How CPU Features Affect CPU Performance, Part 5
|
Scientific calculations
Without Turbo Boost and Hyper-Threading
|
1 core |
2 cores |
3 cores |
4 cores |
Maya ↑ |
7.1 |
8.89 |
25% |
8.97 |
1% |
9.49 |
6% |
SolidWorks ↓ |
40.06 |
35.39 |
13% |
39.03 |
-9% |
38.55 |
1% |
Pro/ENGINEER ↓ |
1632 |
1526 |
7% |
1569 |
-3% |
1554 |
1% |
UGS NX ↓ |
5.47 |
5.47 |
0% |
5.34 |
-2% |
5.41 |
1% |
MAPLE ↑ |
0.2185 |
0.2194 |
0% |
0.2142 |
-2% |
0.2197 |
3% |
Mathematica ↑ |
2.5295 |
2.9533 |
17% |
2.9798 |
1% |
3.2357 |
9% |
MATLAB ↓ |
0.071984 |
0.049676 |
45% |
0.045571 |
9% |
0.038867 |
17% |
Group Score ↑ |
117 |
134 |
15% |
132 |
-1% |
139 |
5% |
There are two main differences from the AMD quad-core CPU: no negative response in any application, when the number of cores is increased from one to two, and an absolutely inadequate response, when the number of cores is increased from two to three. Note: it's already the second case with such response, demonstrated solely by Core i7 (this mode is not standard for this processor).
With Turbo Boost and Hyper-Threading
|
2 cores |
4 cores |
6 cores |
8 cores |
Maya ↑ |
7.44 |
9.11 |
22% |
9.05 |
-1% |
10.24 |
13% |
SolidWorks ↓ |
40.2 |
36.71 |
10% |
38.9 |
-6% |
35.78 |
9% |
Pro/ENGINEER ↓ |
1622 |
1495 |
8% |
1538 |
-3% |
1491 |
3% |
UGS NX ↓ |
5.66 |
5.65 |
0% |
5.43 |
-4% |
5.72 |
5% |
MAPLE ↑ |
0.2193 |
0.2256 |
3% |
0.2124 |
-6% |
0.2227 |
5% |
Mathematica ↑ |
2.5555 |
2.9248 |
14% |
2.8336 |
-3% |
3.1403 |
11% |
MATLAB ↓ |
0.061414 |
0.04521 |
36% |
0.042147 |
7% |
0.03898 |
8% |
Group Score ↑ |
121 |
137 |
13% |
134 |
-2% |
144 |
7% |
There is actually identical response to pure three cores or six cores with HT. As in case of Phenom II X4, MATLAB has the best optimizations for multithreading.
Compared to AMD Phenom II X4
Scalability curve of Phenom II X4 looks much more logical again, as Core i7 shows a significant drop at 3(6) cores. But even if we ignore this drop, performance growth curve is still much smoother.
Write a comment below. No registration needed!
|
|
|
|
|