Benchmark Scores
While the vast majority of PC game and simulation developers do not yet support SMP, the situation will hopefully change in the future as both Microsoft and Sony’s next-generation consoles help evangelize the need for the programmers at larger development houses to learn multi-core support. For simulation fans, however, there is always the halcyon Falcon 4, a title SimHQ has gladly made heavy use of for dual-core testing purposes last year. And with the re-release of the title as Falcon 4: Allied Force, Lead Pursuit has claimed to further optimize the sim’s SMP support, making it an extremely useful and relevant benchmark for SimHQ’s readers with these new dual cores. The two mission tests that are used in this article can be found here; both missions were ran at a resolution of 1024×768 across all tested processors and the simulation itself patched with the 1.03 update.
Test 1: Ground, is a short MUD moving flight that leaves Mandumi, heads over the FLOT, and attacks an enemy supply depot that has heavy AAA and SAM protection. Performance is recorded during the first six minutes of the mission and the scores shown representing the average frame rate.
|
Test 1: Ground
|
|
| FX-57 | 59 |
| FX-60 | 68 |
| 4800+ | 65 |
| 955 | 57 |
| 840 | 51 |
Click for enlarged image.
As shown before, the AMD processors lead by a substantial margin, with even the FX-57 outperforming the Intel dual cores. The FX-60 nudged out the older X2 4800+ by just a few frames, though both parts exhibited a 10%+ advantage over the other tested parts, with the 840 falling significantly behind. Beyond the average frame rates, the frames-per-second flow chart shows that the dual cores from both companies displayed fewer frame rate fluctations than what the FX-57 exhibited. It’s worth noting how the paired dual cores from each company appeared to track similar performance lines, deviating at roughly the same points in the test.
Test 2: Air, is also based on the Korea map. The mission starts with the threat warning siren blaring as the flight comes under immediate attack by Su-27s. The Air mission is shorter than the Ground test, with the frame rate recorded for three minutes.
|
Test 2: Air
|
|
| FX-57 | 65 |
| FX-60 | 74 |
| 4800+ | 68 |
| 955 | 62 |
| 840 | 58 |
Click for enlarged image.
The Air test scores show the FX-60 outperforming the 4800+ more than the part did during the Ground test, outscoring the older processor by almost 10%. Interestingly enough, the FX-57 barely trailed behind the 4800+ in this test, which in turn outscored Intel’s 955 and 840 by roughly 10% and 20%, respectively. Looking at the performance flow lines, the FX-57 shows an enormous amount of deviations in its recorded frame rate in the first half of the mission, though all tested processors display a higher amount of fluctuations than in the Ground mission. The 955 and 840 show the lowest drops in frame rate, dropping into the low 30s toward the end of the test.
With the PC market still heavily dominated by titles that do not support SMP the performance of these parts in such sims still requires some testing. As such, SimHQ decided to run these dual cores with Pacific Fighters (v4.02) and compare how well they perform against AMD’s FX-57. Pacific Fighters was configured to run in OpenGL mode with all video options set to medium and tested using the in-game “F4F vs. G4M” track.
Click for enlarged image.
Unsurprisingly, AMD’s FX-57 led the way across all tested resolutions, outperforming its dual-core siblings by a noteworthy margin only at the lower settings. In contrast, Intel’s dual cores fell behind by a significant amount, with the FX-57 outperforming the 840 by almost 40%, though the 955’s faster FSB and frequency helped close the gap. While the 955 lost by roughly 10% to last year’s 4800+, the FX-60 widens the performance lead for AMD’s dual cores. Readers should bear in mind that Pacific Fighters also exhibits a better relative performance to AMD parts on Intel processors than a lot of other titles on the market, so it is quite possible that further testing of non-SMP simulations would’ve seen an even greater disparity between the two brands of dual cores.
