty basic to me. This feature was explored, and that testing lead to card death. End story.
You seem to imply that W1zz is biased in his reviews, while personally, i think he is critical of all things, and has no brand loyalty.
Look at it this way...both cards are high-level gpus that are "downclocked" to meet PCIe spec. 6990 reaches spec of 2x6970, no problem, via AUSUM bios. GTX 590 is the same, however, does not always seem to meet the spec of 2x GTX580, and does not have dual bios. But testing to each spec, the spec of whatever cards were squished into one, did not require any adjustments other than flipping bios on 6990, but on GTX590, voltage increases were needed...I don't see any issues with how that was done, and why 6990 didn't get the same voltage boost.
if you think i'm biased, then ask yourself why i dont do 3 tests, say "great card" and be done with my testing so nvidia is happy, users are happy. problem solved.
no i love wizz i was talking about in genaral.
anyways I wanted to see how fast i could kill one of these pos's at stock volt overclocks lol
i cant even find one?
no wizz read above
asmof i like your reviews the best you run o3 o5 and 06
Agreed, the particular problem seems to be actually with the SmartDoctor software, not the card. It seems like ASUS has made a rather costly mistake in allowing their software to ignore the limit set in the BIOS, and that is what caused the death of this card.
Now this is a review for the ASUS card, so scoring it baddly because the ASUS software provided with it allowed the voltage to be adjusted way beyond limits seems fair to me. However, I would have like to see a little more differientiation about the actual reference card. Perhaps even a seperate review write up on the card minus the shiney ASUS SmartDoctor software as a general GTX590 review.
well no i thuoght that to only asus but.. i have found one reviewer(tbreak) that killed it with afterbuner yes he did go 1.21v on the wrong sent driver was a zotec card.
i did look at a retail zotec cards bios it was limited to 1.06v.
wont put up a link thats kinda tacky thing to do in 590 review of anothers site
some c/p action of what he said
this guy manned up about it.
sorta what i did with my made for oc'in 460 hawk he got "noobcaked greedy"
That’s a good amount of scaling with a 77MHz overclock. Now given that the GTX 590 has the ability for voltage tweaking, I got really brave and decided to over-voltage the card as well as the speeds and see how far I can go. Using MSI Afterburner I increased the voltage by 75mV and bumping up the Core speeds to 753Mhz which resulted in an incredible score of 9816 on 3DMark 11 (p). Hoping to hit the 10k mark, I decided to bump the voltage to 125mV with Core speeds of 804Mhz. As 3DMark 11 was coming to a close, the whole system shutdown and I could see smoke coming out of the power cable connectors.
As the smell of burnt plastic and metal faded, I contacted Nvidia to try and see what happened. The GTX 590 should have been able to either handle the voltage increase, or simply failed to a BSOD or crashed the system. I certainly didn’t expect it to die on me. Turns out the drivers provided by Zotac were 267.52 and the latest drivers from Nvidia, 267.71, have some fail-safe measures to protect the GTX 590 against over-voltage. So remember folks, a simple driver update can be the deciding factor between the life and death of a graphics card when you’re pushing it to the limits. Always update your drivers.
althuogh he might have had a abx for use.
I run mu 5870 at 1.35 volts, water cooled core, and yes, a cooler for the phases. But even with just paltry aluminum sinks on the phases it would run that voltage OK.
Nvidia made a card, then made it dead by providing a enthusiast card to enthusiasts and not the standard epeen nvidiot. They did the equal of saying, here buy our twin turbocar that can go amazingly fast, and we built for racing.
Oh, and if you ever race it, it will blow up, aand your warranty will be void.
Supercars are notorious for having parts break when driven hard so I think you need a different analogy.
I think both camps have been rushing their products out to market equally. Nvidia has just shipped what should be considered more of a beta as the final product. It was a bad move but unless they feel it financially they could care less TBH. The same goes for AMD and any company that answers to stock holders rather than the end user. I think they need new upper management at Nvidia and perhaps they will be more competitive on the high end without compromising overclockability.
no, why wizz didn't adjust hd 6990 voltage was because there are no program that support new voltage controller in hd 6990,
Pretty sure he could have whipped one up if needed.
