“AMD makes a 450W capable card that can be overclocked without worries…”
“… while nVidia is not only late, but can’t make a decent enthusiast card!”
“My Führer, the GTX 590 isn’t all bad, the cooler is quieter!”
“Enthusiasts don’t give two shits about a quiet cooler!”
“nVidia ALWAYS lies about the power draw!”
“Load up furmark and I’m sure it will burn up!”
“When will Fermi ever work?”
More seriously, KitGuru has posted some details about GTX 590 weak VRM: Nvidia continue to come under fire for poor GTX590 design.
GTX 590 under heavy load – source
14 thoughts on “The truth about the GTX 590?”
“Darkest Of Days” – this describes current NVIDIA status. Hitler in that video says everything what I think about NVIDIA. Let’s say “well done” to AMD and “not nice try” to NVIDIA.
No.. Hitler?! Nvidia will be so bad?
I can see Fudzilla’s next article: Hitler now works for AMD 😀
“As ironically as nVidia want to make CPU’s and a company making CPU’s are beaten them in making GPU’s”
Ah! Kitguru, the same site that said: “The GTX 480 consume about 450-500W” because they aren´t capable of read correctly their EXTERNAL data source (other site with a power consumption test of the ENTIRE system). This crap gave them something to “educated” articles about this “problem”. xD
Masters of Copy & Paste playing here, but the original source of this information ISN´T kitguru, is Behardware or hardwarefr. Kitguru is the echo of various bad and amateur sources. The GTX 590 hasn´t 4 VRMs per core, it has 5 VRMs per core (10), and another 2 VRMs for the memory. The thermic image isn´t saying NOTHING. The gpus are around 80ºC (good report for the bios), and the section of the PCB with VRMs are around 110ºC, this is acceptable for this electronic components (125ºC or more). Take a thermal image of a HD4870 to saw that this VRM´s temperaturas are normal.
Again, amaeteurish press is camping in this site, thanks for this piece of shit. The problem isn´t about thermal problem in the VRMs, is another reason related with the electronic (bad set of components, ex.).
enwgido – yes. At default freq and voltage it runs at 110 deg C, so cooling is ok and temps are quite ok. Now you push freq 10% up, power draw is 10% up (at least, maybe slightly more), vrm produces at least 21% more heat which unfortunately causes it to rise temps even more. While it rises it’s temp Rdson rises too. VRM tries to compensate lower current by increasing duty cycle… and in that moment waste heat rises again blowing the worst element in VRM and if safeties aren’t working all others along the way. Sorry, VRM works already in the safety margin on GTX590 and anything, from ambient temp. through dust to mild overclock might result in damage. Damage of expensive piece of equipment.
Temps on HD6990 backplate reaches only 96 degrees – higher than 590GPUs but lower than VRM – and that’s with 450W limit not 375!
HD4850 and 70 is another history. Fan was set too low on reference boards and some time later fan fix was released addressing the issue. You can see 4890 if AMD learned it’s lesson. Or 4870X2, or 5970…now it’s time for nvidia’s fan fix 😉 And magically most issues with 590 will disappear but one will emerge – fan noise.
Why burning such a great GPU,
Enjoy its Potential
I know I am starting to sound like a brokken record, but Fermi generation was a lost generation a total failure. Nvidia has to change their philosophy and strategy big is better doesn’t work anymore!
i dont think so that fermi is a total failure
As nivia has made some nice GPU’s like 580, 570 and 560 which are truly good on games as well as in 3D softwares too.
I am a Special Effects Artist and i use Vray 2.0 and Iray which are GPU exhaustive renderes and pretty much satisfied with my 570 dual SLI.
even it gives good Physical Effects than compared to my friend’s HD 5870 AVIAVO tech
So 590 may be not up to the mark but rest of the fermi force is.
Dr you’re right that Nvidia has to change their strategy with the GPU making but they have made some good cards that are problems to ATI.
All their chips are far bigger than any of AMDs equivalent segment chips and are barely faster. That shows me that nVidia has definitely dropped the ball in terms of engineering. With an idiotic/overpride CEO on the top this company is going to sink in the not so distant future.
@DrBalthar (cylon colaborer :>)
From consumers point of view chips size ABSOLUTELY doesn’t matter. All that matters:
2. Power Draw
Temperature, size – doesn’t matter at all.
For GRAPHIC CARDS it doesn’t matter how fast they raytrace or calculate Milkyway projects. Those cards must be fast in games (or graphic/modeling apps). CUDA/OpenCL should go slightly separate ways. Slightly because OpenCL and CUDA usually doesn’t need ROPs and TMUs or Tesselators in their calculation tasks. Just bunch of cuda cores and cache. Gamers on the other hand might use cuda/opencl/directcompute partially but most time they need strong graphics performance, and not scientific.
“but Fermi generation was a lost generation a total failure”
1. yeah the GTX590 has issues (but it will be solved with non-reference designs by the vendors)
2. that card costs 600€! It is everything else than a mainstream product! The only market I see for it are developers, who run them in a 4-SLI to develop the render-engines for the next 3-generations (see that Unreal techdemo released a week before).
3. The other Fermi cards perform well, esp. the middle-range ones (GTX460, GTX560TI and maybe GTX470). Even the low-end ones perform equal as bad as the ATi ones (performance/costs). So the Fermi gen is everything else than a failure, it is a SINGLE reference card design that fails!
Analyzing the thermal photos we mast take in account that the GTX 590 has the VRMs exposed on the back of the card (right in front of the camera) without any cover while the 6990 has the VRMs behind the PCB that is also covered by a metal plate.
We don’t know 6990’s VRMs temp, it could be as much high as the GTX 590 ones
@Promilus: yeah you’re right, Thats why CUDA rocks
Comments are closed.