Sign in with
Sign up | Sign in
Your question

Is my AMD CPU gimping my Video Card? MSI R7950

Last response: in CPUs
Share
August 13, 2012 4:25:39 PM

Hello, so to start this off here are my PC Specs:

CPU: AMD Phenom II 960T 4 cores @ 4.0ghz, but only by a x20 multiplier, no fsb overclock.
CPU Cooler: Antec Kuhler H20 620
Mobo: Asus M5a880-V EVO
GPU: MSI R7950 Twin Frozr

So I was trying to bench my gpu last night and I cam up with some very strange results:

First off I was using Furmark, which was so messed up.
Preset 720p: 3785 score at stock 880/1250. 3785 score overclocked at 1100/1400. So idk why this didn't change, also my score is about the MSI 580 Lightning.
Preset 1080p: 2159 stock and overclocked same score, again. This time it's a maybe 5% higher than the 580 Lightning.

So why isn't overclocking giving me a better score with Furmark?

Next I used 3DMark11 where things get worse:
At Stock I Scored followed by benches for the same card but with an i7 970 cpu (I think, either way it wasn't even
Overall: P6294 me // P7432 i7
Graphics: 7277 me // 7016 i7
Physics: 4524! me // 11378 i7
Combined: 4413 me // 6913 i7

clearly my graphics score is better, but I am smoked in every other category, is this indicative of gaming performance? It gets even worse overclocked:
O: P7037 // P8702 = 1700 24% difference
G: 8692 // 8406
P: 4558 // 11361
C: 4268 // 8006

Now clearly the main difference is the Physics score, but it seems to make a humongous difference in the rest of the scores. I don't really know what this bench means in actual game performance though, I'm not a bench head, but I wanted to see how much overclocking was improving my performance, which is about 19.5%, very nice, but am I holding back performance by a huge amount with my AMD cpu?

Lastly I used Unigine Heaven, which really wasn't too much of a difference, but comparing clock speeds it seems to be big.
The settings used were:
1080p, high shaders/textures, tri-linear, anisotrophy x4, occlusion/refraction/volumetric enabled, tessellation normal.
comparing my 960T @4ghz to an i5 2500k @ 3.3ghz
FPS: 70.6 // 71.5
MIN: 33.4 // 30.8
MAX: 154.8 // 158.8
Score: 1778 vs 1800 for the i5

Not bad, except the i5 is clocked at a measly 3.3 when I know they can easily push 4.5

So, I am getting killed in benchmarks by an i5 system's benches, even though I appear to have a better card than was used in the test (perhaps due to newer drivers). Is this indicative of perfomance in gaming? Will I see a significant boost in FPS and performance by switching to an i5 sandy or even ivy bridge? It appears so from the benchmarks. Does anyone have any personal experience? What about going from PCIE 2.0 to PCIE 3.0, right now when I am running my card I see this from HWinfo:
Radeon 7000 series: PCIe v3.0 x16 (8.0 GB/s) @ x8 (5.0 GB/s or 2.5 GB/s at idle). I am plugged into a PCIe 2.0 x16 slot. I don't know if I am saturating the bandwidth or not, or if I would if I switched to a Sandy or Ivy Bridge (obviously if I would see any real improvements I would move to Ivy to get support for PCIe 3.0)

And thanks for all the help from anyone willing to help or just put their 2 cents in.

btw. I went with a 960T and an Asus 970 motherboard because I got them for $130 total used, I ended up putting that motherboard in my girlfriends build with a 955 BE, but then my motherboard died and I used this extra Asus 880. I don't think I would see much of a performance increase using a 990FX board, maybe I could clock a bit higher, but it doesn't seem like it's worth the money when I should probably be moving to i5 3570K and z77 board. I was thinking the Asus Sabertooth z77.
a b à CPUs
a b À AMD
August 13, 2012 4:53:02 PM

Dont completely trust synthetic benchmarks as they are not always indicative of real game performance, one thing you can do is to get msi afterburner and enable the OSD to show GPU load and fps, if the load is around 96 - 98% then you are not cpu limited, sometime you will get slowdowns but also have the fps counter showing if you use vsync as it could just be that your gpu is coping fine with the scene at 60fps.
m
0
l
August 13, 2012 6:46:53 PM

Quote:
Is this indicative of performance in gaming?
Why not see for yourself? Test your games for FPS and GPU/CPU usage. Compare your FPS to systems with the same graphics card. If your GPU usage is below 90% and your CPU usage is a constant 100% you may be suffering from a CPU bottleneck. This is entirely game dependent though. May I ask what game you plan to play on this system?

m
0
l
Related resources
!