A friend's machine dual boots between w98se and w2k. I put 3dmark 2000 v1.1 on both os's to do some testing. The important parts of the machine are as follows:
ASUS A7V mobo
Athlon 800
ASUS V7700 Geforce 2
256Mb RAM PC133
Drivers:
w98se - NVIDIA reference 6.34
w2k - NVIDIA reference 6.34 (tho dxdiag says it's still 6.31... hmm...)
w98 has DX7 (It had DX8 but I reverted)
w2k has DX8 (don't know how to revert that back to DX7
)
Now, when I do the high polygon count tests in 3dmark the results differ greatly between 98 and 2k. I haven't bothered to compare the other tests but when you see the numbers below you'll see that there's definately something wrong...
98, 800x600x32, D3D Hardware T&L, loop 3 times:
1 light - 11153
4 lights - 9884
8 lights - 5585
2k, 800x600x32, D3D Hardware T&L, loop 3 times:
1 light - 3323
4 lights - 1981
8 lights - 1015
The difference is clearly visible when watching the tests running. In 98 it's smooth in all cases but in 2k only the 1-light test is smooth. The 4 and 8 light ones are very jerky and possibly even in the single-digit fps for 8 lights.
I tested it with software t&l too:
98, 800x600x32, D3D Software T&L, loop 3 times:
1 light - 5731
4 lights - 3028
8 lights - 1936
2k, 800x600x32, D3D Software T&L, loop 3 times:
1 light - 4281
4 lights - 3026
8 lights - 2235
Now these seem much more similar. Sure they differ, but not like the difference when using hardware t&l!
I don't know how to revert 2k back to DX7 to see if the performance increases or not.
I've seen this same sort of problem on another friend's machine that has almost the same setup except it's an Athlon 900 with a Geforce 1 DDR (DX8 in 98 and 2k). The 2k high polygon tests are very bad , but they're fine in 98.
Does anybody get similar results? Do you know if it's DX8 causing the problem?
ASUS A7V mobo
Athlon 800
ASUS V7700 Geforce 2
256Mb RAM PC133
Drivers:
w98se - NVIDIA reference 6.34
w2k - NVIDIA reference 6.34 (tho dxdiag says it's still 6.31... hmm...)
w98 has DX7 (It had DX8 but I reverted)
w2k has DX8 (don't know how to revert that back to DX7
)Now, when I do the high polygon count tests in 3dmark the results differ greatly between 98 and 2k. I haven't bothered to compare the other tests but when you see the numbers below you'll see that there's definately something wrong...
98, 800x600x32, D3D Hardware T&L, loop 3 times:
1 light - 11153
4 lights - 9884
8 lights - 5585
2k, 800x600x32, D3D Hardware T&L, loop 3 times:
1 light - 3323
4 lights - 1981
8 lights - 1015
The difference is clearly visible when watching the tests running. In 98 it's smooth in all cases but in 2k only the 1-light test is smooth. The 4 and 8 light ones are very jerky and possibly even in the single-digit fps for 8 lights.
I tested it with software t&l too:
98, 800x600x32, D3D Software T&L, loop 3 times:
1 light - 5731
4 lights - 3028
8 lights - 1936
2k, 800x600x32, D3D Software T&L, loop 3 times:
1 light - 4281
4 lights - 3026
8 lights - 2235
Now these seem much more similar. Sure they differ, but not like the difference when using hardware t&l!
I don't know how to revert 2k back to DX7 to see if the performance increases or not.
I've seen this same sort of problem on another friend's machine that has almost the same setup except it's an Athlon 900 with a Geforce 1 DDR (DX8 in 98 and 2k). The 2k high polygon tests are very bad , but they're fine in 98.
Does anybody get similar results? Do you know if it's DX8 causing the problem?
Facebook
Twitter