Perception vs. Reality
I think the vast majority of enthusiasts understand that the Ryzen 3000 series can’t hit the advertised boost clock on all of the cores, and I think this is fine so long as AMD communicated this adequately. I think it has. Unfortunately, many users can’t get the advertised boost clocks on a single CPU core anywhere on the CPU. As I said, we have never seen the advertised boost clocks on any single AMD Ryzen 3000 CPU on any motherboard to date. With the first round of AGESA code updates and BIOS updates from MSI and ASUS, I have seen our sample CPU’s achieve boost clocks that are very close to the advertised boost clocks, but never have I seen the actual advertised boost clocks.
For example, Our Ryzen 9 3900X achieved a maximum clock speed of 4.575GHz on the MSI MEG X570 GODLIKE motherboard. This is close to 4.6GHz, but not quite 4.6GHz. Many people also point out that these CPU’s do not sustain the boost clocks for any length of time. In contrast, an Intel CPU will. Older AMD CPU’s would too and thus the perception is that the Ryzen 3000 series doesn’t do what AMD advertised.
Of course, some people believe that 4.575GHz is close enough and that 25MHz wouldn’t make much of a difference. Unfortunately, experiences vary substantially between individual configurations. I have seen one 3700X achieve nearly the correct boost clocks on one motherboard and not another. When it doesn’t boost much at all, you can sometimes change the RAM and it will behave differently. One 3700X I tested wouldn’t achieve more than 4.312-4.35GHz under any circumstances. Even clocking that high would sometimes require adjusting the base clock over 100MHz, which can create problems with the system. FCLK, SOC voltages and other values adjustable in BIOS don’t help either. The poor boosting behavior is something that can rear its ugly head with a BIOS update, or simply by altering your configuration slightly. There doesn’t appear to be anything beyond luck of the draw that makes a difference consistently.
For the most part, the benchmarks on one of these CPU’s that doesn’t boost correctly yields results within a margin of error compared to CPU’s that do boost correctly or close to it. So, even though the performance AMD advertised is there, the perception remains that these CPU’s do not work correctly. The actual reality is that none of these CPU’s sustain the maximum single-core boost clock for the duration of the test. As a result, the difference between a CPU that achieves the advertised value and one that doesn’t is so small, as to usually be statistically insignificant.
Multi-Threading
Under multi-threaded application loads and benchmarks, core clocking is generally consistent within a given model, and though AMD was clear about what to expect, the boxes for these CPU’s in no way indicate the all-core boost frequency leaving some disappointed that the maximum clocks listed for single-core boosting doesn’t apply to all the CPU cores. While these people are in the minority, they are vocal none the less. I don’t think that this group of people truly understands boost clocks and how they work. After all, while the Ryzen 3000 series does behave a little differently, Intel CPU’s haven’t worked that way either. I’m not sure where this unrealistic expectation comes from?
I think most enthusiasts understand that firmware, VRM design, motherboard quality, OEM UEFI BIOS implementation, cooling, and settings can adversely impact the results one might achieve in any overclock, even factory boost clock. However, users who go all out on custom water-cooling solutions with $700 motherboards and very expensive RAM seem just as likely to get a dud in the boost clocking department as anyone else. Manual overclocking doesn’t change anything either with many CPU’s being unable to sustain 4.2GHz on all cores.
Indeed, with custom cooling, neither Ryzen 7 3700X I tested could achieve 4.3GHz with stability. One CPU would do it some of the time and running the other at that speed resulted in a hard shut down of the system in less than 20 minutes. This wasn’t even in unrealistic benchmarking or arbitrary workloads that you wouldn’t encounter in the real world. This was experienced playing games in high-quality mode to leverage the GPU and so on. Again, the lack of rhyme or reason for this outside of simple luck with the silicon lottery is one major cause of the frustration for even ardent AMD fans.