Early review of Quadro 2000, The overheating issue

Very disturbing

http://www.3dchips.net/main/modules.php?name=Content&pa=showpage&pid=83

This attitude had however no influence on the graphic card, for in a Dauertestlauf under Maya with abundance Scene Anti Aliasing 4x in a solution of 1280X1024 picture points, the surprise came then: the card first ran for this extremely demanding Benchmark in 13 pictures per second (FPS). After short time the ventilator of the FX 2000 caught was choked to revolve at maximally, at the same time the achievement of the graphic card, delivered the Benchmark only 9 FPS. Subsequently the ventilator regulated itself again on moderated numbers of revolutions, the 9 FPS remained however admissible the measure of all things. First with a new start of the Benchmarks, again the 13 FPS are reached, to the renewed overheating, goes the game again by ahead loose. In spite of costly Belüftung, open housing or other player eggs, it is not arrive us to hold the GeForce FX on a durable achievement level.

Read that, or translate it if necessary. It says point blank that the Quadro 2000's cooling solution is uneffective. They start a test it gets 13FPS then it starts to overheat and the benchmarks drop to 9, then they shut it off and let it cool, then it gets 13 again.. then it overheats.. then they cool it down..

You get the idea. Just posting this becuase I am already seeing posts where people at Some specific sites are FALSLEY Reporting (like riva station) that

"look see??? I told you GFFX does not need that dustbuster"

It also begs the question of wether the GFFX ULtra is really running at a constant 500mhz or not. Multitexture tests posted at Anand, and Extremetech would indicate a clockspeed closer to 450mhz. At least by the time it gets to the fill rate tests.

3dMark Fillrate test multitexture

From extremetech.com

GFFX ULTRA 3557.8
9700 Pro 2564.1

From Anand

GFFX ULTRA 3477.9
Radeon 9700 2536.7

I'm surprised no web sites commented on these scores. The 9700 Pro is clocked at 325 mhz engine. That gives a theoretical maximum of 2600 in the 3D Mark multitexture fillrate test. The actual score is 2564, which is 98.6% of the maximum. Anandtecchs scores are nearly the same. Now look at the GeForce FX. Theoretical maximum at 500 mhz would be 4000, yet it achieves 3558 which is about 89% of the maximum. That seems very odd. Now, take 3558 and round it up to 3600 and now compute the clock speed: 450 mhz. Quite frankly, the multitexture test should be a very reliable way to calculate clock speed. Not that its indicating EXACTLY 450mhz.. but something less than 500 seems certain.

Interesting eh?
 
fxtemp.jpg


This attitude had however no influence on the diagram map, because came in a lasting run under Maya with Full Scene anti Aliasing 4x with a dissolution of 1280x1024 pixels then the surprise: First the map for this extremely fastidious bench mark ran with 13 pictures per second (FPS). After short time the exhaust of the FX caught 2000 actually maximally to turn, the achievement of the diagram map was throttled, to that with bench mark supplied only 9 FPS. Subsequently, the exhaust adjusted itself again on moderate numbers of revolutions, the 9 FPS remained however permitted the measure of all things. Only with a restart of the bench mark again the 13 FPS are reached, up to the renewed overheating, then the play goes again from the front loosely. Despite complex ventilation, open housing or other Spielereien we did not succeed in holding the GeForce FX on a durable leistungsniveau.

Nopefully nggalai can give a better translation ..from what I read...the card didn't overheat until they ran 1280 x 1024 with 4X AA even with the case open..

I'm more concerned about the tempeture reading there...that is smokin.
 
Holy crap! Are they sure that temperature reading is right? How the smeg didn't the chip fry, at those temps?
 
Read the Info at the bottom of the first page in red.

Basically it states that after a BIOS update, the problem is resolved...

UPDATE 28.01.03: Nach einem Biosupdate (Neu: Version 4.30.20.17.01) konnte das Problem behoben werden
 
221 Celsius!!!!! They should take that card apart. Sounds like someone forgot to put thermal paste on it, or its not making contact to the heatsink.
 
"Basically it states that after a BIOS update, the problem is resolved... "

So it wasn't actually that hot, or what? Did the FPS go back up?
 
So it wasn't actually that hot, or what? Did the FPS go back up?

Good question. Several answers...

If there was a problem, a heat issue that caused a decrease in clock speed, and if the problem is fixed then the RAM GPU does not heat up as much.

Or, the temperature was read incorrectly and the card was prematurely clocked down when in fact it wasn't exceeding specs, which is now fixed.

I'm sure there are others... ;)
 
Actually, now that I think a bit, it is unlikely the card was at a core temperature of 221 deg C. The "Grade" is set to 140 deg C in that screenshot, which would have immediately underclocked the card until the it was under the 140 deg C threshold.
 
um, who deleted my post? am I not entitled to an opinion?

Edited by JRR: I did. I also deleted someone's response to your post in the hopes of keeping this thread calm. Of course you're entitled to an opinion, but your "opinion" is just going to turn another thread into a flame war, which I'm sick of.
 
BoardBonobo said:
Well I guess they're OK until it hits 1414 Celsius. That's when the silicon would melt 8)

I think your leads or bond wires would delaminate before then. Or if you live in a humid environment, your package might pop from the trapped water vaporizing.
 
Dopant migration occurs at much lower temperatures than that. I would have said that something around 120 degrees was the maximum sustained temperature that the silicon itself will stand, but obviously I'm a bit out there.

This is a very interesting white paper Intel did, several things on there relevant to heat as well as a particular problem in chip design...
http://www.intel.com/technology/itj/q12001/pdf/art_4.pdf
 
THe_KELRaTH said:
There's no doubt that this card is going to be "THE CARD" recommended for use in artic expeditions!
The polar icecaps aren't melting fast enough for you? ;)
 
Um, JR, you mean my opnion of 3dfx? Common man let it rest, that Company is long gone. I really don't see how you can have a flame war invovling a company that no longer exsists.

What I said is a perfectly resonable hypothesis. This really is the first problemtic card I've seen from Nvidia and there IS a striking resemblence from the voodoo 5500 days and the problems it had through development. I've even seen someone else make remarks like this and you didn't edit that post. I didn't even consider that until I read someone else mention it.

There's nothing to argue about really, either you agree or don't agree. It's just a guess, so I'm not going to argue about it.
 
Quincy, you are making a tenuous link at best - the is very little to no similarity to the actual technology used from 3dfx.

Its quite clear by now that the majority of the issues stem from the fact that .13u wasn't ready in the form that NVIDIA required, thus the power/heat requirements have strayed from their initial specification. Being aggressive in in the process technology is far from a 3dfx trait, as you've said yourself many times, and actually follows the same path that NVIDIA (sans 3dfx) have traditionally taken. Its just the case that in this instance their planning hasn't panned out as they would have liked, and their competitor (evidently heading the advice of the FAB) made a very competetive part on a more mature process.

Its possible that the choice to opt for .13u was made even before 3dfx engineers had any input at all.
 
Back
Top