B. KU8
1600*1200
2025-10-31 06:23:42 - hl2
Frames: 3925 - Time: 60000ms - Avg: 65.417 - Min: 17 - Max: 91

1280*1024
2025-10-31 06:26:28 - hl2
Frames: 4227 - Time: 60000ms - Avg: 70.450 - Min: 54 - Max: 109

1024*768
2025-10-31 06:25:07 - hl2
Frames: 4236 - Time: 60000ms - Avg: 70.600 - Min: 57 - Max: 99

C. KV8
1600*1200
2024-12-26 09:00:15 - hl2
Frames: 3969 - Time: 60000ms - Avg: 66.150 - Min: 51 - Max: 89

1280*1024
2024-12-26 09:03:13 - hl2
Frames: 4238 - Time: 60000ms - Avg: 70.633 - Min: 17 - Max: 116

1024*768
2024-12-26 09:04:50 - hl2
Frames: 4711 - Time: 60000ms - Avg: 78.517 - Min: 57 - Max: 99

Again ULI takes a severe hit because of AC97 codec poor implementation. Either Abit deliberately cheaped out or engineers made a few errors. I am inclined to believe the first as - contrary to NF8 and KV8 - KU8 has only 3 audio jacks so they cut corners everywhere!

46. Doom 3 is the first opengl game to be tested and, because of performance issues I chose to test with AA disabled but at ultra quality and all advanced settings ON. Again the in game test does NOT use sound.

1024*768
A. NF8
B. KU8
C. KV8

47. We are already below 60 FPS average so taking Doom to the next resolution will make the framerate even lower:
A. NF8
B. KU8
C. KV8

48. At 1600*1200 we are completely GPU limited at during actual gameplay it will drop to the low 30's - still playable but only just!
A. NF8
B. KU8
C. KV8

49 & 50 & 51. The second opengl game is Quake 4. Extremely demanding, I used hardwareOC quake 4 bench tool for testing which is a blessing - during actual gameplay you will run out of memory and out of FPS with the poor 7800gs!
A. NF8
B. KU8
C. KV8

There is nothing to see as the game is actually unplayable - you need more RAM and a much more powerful video card!

52. The last game to be tested is the 2005 system killer FEAR. For ease of use and ease of comparison the first screenshots at 1024*768 have maximum selected for both computer and graphic card. However when you select both AA and soft shadows (SS) simultaneously SS will actually be disabled. You have to select sequentially - AA first and exit to main option menu and SS second in order to have them both enabled. SS would annihilate the geforce 7800gs so it is far better to leave it off for any directx9 GPU!
1024*768
A. NF8
B. KU8
C. KV8

When you start with a 42 average FSP you know things are bad!
53. 1280*1024
A. NF8
B. KU8
C. KV8

Things are looking really bad at 1280*1024!
54. However at 1600*1200 are looking horrible!
A. NF8
B. KU8
C. KV8

FEAR is not playable on such a system regardless of the resolution! You need a geforce 8800 GTX at a minimum to actually be able to enjoy the game and even with that card you will be limited to 1024*768!

So 3400+ 2.4Ghz is a good alternative. There is a very small performance penalty in games due to smaller cache, but at ~60 fps it is only about 1-2 fps thus ok. It is slightly better in synthetic benchmarks at the same time so the PR rating seems fair. Clawhammer has larger L2 cache but this cache has terrible latency.

Now the question is why should one buy s754 Athlon 64 with AGP instead of s462 Athlon XP. What does one gain in practice. What is the difference in performance between nForce2 and s754 boards.