Resident expert and veteran overclocker Roman Hartung/ Der8auer conducted a survey with very interesting results. This is to determine how many of the AMD Ryzen 3000 series SKU’s advertised turbo clock speeds. This all started with him unable to reach the CPUs advertised boost clock speeds irrespective of the motherboard/ BIOS variants.
This analysis is done with 2,726 submissions, 40% with Ryzen 7 3700X CPUs. It also included 26% Ryzen 7 3900X submissions and 21% Ryzen 5 3600 CPUs. The rest were 6% Ryzen 5 3600X and 7% 3800X submissions. The submissions used numbers from single-core Cinebench R15 and monitored with HWInfo. AMD told him to use these to validate the observation and hence the end result. Its strange AMD would say this and expect a contrary analysis. Did even test the CPUs ability to reach advertised turbo clock speeds?
Der8auer also said that he limited the submissions until Sunday as he did not want to give the vendors an opportunity to figure what he was doing and potentially providing submissions which may affect the real-world observations. It also addressed many did not observe or report about this issue.
The results are very surprising and need to be addressed. Since the required information wasn’t coming from the usual channel, an issue which I also face, it had to be crowdsourced.
Challenges of relying on crowd-sourced information
Naturally, relying on a crowdsourced data would be dismissed by some (irrespective of their intentions) as it introduces certain inaccuracies or troll submissions from fanboys who are either for or against Ryzen 3000 CPUs. But Derbauer said that was able to identify and exclude them from testing, which was a very small and insignificant number. Regardless, dismissing the analysis on minor inaccuracies (which is screened out) is immature at best.
About the Ryzen-9 3900X
The 3900X’s analysis is interesting since it is the highest CPU spec you can buy for now, with Ryzen 9 3905X expected to come out ‘soon’. These analyses were also important as HardwareUnboxed said they were able to reach its 4.6 GHz advertised boost clock speed and base clock of 3.8GHz.
Contrary to Hardware Unbox’s claim, Derbauer said this is not depended on the motherboard, referring to the 3900X and the AORUS X570 Extreme. He was able to back this up with the survey information with 3900X and AORUS X570 Extreme submissions. Naturally, expecting everyone to pick up the AORUS X570 Extreme to run at advertised speed is unrealistic.
While the issue depends mostly on the AMD Ryzen 3000 series’ ability to run at its advertised boost clock speed, it also needs to factor in the motherboard’s AEGESA and SMU version. According to the information he received via the survey, while many X570 motherboards had newer AEGESA updates, it would be dependent on the SMU versions. He also addressed that it shouldn’t matter which Windows version was used as CPUs should reach its advertised boost clocks rates. If the end result affects specific Windows versions, AMD needs to specify the versions. Till date, it did not.
About the Ryzen-5 3600- the mid-end workhorse
The only CPU that had the best results to reach advertised boost clock speed was on the Ryzen-5 3600, with 49.8% of its submissions (270 out of 542).
Comments by Shamino
He further emphasized on what ASUS ROG R&D’s veteran and prolific extreme overclocker ‘Shamino’ said. He said that AMD may not have any idea to fix the issue, nor showed any indication/ intent to fix it. He speculates that if AMD has to push to reach those advertised boost clocks. it would draw very high VCORE and maybe damaging the CPU.
Shamino also addressed this concern at Overclock.net
Every new BIOS i get asked the boost question all over again, i have not tested a newer version of AGESA that changes the current state of 1003 boost, not even 1004. if i do know of changes, i will specifically state this. They were being too aggressive with the boost previously, the current boost behaviour is more in line with their confidence in long term reliability and i have not heard of any changes to this stance, tho i have heard of a ‘more customizable’ version in the future.
Lack of AMD’s involvement
Obviously, this is not an official statement from ASUS. But the point is very valid. AMD isn’t addressing or showing any signs of fixing it. This applies even with the results from software recommended by the chipmaker. It also doesn’t help when companies (especially chipmakers) that make enthusiast-centric components do not address enthusiast-centric concerns on time. Even if they do, it is mostly covered with marketing lingo. Contrary to their belief, that simply makes it worse. Wisdom can be extracted from its Bulldozer lawsuit.
This wouldn’t hamper the popularity or sales of the AMD Ryzen 3000 series CPUs. People don’t pay just for the base/boost clock speeds. But where do we start drawing the line? This argument is no different from the time Nvidia GTX 970’s issues with its VRAM distribution. We can’t compromise on the accuracy of advertised boost clock speed and dismiss it as insignificant. If we will keep dismissing every discrepancy as it comes up, as a whole would be a very large oversight. This also would apply to Intel and even Nvidia/AMD Radeon graphics cards. Nobody will appreciate gaslighting either, which creates a further gap in trust.
It is an understatement to say that Roman’s efforts in compiling this analysis and observations are very helpful. It represents what hardware enthusiast love to do.
— Hardware BBQ (@HardwareBBQ) September 3, 2019