Sign in with
Sign up | Sign in
Your question

Are benchmarks biased for video cards?

Last response: in Graphics & Displays
Share
October 28, 2008 4:32:05 AM

hi i was wondering if all of the benchmarks everyone gives is biased. i ask this because are video cards dependent on cpu's? if so should i not expect the same scores from my video card if the benchmarks are using a qx9770? i only really address this issue in 3dmark06 and 3dmark vantage for the gpu. obviously they will have higher fps but will this affect the gpu alone? as well as their operating system and motherboards? im on average 20% lower on all scores and fps than the benchmarks just for my gpu. i dont have 64bit vista with sp1 or a ddr3 ram compatible mobo, but a really good mobo, ddr3 ram, and a better os wont increase performance/fps that much will it? im just tryin to figure what is going on before i rma my video card for being on average 20-40% lower than the benchmarked cards. i get great fps in hl2 games like tf2. but cod4 i only get about 30-60 when im outside and around 100 inside on max detail at 1680 detail. is this normal?
October 28, 2008 4:59:20 AM

*psychically attempts to ascertain your hardware configuration*
October 28, 2008 5:21:03 AM

its a q9450 stock at 2.66, ocz hpc reaper pc2-8500 at 1066, foxconn p45a-s motherboard, 10k raptor for hard drive, corsair 650tx psu, and of course evga gtx 260 core 216 superclocked..i just got a score of 13807 on 3dmark 06. thats seems pretty aweful if you ask me. SM 2.0 score of 5311, and SM 3.0 score of 6496. btw im using a stripped (hacked) version of windows vista 32 bit without sp1.
Related resources
October 28, 2008 5:57:36 AM

shortshift23 said:
hi i was wondering if all of the benchmarks everyone gives is biased. i ask this because are video cards dependent on cpu's? if so should i not expect the same scores from my video card if the benchmarks are using a qx9770? i only really address this issue in 3dmark06 and 3dmark vantage for the gpu. obviously they will have higher fps but will this affect the gpu alone? as well as their operating system and motherboards? im on average 20% lower on all scores and fps than the benchmarks just for my gpu. i dont have 64bit vista with sp1 or a ddr3 ram compatible mobo, but a really good mobo, ddr3 ram, and a better os wont increase performance/fps that much will it? im just tryin to figure what is going on before i rma my video card for being on average 20-40% lower than the benchmarked cards. i get great fps in hl2 games like tf2. but cod4 i only get about 30-60 when im outside and around 100 inside on max detail at 1680 detail. is this normal?


Yes, you will see different results in benchmarks than those published by someone using a different hardware configuration. Frequently video cards are benchmarked on very high end systems in order to minimize the likelihood that the video card will be bottlenecked by the rest of the system. If you are using a different OS, different cpu, and different memory you will likely see different scores. In your particular case, you are using a processor that is significantly slower than the QX9770 used in the benchmarks you are comparing to so it is possible that the processor is bottlenecking you somewhat. I suspect that the majority of the discrepancy is caused by you using a different OS however.
October 28, 2008 6:04:33 AM

you think switching to vista 64bit with sp1 would help a bit? and isnt a q9450 a pretty good cpu which wouldnt bottleneck a gtx260 core216? i see people with amd 6200x2+ have better scores then me on 3dmark06. i dont get it =(. could it also be my mobo? p45 doesnt have the best performance....but a mobo doesnt really add that much difference of performance...last time i heard
a c 130 U Graphics card
October 28, 2008 7:44:33 AM


You cant possably expect to get the same figures without exactly the same system and running the same clocks etc. Even then there will be some discrepancy. All of the things you listed can andn will make upto that amount of differance. Its the same old apples to oranges argument that comes around when people compare Nvidia and Ati cards, if its not the same hardware doing the same thing the same way then you just cant expect comparable results.

