Disabling the 3600X's Fastest Cores
Disabling the Fastest Core
To see the maximum speed possible with the fastest core removed, we disabled Core 4.
Ryzen Master marks Core 4 (in CCX 1) as the fastest single core in the chip, and Core 2 as the fastest core in the adjacent CCX 0 (but second-fastest overall). However, Ryzen Master only allows cores to be disabled in groups of two, one in each CCX, so we also removed Core 3, which is the slowest core in CCX 0.
This means we are now testing a 6-core 12-thread chip as a quad-core eight-thread processor. It's important to note that disabling these cores removes their L1 and L2 caches from the equation, but the same 16MB of shared L3 cache is still available, meaning that shouldn't adversely impact performance of the individual cores.
The utilization chart shows that Core 2 is now being targeted with the workloads. This makes sense, as it is marked as the second-fastest core in the chip. However, as you can see, Core 2 can only sustain a frequency of 4.25 GHz, which is 100 MHz below the sustained 4.35 GHz we measured with Core 4 in our prior tests. We do see three single measurements of 4.35 GHz, but these aren't nearly as impressive as the sustained boost periods we recorded with Core 4.
Also, notice that the workload now spans the entire length of the chart – and then some. This iteration of the test took 30 minutes (94 data points spill off the chart), while the testing with the fastest core enabled (prior page) only took 28 minutes and 25 seconds.
That ~100 MHz difference in average speed during the test seems slight, but it does equate to lower performance.
Disabling Both Fastest Cores
This time around, we disabled both fastest cores (Core 2 and Core 4), so we're testing it in a quad-core configuration again. The workload now moves to Core 1, which is designated as the second-fastest core in CCX 0. This core oddly sustains a 4.3 GHz boost frequency, which is higher than what we recorded with the fastest core in the same CCX (Core 2). This could be the chip adjusting for power density due to less active cores nearby, but we aren't sure of how the cores are mapped on the physical die, so this is just a theory.
In either case, the targeted core still doesn't reach the same peak speed we recorded with the fastest core, lagging by 50 MHz.
With the incrementally faster performance comes a shorter run time, as you can see from the length of the test.
Disabling All Fast Cores
Here we disabled all four cores marked as faster cores, in effect turning the processor into a dual-core. "Core 1" is targeted with the workload, but this isn't the same Core 1 we tested earlier. Because we disabled Core 1 and Core 2 for this test, this is actually physical Core 3. This core sustains an average speed of 4.3 GHz, but we spot periodic bursts of 4.325 GHz at the tail end of the run during the Cinebench sequence.
MORE: Best CPUs
MORE: Intel & AMD Processor Hierarchy
MORE: All CPUs Content