|

Now comes the all important performance analysis.
First up, we are going to look at gaming benches, and then we will
wind up with some CPU performance benchmarks.
Gaming Performance
To test the gaming performance of the Phenom II
940, I pitted against what I believe to be the best gaming CPU on
the market right now: intel's E8600. We tested head to
head with maximum overclocks on both; 3.7GHz on the AMD and 4.4GHz
on the intel, both completely stable. The E8600 makes for an
interesting comparison for gamers because both CPUs should be right
around the same price, and since gaming performance is important for
so many people, this should prove to be very interesting.
Below is the intel comparison system for the gaming benches:
| CPU: |
Intel E8600 @ 4.4GHz |
| Motherboard: |
FoxConn BlackOps X48 |
| GPU: |
Sapphire 4870X2 |
| RAM: |
2x2GB OCZ Reapers, DDR3-1760 8-8-8-27 |
| Case: |
Danger Den Torture Rack |
| Sound: |
Soundblaster X-Fi |
| Cooling: |
Swiftech H20-220 Compact |
| PSU: |
Corsair 1000HX |
| OS: |
Vista x64 Ultimate SP1 |
X3: Terran Conflict
First up, we are going to look at some benchmarks
I know to be very CPU dependent. First up for testing is X3:
Terran Conflict. This benchmark has four different tests to
represent different scenarios with in the game. This test
loves pure clock speed on both the CPU and the GPU for effective
testing. Testing was done at 1920x1200, 8xAA, 16xAA, with all
the goodies turned on. Here are the results:

| |
Phenom II 940 @
3.7GHz |
E8600 @ 4.4GHz |
| Trade |
43.7 |
60.8 |
| Fight |
94.9 |
127.5 |
| Build |
135.1 |
171.4 |
| Think |
60.7 |
75.5 |
| Average |
83.6 |
108.8 |
As we can see here, the clock speed of the E8600
cruises past the Phenom II, with about a 20% performance advantage
across the board.
Supreme Commander
This is not a test I usually bench with, however
it is notoriously CPU dependent so I thought it would make for
another good candidate. This was also tested with all settings
maxed at 1920x1200, using the /perf option in the command line to
invoke the built in bench:

While the minimum framerate count is about even
between the two CPUs, the E8600 again takes the lead, which is quite
surprising for a game that is supposed to be able to take advantage
of four cores.
Far Cry 2
Next up is Far Cry 2. This first person
shooter features a wide-open world environment with tons of AI.
It also happens to feature a pretty nice benchmarking tool.
This test was run with the "Ultra" preset, at all common resolutions
and AA settings.

|
|
E8600 @ 4.4GHz |
X4 940 @ 3.7GHz |
|
|
Min |
Max |
Avg |
Min |
Max |
Avg |
|
1280 0xAA |
56.17 |
133.6 |
84.78 |
41.65 |
103.29 |
64.39 |
|
1280 2xAA |
57.4 |
132.08 |
86.85 |
41.5 |
99.4 |
63.07 |
|
1280 4xAA |
58.45 |
120.91 |
82.48 |
41.88 |
100.76 |
64.14 |
|
1280 8xAA |
42.97 |
92.78 |
63.07 |
41.64 |
88.94 |
59.18 |
|
1680 0xAA |
57.24 |
135.69 |
85.91 |
41.81 |
98.6 |
63.43 |
|
1680 2xAA |
58.63 |
118.7 |
82.72 |
41.68 |
100.12 |
63.46 |
|
1680 4xAA |
50.58 |
103.03 |
72.68 |
42.32 |
98.76 |
63.54 |
|
1680 8xAA |
33.01 |
79.74 |
52.14 |
32.87 |
74.86 |
50.38 |
|
1920 0xAA |
56.35 |
125.85 |
84.12 |
41.79 |
102.36 |
63.67 |
|
1920 2xAA |
57.03 |
108.66 |
79.53 |
41.82 |
98.42 |
63.41 |
|
1920 4xAA |
48.07 |
93.95 |
66.97 |
42.71 |
91.69 |
61.62 |
|
1920 8xAA |
23.86 |
74.66 |
39.56 |
28.02 |
66.74 |
44.34 |
It would appear here that Far Cry 2 is much more
CPU limited than I expected; it appears that for the most part the
only testing that is GPU bound here is when 8xAA is applied.
The E8600 definitely wins out here.
Crysis: Warhead
This is turning into one of my favorite
benchmarks. It's not as brutal as Crysis, but it is truly one
of the funnest games out there, and very pleasing to the eye.
Just like Far Cry 2, the test was run with the "Enthusiast" preset,
at all common resolutions and AA settings.
|
|
E8600 @ 4.4GHz |
X4 940 @ 3.7GHz |
|
|
Min |
Max |
Avg |
Min |
Max |
Avg |
|
1280, 0x |
31.83 |
61.75 |
47.92 |
22.08 |
60.21 |
41.99 |
|
1280, 2x |
30.44 |
54.66 |
42.92 |
21.94 |
54 |
40.17 |
|
1280, 4x |
29.44 |
55.18 |
42.73 |
21.74 |
54.75 |
40.1 |
|
1280, 8x |
25.95 |
52.11 |
39.53 |
21.44 |
51.11 |
37.99 |
|
1680, 0x |
23.5 |
50.38 |
39.95 |
20.4 |
48.95 |
38.01 |
|
1680, 2x |
22.47 |
44.56 |
35.27 |
20.21 |
43.08 |
34.44 |
|
1680, 4x |
21.96 |
45.38 |
35.4 |
20.99 |
43.91 |
34.43 |
|
1680, 8x |
21.75 |
42.09 |
32.34 |
19.26 |
42.24 |
31.6 |
|
1900, 0x |
21.68 |
42.81 |
34.33 |
20.69 |
42.88 |
33.54 |
|
1900, 2x |
20.15 |
37.56 |
30.07 |
19.81 |
37 |
29.53 |
|
1900, 4x |
19.62 |
38.22 |
30.04 |
18.37 |
37.81 |
29.46 |
|
1900, 8x |
8.97 |
35.98 |
27.2 |
12.62 |
35.65 |
26.63 |
The story here is a bit different than the one for
Far Cry 2. While the E8600 does indeed win at 1280x1024, once
you hit 1680 or higher, the benchmark is entirely GPU bound.
|