Mactronix
October 28, 2008 8:50:53 AM

Personally I believe it all depends on the hardware your running. Now I'm far from running top of the line equipment. Current specs for my pc- Athlon x2 5600+running stock 2.8 Ghz, Gigabyte m55sli-s4(nforce 4 sucks lol) 8800gt in sli config , 2 GB g.skill ddr2 6400 800Mhz memory, Hitachi deskstar 320GB, SB audigy and 700w psu in an antec 900 case. Well I decided to give the radeon 4850 a try on my pc to see how much faster it ran than a single 8800gt and in my particular instance the 8800gt beat the 4850 at stock settings in both 3dmark06 and vantage but in crysis demo, obivion and all my other games the 4850 was faster by far, except Hellgate where the 8800gt faired better. (stock settings single 4850 vs single 8800gt) Yea its nice to have high 3dmark scores for bragging rights but those benchmarks can vary widely and I dont believe are very accurate. Several things can affect them especially in vantage where physics can be run on nvidia cards resulting in higher benchmarks. Personally i dont take much stock in those benchies. Im aiming for more real world performance. Now Im not a fanboy and I dont favor nvidia over ati or vice versa(Not gonna lie, i go for what the best deal is at that time)
Also game performance can vary from one manufacturer to another. One game may run better on ati than on nvidia and another may perfer nvidia over ati. As far as your OS is concerned maybe thats you problem. Maybe that "hacked" version is buggy and i dont think it would hurt to get vista 64bit w/sp1. Also o.c. your cpu.(at this time theres not many games that take advantage of 4 cores) You should be able to get over 3Ghz easy. (lol wish I could say the same about mine)
Good luck and good gaming.
a b U Graphics card
October 28, 2008 10:41:50 AM

Why don't you go out and buy a complete version of Windows for starters and install it legally. Nothing irritates me off more than an idiot doing something illegal, trying to get something for nothing, then wondering why things don't work as good as they should.
October 28, 2008 2:38:45 PM

i guess im just gonna have to save up for vista then >.< and what temps do you think i should be hitting if i overclock to 3ghz?
a c 130 U Graphics card
October 28, 2008 4:51:22 PM


The short answer to the question is yes they are. Most sites (allegedly) lean one way or the other. Corporate reasons or just plain favouritism from the point of view of the reviewer. Dont forget someone has to supply the cards, now call me sceptical but i recon supplied free of cost cards have got to be worth an extra FPS or endorsement.

Mactronix
October 28, 2008 7:34:10 PM

i overclocked to 2.8ghz and i was getting around 54-58c under coredamage stress test(the hottest of the stress test) is that a good range? btw im using a zerotherm nirvana 120. will going to 3ghz make it alot hotter? and will this help my video card a little more if i overclock my cpu?
October 28, 2008 7:39:19 PM

shortshift23 said:
i overclocked to 2.8ghz and i was getting around 54-58c under coredamage stress test(the hottest of the stress test) is that a good range? btw im using a zerotherm nirvana 120. will going to 3ghz make it alot hotter? and will this help my video card a little more if i overclock my cpu?


As long as it's below 71C it's okay. So it's something else then. But that system definitely have problems. Here's 3dmark06 with q6600 and 8800gts, under Vista x64.
October 29, 2008 2:50:55 AM

so if im running around 13.8k with a system like mine you think it sounds slow? it could be a number of things. i have the rma number for my gpu so im gonna send it back and hopefully they find something i didnt cause i dont want to waste another 2 weeks of my time. and i am just going to buy vista 64 with sp1 seems like the logical thing to do. also could my motherboard a p45 chip hurt my performance alot? i heard alot of bad things towards performance lately. maybe its just me. btw at stock my cpu speed was 4100 at 2.6ghz. is that normal too?
October 29, 2008 5:18:25 AM

You should stop worrying much about 3d mark and just worry about real world results. I have a similar setup and my 3d marks scores are crap, only around 15k to 16k with sli and oced 9450, but my real world benchmarks are far superior.Crysis maxed out I get around 30 to 50fps at 1920x1200, Far cry 2 I get around 80 to 100fps, Cod4 I get around 200fps..
First you should purchase a legal copy of all the software you are using and then try real world benchmarks.
October 29, 2008 6:17:30 AM

lol mactronix=rip off of computronix?
a c 130 U Graphics card
October 29, 2008 7:00:19 AM


Never heard of it/them
Mactronix
October 29, 2008 11:34:43 AM

chef_jd said:
You should stop worrying much about 3d mark and just worry about real world results. I have a similar setup and my 3d marks scores are crap, only around 15k to 16k with sli and oced 9450, but my real world benchmarks are far superior.Crysis maxed out I get around 30 to 50fps at 1920x1200, Far cry 2 I get around 80 to 100fps, Cod4 I get around 200fps..
First you should purchase a legal copy of all the software you are using and then try real world benchmarks.


Usually systems that underperforms in 3dmark also underperforms in real games. But anyway, there are frontend benchmark utilities for real games. Try something like this. You have to have the corresponding game installed, of course.
http://downloads.guru3d.com/Crysis-Benchmark-Tool-1.05-...

!