Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Some Benchmarks Seem Totally Bogus!
- Thread starter Tesselator
- Start date
- Sort by reaction score
Anything is possible of course but one wouldn't ordinarily expect an OpenGL slash DirectX benchmark to be CPU dependent.
My guess would be that it's a lame-ass windows port and isn't doing almost anything right. Just a guess tho.
I would say that it is something fishy with your 570 (even though, yes the 8800GT was a beast for its age).
My 560Ti even get better scores than your 570 (56.5fps/1423 points on the same 1280-settings you used for the Heaven 4.0 benchmarks)
Try increase the settings to 1920*1200-ish and higher settings to see if the difference is larger then? (my 560Ti got 32 fps on 4xAA and medium quality, 1920*1200)
My 560Ti even get better scores than your 570 (56.5fps/1423 points on the same 1280-settings you used for the Heaven 4.0 benchmarks)
Try increase the settings to 1920*1200-ish and higher settings to see if the difference is larger then? (my 560Ti got 32 fps on 4xAA and medium quality, 1920*1200)
Could be you're right... I mean WTH is this (notice PCIe bus is erring somehow):
Attachments
Last edited:
I get the same error using a Windows Quadro 4000.I get the same error... I think it's due to running non-EFI cards.
Heaven Benchmark seems to be working fine for me though....
Hmm, these errors don't sound real good. I guess we need to flash the cards to get rid of them? Great, more trepidatious procedures to undertake.
I tried upping the specs and the score remains about the same. Between 1010 and 1070 no matter what I turn on or off. And the average frame rate stays the same too (between 40 and 50) - again no matter what I turn on and up to 1080p. I guess that's different than the 8800GT which crawls set like that but it's still kinda weird that MUCH lower settings wouldn't produce MUCH faster frame-rates. <shrug>
I tried upping the specs and the score remains about the same. Between 1010 and 1070 no matter what I turn on or off. And the average frame rate stays the same too (between 40 and 50) - again no matter what I turn on and up to 1080p. I guess that's different than the 8800GT which crawls set like that but it's still kinda weird that MUCH lower settings wouldn't produce MUCH faster frame-rates. <shrug>
Hmm, these errors don't sound real good. I guess we need to flash the cards to get rid of them? Great, more trepidatious procedures to undertake.
You'd need to create EFI32 ROM for it, because all existing are EFI64.
Regarding benchmark score: I'll test my 570 in 1,1 when I'll back home. I use it in 4,1 currently.
Other benchmarks seem to be appropriate (in comparison with the 8800GT) tho. The 8800GT gets about 5FPS in FurMark (right) and 0.1 FPS in GiMark (left).
Although interestingly, Furmark in full frame at 1080p (on the 570GTX) shuts the power off to the card about 20% of the way thru the benchmark. I hear a click, the display goes black, and the card turns completely off. Weird. I assume the PSU is being over-taxed and this is a safety feature. <shrug>
Although interestingly, Furmark in full frame at 1080p (on the 570GTX) shuts the power off to the card about 20% of the way thru the benchmark. I hear a click, the display goes black, and the card turns completely off. Weird. I assume the PSU is being over-taxed and this is a safety feature. <shrug>
Attachments
Hmm, I wonder how much of a pain that would be?
Really, wow! That would be awesome! Thanks man!
Thanks man! That a different (higher spec) system tho. I'm waiting to see what 666sheep comes up with.
If it's at all close to mine I'll just relax and accept it. If it's significantly higher I have to do some more testing or return the thing.
Thanks man! That a different (higher spec) system tho. I'm waiting to see what 666sheep comes up with.
If he doesn't get back soonish I'll throw the card in my 1,1 and run the test
Thanks man! That a different (higher spec) system tho. I'm waiting to see what 666sheep comes up with.
Here is the 1,1..
5345's
16GB
570
It looks right in line with your score
Attachments
hmm, yeah, very similar. It's looking more and more like Pastrychef was right in the first place.
In this specific test, yes. But this benchmark is too bogus for me. I mean a game engine which is this dependent on CPU is really too lame to be considered seriously as a graphic subsystem benchmarker. Maybe there's something to their name Unigine (Un + Engine), an un-game engine. LOL
Even when I benchmarked The Witcher 2 comparing the 8800GT to the 570GTX there was like a 6 to 10 times speed increase. In Unigine; almost nothing... just a few extra features could be enabled but no score or speed improvement. In fact actually the 8800GT benchmarks faster than the 570 in low and medium settings - heh, go figure.
40 FPS at 720p on a MacPro1,1 with low quality settings definitely suggests that you're CPU limited. You can confirm this by checking the value of "GPU Core Utilization" with the OpenGL Driver Monitor tool, which is part of the free Xcode tools package.
In this specific test, yes. But this benchmark is too bogus for me. I mean a game engine which is this dependent on CPU is really too lame to be considered seriously as a graphic subsystem benchmarker. Maybe there's something to their name Unigine (Un + Engine), an un-game engine. LOL
Even when I benchmarked The Witcher 2 comparing the 8800GT to the 570GTX there was like a 6 to 10 times speed increase. In Unigine; almost nothing... just a few extra features could be enabled but no score or speed improvement. In fact actually the 8800GT benchmarks faster than the 570 in low and medium settings - heh, go figure.
Most games are CPU dependent, thats all it's really testing. I'm not going to argue for or against it because in the end it's a bench just like Cinebench dependent on more than one thing. meh
GPU usage can hit 100% and draw the full 250 Watts in other tests. So profiling Unigine would be a bit of a wasted effort unless I'm misunderstanding something. I would only be showing how lame their code is really.
In some other games I can get well over 1,000 FPS in 720p on the GTX570 where the 8800GT tops out at 60 to 80 FPS.
----------
Tru dat...
Right, but every test/application is different, and will have bottlenecks in different areas. Checking the GPU utilization with Driver Monitor is a really quick and easy way to see if you're limited more on the CPU/software side of things versus the GPU side of things.
Yeah I guess. Unfortunately almost all the tools in my XCode monitoring package crash profusely. Not sure why but I've stopped using it for pretty much everything since upgrading to 10.7.5