Following setup was used during the tests:
Asus Radeon 6850 1024MB
4x2GB Patriot DDR3 1600MHz
Verbatim 120GB SSD SATA-III
Enermax EMG750AWT PSU
CoolIT ECO ALC Watercooler
Intel Core I5 2500k
Asus P8P67 rev 3
Generally speaking the faster the CPU is, the more bandwidth it has and the lower the latency is. FX-8120 results in this test are bit weird. While it has higher bandwidth versus the i5 2500k, the latency is rocketing sky high. The results were consistent throughout our test periods (we did run a test multiple times) and did not find any reason why the latency was so high.
We use Sisoftware Sandra 2012 to measure processors ALU and FPU performance. FX-8120 beats i5 2500K in Whetstone test with a neglible margin but in the Drystone test the 2500k takes the lead. The difference in Drystone test was 30,59Gips.
We used Cinebench to measure the rendering speed and the results are highly affected by the amount of cores available. Despite that, the AMD FX-8120 did not quite manage to beat I5 2500k. The difference was 0.79 points which is not major, but still noticeable.
In the single threaded computing test we used Superpi to calculate 8M digits of pi. Like in the past, AMD has problems managing with good single threaded performance. i5 2500k manages to push three times faster results in this test.
PCMark is a great tool to measure a general performance of your processor. We did both a general test and computation test in this program. I5 2500k manages to beat FX-8120 in both tests with quite a good margin.
We used wPrime to test multithreaded performance of our test subject. The results we’re quite similar between i5 2500k and FX-8120. Both manage to perform about the same and while there is some difference in test results, the difference is not huge and should not be worried about.
In overclocking we managed to push our sample unit to 4.6GHz speed. This required raising the multiplier to 23x and raising the voltage to 1.40V. We did not try to finetune our processor by raising the FSB as we thought the difference would not be mind blowing and we tried to get a general idea how these processor overclock instead of finding the absolute maximum of our test unit. The temperature of our processor was around 79c with CoolIT ECO ALC cooler and the power consumption of the test setup was in total of around 380W (compared to around 270W without overclocking). This leads to a conclusion that the power consumption of Bulldozer processors rockets sky high when they are overclocked and raiding the vcore should be done very carefully.
FX-8120, like the Bulldozer platform in whole, is bit of a weird product on the market. While it has mediocre performance in multithreaded programs, it falls far behind similarly priced Intel processors in singlethreaded programs. Intel i5 2500K did beat the FX-8120 with more than three times better results in some tests and throughout our tests it showed better or similar performance. As they are both priced similarly around 200USD, it may be hard to justify a reason why you should buy the FX-8120. The only reason I could think of is if you need more than four cores and know the downsides of the bulldozer architecture. Other reasons than that, don’t bother!
You can pick this cpu up from Newegg for a good price.
+ Good performance in multithread programs
- Bad performance in single thread
- Power consumption (especially if overclocked)
Pages: 1 2