CPU Performance
Both Winbench99 and Passmark provide CPU specific performance results. Passmark is particularly nice in that it separates out not only integer vs. floating point, but actually compares the them by operation, and also includes tests to test MMX instructions specifically.
Passmark Math Tests | AX6C/DRDRAM | AX6BC/SDRAM | AX63/SDRAM | AX63/VCSDRAM |
Add | 93.7 |
94.2 | 92.5 |
92.3 |
Subtract | 93.9 |
94.2 | 92.6 |
92.5 |
Multiply | 89.7 |
90.0 |
88.3 |
88.3 |
Division | 13.1 | 13.1 | 13.1 | 13.1 |
FP Add | 85.9 |
86.0 | 84.6 |
84.3 |
FP Subtract | 85.9 |
86.0 |
84.6 |
84.6 |
FP Multiply | 85.1 |
85.2 | 83.8 |
83.7 |
FP Division | 14.9 | 14.9 | 14.9 | 14.9 |
Math Mark | 64.4 |
64.6 | 63.5 |
63.4 |
Passmark MMX Tests | AX6C/DRDRAM | AX6BC/SDRAM | AX63/SDRAM | AX63/VCSDRAM |
Add | 164.2 |
165.5 |
162.8 | 163.4 |
Subtract |
165.3 | 163.6 | 163.6 |
163.2 |
Multiply | 162.2 |
162.3 |
159.5 |
159.5 |
MMX Mark |
84.3 | 83.4 |
82.4 | 82.5 |
WinBench99 | AX6C/DRDRAM | AX6BC/SDRAM | AX63/SDRAM | AX63/VCSDRAM |
CPU Mark99 | 38.3 |
38.7 |
38.2 | 38.3 |
FPU WinMark | 2560 | 2560 |
2550 | 2560 |
Even though the same processor was used on all platforms, the differences in the other subsystems appear to have an effect on the CPU performance. Not all of this difference can be attributed directly to memory bandwidth or throughput, since we have seen that they are fairly close between platforms. It would seem that latency has a significant impact here also. If we can rely upon the latency values provide in the addendum by Cachemem, there would seem to be some correlation here.
Pages: « Prev 1 2 3 4 5 6 7 8 Next »
Be the first to discuss this article!