- Joined
- Sep 5, 2002
- Location
- Toronto, ON
I thought that some might be interested in the CPU scaling performance of the 9500 Pro, so I ran some benchmarks at a few different CPU speeds to get an idea of it. I think that CPU scaling doesn't get enough attention from the reviewers in general.
It's nice to know how well a card performs with the fastest processor out there, but I don't think that the majority of the users out there have a $600 CPU in their systems, so how well the card performs with lower speed CPUs is important.
I ran my XP 1700+ at four different clock speeds and benchmarked Quake 3 and 3dmark under each. The memory settings were set to turbo (cas2), and the fsb and memory bus were held constant at 166MHz in all the tests. So at 1.0GHz this chip is significantly faster than a Thunderbird at that speed, as an XP is faster at the same clock speed, and you have a higher bus speed.
All tests were done at 1024 at default settings. The Radeon 9500 Pro was also left at default speeds (still haven't figured out a way to OC it anyway).
CPU Clock (GHz) Quake 3 (fps) 3dmark
1.00 137.7 9238
1.33 164.5 10520
1.66 184.7 11279
1.91 194.2 11626
As you can see there is a big jump in performance from 1 to 1.33GHz, then it starts to tail off, but it's still significant from 1.33 to 1.66GHz.
It might be beneficial to look at the same results at percentage increases:
CPU increase Q3 increase 3dm increase
33.00% 19.46% 13.88%
24.81% 12.28% 7.21%
15.06% 5.14% 3.08%
The Q3 benches jump more than the 3dmark benches, probably because the game is more CPU-dependent.
I would've liked to use 2.0GHz instead of 1.91, but I couldn't complete 3dmark at that speed. Quake3 gave me 198.9 fps at that level though, so it shows you that it's not done scaling. I would imagine that with a 2.0GHz Athlon XP you're getting close to fully utilizing the card.
Not to say that these results are completely conclusive, they were done pretty quickly. I think it gives a pretty good idea of the scaling though, and if you're close to that 1.33GHz Athlon XP level there or lower, you're probably not going to be close to utilizing the full potential of this card.
It's nice to know how well a card performs with the fastest processor out there, but I don't think that the majority of the users out there have a $600 CPU in their systems, so how well the card performs with lower speed CPUs is important.
I ran my XP 1700+ at four different clock speeds and benchmarked Quake 3 and 3dmark under each. The memory settings were set to turbo (cas2), and the fsb and memory bus were held constant at 166MHz in all the tests. So at 1.0GHz this chip is significantly faster than a Thunderbird at that speed, as an XP is faster at the same clock speed, and you have a higher bus speed.
All tests were done at 1024 at default settings. The Radeon 9500 Pro was also left at default speeds (still haven't figured out a way to OC it anyway).
CPU Clock (GHz) Quake 3 (fps) 3dmark
1.00 137.7 9238
1.33 164.5 10520
1.66 184.7 11279
1.91 194.2 11626
As you can see there is a big jump in performance from 1 to 1.33GHz, then it starts to tail off, but it's still significant from 1.33 to 1.66GHz.
It might be beneficial to look at the same results at percentage increases:
CPU increase Q3 increase 3dm increase
33.00% 19.46% 13.88%
24.81% 12.28% 7.21%
15.06% 5.14% 3.08%
The Q3 benches jump more than the 3dmark benches, probably because the game is more CPU-dependent.
I would've liked to use 2.0GHz instead of 1.91, but I couldn't complete 3dmark at that speed. Quake3 gave me 198.9 fps at that level though, so it shows you that it's not done scaling. I would imagine that with a 2.0GHz Athlon XP you're getting close to fully utilizing the card.
Not to say that these results are completely conclusive, they were done pretty quickly. I think it gives a pretty good idea of the scaling though, and if you're close to that 1.33GHz Athlon XP level there or lower, you're probably not going to be close to utilizing the full potential of this card.