This bench is much more difficult than its predecessor.
The first three scenes are progressively more and more difficult,
and it shows. Whereas I easily attained 200 frames in the first scene
(B17s attacked by Messerschmits), I was at great difficulty to get 20
frames in the (third) Troll's Lair scene.
Of note : installing version 4.11 of Catalyst drivers boosted the score
by 200 points.
Which basically means that next year I'll be blowing another wad of cash. Oh bother.
So this is the culprit. This is the bench that started it all. I must say that that 43k over 17k is music to my ears, but that only lasts until I realize that, 12 months from now, I'll be wanting 80k.
Or, more realistically, I'll be wanting 43k minimum. As it stands now, my min score is 11.46 fps, which translates to 11 461 points. High score is 108 000, or 108 fps. Fine, but we all know that high or average count for nothing. More fluid than fluid is useless, the rule is to avoid stutter. Unfortunately, at 11 fps, stutter is still unavoidable.
Okay, the stuttering only occured in the Massive Overdraw sequence. It is obvious that things will rarely get that bad, but still, they might. And given the industry's tendancy to gobble up all available processing power, things will only get worse.
Codecreatures is about as bad as 3DMark 2003. What is shown is mostly fluid, but framerates are in the low fourties most of the time. Hey, I probably shouldn't complain, it is a clear improvement from before. But we all know that 40fps is the minimum to ensure fluid video, and that minimum is still not attainable on the best hardware available for AMD processors.
The full table is here :
| Total | Catalyst 4.11 | |
|---|---|---|
| Avg fps | 45.7 | 49.6 |
| Max fps | 60.0 | 60.0 |
| Avg polys/s | 16.3 mio | 18.1 mio |
| Max polys/s | 23.2 mio | 26.1 mio |
| Total frames | 5 988 | 6492 |
| Grass Section | Catalyst 4.11 | |
|---|---|---|
| Avg fps | 50.1 | 52.6 |
| Max fps | 60.0 | 60 |
| Water Section | Catalyst 4.11 | |
|---|---|---|
| Avg fps | 42.1 | 47.2 |
| Max fps | 58.0 | 60 |
Interesting to note that the water section was actually slower than the grass section, whereas visual observation suggested the reverse.
For once, an upgrade did indeed increase Sharok's performance by a quite noticable degree. I have over 100% increase in average figures for my benchmarks, which is good.
What is also good is that I will be upgrading Falmir shortly with the pieces I took from Sharok. That will give a nice level of performance across the board.
What really got my attention is the lack of noise from the cooling
fan on the Radeon. Compared to the GF4 Ti 4400 I had before, this baby
is silent. Heck, the video card is more quiet than the
PSU - under full load, no less. So, I'm really happy with this card. I
have a gut feeling that next year's video card purchase will be another
ATI card.
Update !!
3DMark 2005 is out ! Here are my results.
Warning : it hurts.