Well, i’m just looking for your guys opinions, to see if you guys think this is normal or not.
My engine runs at about 60fps on a AMD 1300 /Gf2 mx , and at about 100fps on a AMD 800/GF3 , so, it’s not cpu limited…
Now, i have added some trees to the engine, and on the first computer (AMD 1300/GF2 mx) it runs at an average of 48/56 fps…, no big problems here…
On the AMD 800/GF3 it drops down to around 35/48 fps…
What’s wrong with this picture ?
Do you guys think that cpu speed is making the diference here?
I’m just doing a pushmatrix/ render mesh / popmatrix inside a for loop…
I’m so curious that i’m holding myself to opening the computers and change the boards to see what happens…
Maybe you were fill rate limited, now your AGP limited because of the extra vertices you’re sending. Is your AMD 800 motherboard AGP2x and your AMD 1300 AGP4x?
Is this SDRAM vs DDR?
When your geometry is coming from system memory that would explain it.
I was astonished quite a bit when I saw the performance jump going from SDR to DDR.
A Duron1200 (9*133) on PC2100-CL2 yielded up to a 160% performance increase over an Athlon Classic 700 on PC100-CL3. The best gains came in high-poly areas with heavy software culling load (ie worst case conditions). I attribute that to higher platform bandwith.
Both motherboards have agp4x enabled…
Could it be memory ?
The 1300 amd has 192 Ram memory, the 800 has 128 Ram.
And the 1300 in under XP, while the 800 is on WinME.
I’m not sure if this could make a diference, but i supose XP is a much better memory management.
One thing interesting…, before the trees, if i changed resolution the fps would be the same., now if i decrease the resolution on the gf2mx machine i have an increase of fps…
This is quite frustrating…, having a gf3 beaten by a gf2…