Originally Posted by snoball
This thing is really starting to P*** me off. I recently ran some benchmarks comparing my results to those from Tech Power Up and Hardware Canucks. My results are nowhere near where they should be...
I ran these setting and got an average of 91 FPS.
Ran these settings and got roughly 55 average.
Once again, ~30 FPS average.
One this I scored 19303 marks... About 2000 below the expected...
This is the case for most benchmarks I run. The card does not preform as advertised and I can not imagine my Phenom II X4 955 being that bad of a bottleneck. The card is passed return date so I would have to resale and take a loss if the thing is that bad. When it came out many reviewers called it the best bang for buck. Now I am hearing about these things dying and specifically mine not preforming up to par. What the eff is the issue?!
Sincerely, Disgruntled NVIDIA Customer, snoball
#1 ... online benchmarks are for comparing a bunch of cards against EACH OTHER, on the exact same rig. They don't always tell you the details of the tests they ran, either. You should not assume that what they show is what you'll get 'at home'. That's not what they're for.
But looking at specifics here:
Dirt 2: With FPS that high, the CPU could well be bottlenecking. Does it matter at 91FPS?
BFBC2: There's no benchmark, so how can you be sure you're comparing apples/apples? Also, this game likes a lot of CPU power, so a BN is not unthinkable.
Metro: Are you SURE you ran the same settings, and that you're running the same test as they did? Bear in mind that in 'High' mode, it looks like you're setting AAA but it's actually 0xAA, as AAA mode only works in VH.
This is the one test of this bunch where I would say fairly unequivocally that the CPU should not matter much, if at alll. But actually Windfire's results suggest that CPU *could* be involved a tiny bit, based on the increases he got from OC'ing the proc ... still, it's the least CPU-dependent test scenario of these four, I'm certain.
Vantage: So you're saying your GPU score is off by about 10%? In regular performance mode, you could well be getting a bit of a BN here as well. And if you're comparing total P-score you definitely will since CPU tests are directly involved in that score.
I'd try to compare GPU
scores on like Vantage High, or 3dMark11, with others on the ORB ... CPU won't matter on those tests for sure. Metro I think you haven't matched settings properly, or just aren't running the same test as HC did.Edited by brettjv - 2/7/11 at 10:49pm