Oh wait...it works now, doesn't it.
why would ppl compare apples to bananas, different hardware takes different voltages, AMD Cpu's carry's much higher voltage configuration than Intel's and overclocks worse.
Same story here, generally, Nv's high end overclocks waaay better than AMD's, overvolt'd or not.
AMD pushed their gpu/cpu clocks way too high (concluded from their smaller headroom for overclocking) which will result a higher failing rate than their competitors(theoretically).
Not to mention those HD6990's died without any tweaks, while 590's only died on excessive overclocks(and faulty drivers, with the new drivers, the card's OCP works just fine, but NV shouldn't let this problem happen at the first place)
My 4870X2 died within like what, 2 weeks, and it came with Arctic Xtreme somthing cooler, so heat was never a problem. I'm not based on my opinion just because one card died on me, but AMD's dual cards do have some reliability problems, I thought they were getting better with their 5970, but this one's probably as bad as the 4870X2, or even worse.
Also if NV makes their cards as long as the 6990, the extra heat sink NV could fit in the huge spaces here and there will definitely lower the cards' overall heat output and make their cards more quiet.
My unlocked 6950 @ 1030MHz and my previous 5850 @ 1080MHz disagree. Try that with a GTX 570 and it will blow up.
Do you really think you understand how to clock hardware better than the company's own engineers? Think about that for a second and how ridiculous it sounds. Also, building false logic on incorrect statements just leads to a bunch of nonsense - AMD's cards, despite being clocked higher, are still much more power efficient than NVIDIA's and don't require near the circuitry or the stress. If anyone's trying to overclock their hardware to (barely) compete, it's NVIDIA. That'd explain why their GTX 570's failed and why the GTX 590's are failing now. With that track record, nevermind the 8800's from back in the day, I wouldn't bet a nickel on NVIDIA's quality control.
There are multiple reports from several different review sites of GTX 590's dying at stock. The only documented 6990 death I've seen is the one from Neoseeker.
Again, after the 8800's, the GTX295's (I had three die on me), GTX 570 reports, and now the GTX 590, I have no faith in NVIDIA's quality control.
Maybe if they made the card longer they might actually have a card that works. As it is, they had to release a driver to downclock the cards just so they don't blow at stock. I'd like to see reviewers retest with the newest driver to see how much a performance hit there is when the card actually runs safely.
I'm going to go out on a limb here.
The 590 isn't an enthusiast card for overclockers. It's and enthusiast card for really rich folk or people with fairly reasonable disposable incomes. It doesn't come with two BIOS and it doesn't have specific points for voltage measurements like the MSI Lightning range or the Gigabyte SOC card.
It is in fact just a really fast and relatively quiet card. It isn't designed for overvolting to the extent other cards may be.
Just because it's expensive, doesn't make it enthusiast. An enthusiast will buy a product known for it's 'character', not based on it's price. The argument this is enthusiast because of it's price is misleading on all fronts. In other words, having cash does not an enthusiast make!
Let's face it, if you want to get maximum graphics scores you'll probably go for 4 way GTX 580 on LN2, not two of these (likewise of you're an AMD OC'er you'd use 4 way 6970's).. Dual graphics cards are nice little speed devils that don't ask to be tampered with.
We seem to have forgotten the whole 6990 AMD switch warranty (i know XFX will honour that warranty now). Can that voltage be adjusted? Will it melt the card if it's bypassed? WHO CARES? It's not designed for it.
Price doesn't exemplify enthusiast. It's an expensive well performing (but not best) dual gpu. The 6990 pretty much takes the crown in neutral vendor games but is way noisier and needs a better cooler.
Give me a 6970 or a 580 lightning over either dual card and i'd be much happier.
Agreed with that, but I think this card can be interesting for overclockers and the people wanting to set records with single cards.
I'd like to ask all the people criticising the 590: who really wouldn't want one? Yeah, it's pretty nice, regardless.
Separate names with a comma.