Page 12 of 16
Passmark Performance Test Benchmark
PassMark PerformanceTest is a PC hardware benchmark utility that allows a user to quickly assess the performance of their computer and compare it to a number of standard 'baseline' computer systems. The Passmark Performance Memory Test looks at a variety of memory functions:
- Allocate Small Block
- Read Cached Memory
- Read Un-Cached Memory
- Write Memory
- Large RAM
It then calculates an overall performance number for Memory Marks. Unlike the Everest memory benchmarks, this test uses all available CPU cores. At similar clock speeds the Intel i5 beats the AMD 555BE by 39%. Even though I saw all cores being used by this test, using task manager to see the individual CPU loads during unrecorded trials, the results look more like a single-core test.
The CPU tests benchmark all the mathematical operations, compression, encryption, SSE, and 3DNow! instructions of a modern processor. There are several areas of concentration for each benchmark, which are then combined into one compound score. This score is referred to as the CPU Mark, and is a composite of the following tests:
Floating Point Math
Find Prime Numbers
The CPU test shows a 69% performance gain for the Core i5-750 over the Phenom II 720BE at roughly similar CPU clock speeds. If you factor out the difference in the number of cores, the difference between the i5-750 and 555BE gets narrowed down to 17%. There's little doubt about the power of the P55/i5 platform, especially in CPU-bound applications or where memory performance has an impact. But that performance comes at a price, literally.
Wrapping up the synthetic benchmarks, what can we say about the overall results? Obviously, in non-gaming situations, four cores beat two or three cores. Applications are getting better at using multicore processors to their fullest capability and computationally intensive tasks like video and audio encoding really benefit from the additional cores. If you've got a gaming rig though, you are better off running those tasks on your GPU. It'll smoke the CPU every time. If you don't, then every additional CPU core makes a difference.
The other thing that struck me was the way the M4A88TD-V EVO/USB3 motherboard and the AMD 555BE CPU scaled their performance in every benchmark, with a fairly simple overclock to a very respectable 4.0 GHz. The results for the 790FX/720BE and P55/i5-750 each show the best 24/7 overclock I was able to achieve with that MB/CPU combination, so I set the bar pretty high for the 880G/555BE. There is a limit to how much performance you can squeeze out of two cores, but in the end the actual performance is dictated by how well the application uses all available cores. CINEBENCH gives almost perfect CPU scaling results, while PCMark Vantage mixes up a variety of tasks and threads, and provides more of a system-level benchmark. The lower cost solution looks more viable at the system level, which reinforces the accepted wisdom that for general purpose usage, people tend to buy more computing power than they need.
Let's take a look at two gaming benchmarks to see how well this CPU performance translates in the graphical world.