Application and Synthetic Testing
Sandra Memory Bandwidth
Note: All systems were run in dual channel or quad-channel modes where applicable. The 9900K and 3900X used timings of 16,18,18,36@1T. The 10980XE was restricted to the timings of the RAM I had on hand which were 18,19,19,39@1T. For the memory bandwidth test, the Core i9 10980XE test system was run with the memory clocked at 3800MHz.
In this test, we see a result of 31.6GBs at DDR4 3200MHz and 39GB/s when using DDR4 3600MHz memory. For a dual-channel system, these are solid results. Generally, AMD turns in better results here, but Intel isn’t far behind.
Sandra CPU Dhrystone
The Core i9 9900K produces a result of 428GiPS when overclocked and 416GiPS stock. The 9900K achieves a slightly better result at stock speeds which is something you should get used to. This can probably be chalked up to Windows build differences or additional mitigation patches.
PCMark 10
In this test, we see the 9900KF fall slightly behind the 9900K once again. Overclocked, we get a slightly higher result.
WinRAR – Multithread
As stated many times, this is an odd test. The results favor clock speeds as well as multi-threading. Here, the 9900KF is competitive with the 3900X at both stock and overclocked speeds.
WinRAR – Single Thread
In the single-threaded test, these test systems are all virtually the same. Massive clock speed changes would be needed for a substantially different result.
wPrime v2.10
Again, we see a bit of a backslide in performance from our 9900KF vs. our 9900K.
POV-Ray
POV-Ray shows almost no improvement from clock speed changes. All three of our Coffee Lake CPU’s perform virtually the same.