Editorial: Boostgate: AMD’s Boost Clock Controversy

ASRock just released their official ABBA update yesterday. On my X570 Steel Legend, my 3600X is now boosting to the suggested boost frequency on all cores when they get hit with a single threaded workload. In BL3 last night HW Info showed all 6 cores as hitting 4.4GHz, when it would usually max at 4.3GHz before. Real world performance, as would be expected from 100MHz on a single core, was negligible to nothing at all in every test.
This was kind of my point... The article was posted at the end of last week, official released are already coming out... ill timed article. Congrats on getting the increased clocks! And as expected, not much in general on real world tests, thanks for the confirmation.
 
Do... you... I mean....

You realize it takes work and effort to gather the results in a repeatable scientific manner as to make them worthwhile and not just man on the street level data. I mean... it probably takes a week or more just to correlate the data, THEN write, edit, and publish an full blown article.

If you have some magic effing wand to do this more power to you. But get over yourself otherwise. Their results were from DAYS before the leak happened and they published the article they had on hand so as to NOT waste their effort. Just because it doesn't include the beta reviews you wan't doesn't make it a bad article.

Wow... just... wow...
It has little to do with not including beta bios... The fact that betas where out meant we would probably be seeing officials soon, which is now confirmed. As I said, this article would have made more sense 2 weeks earlier or 2 weeks later. I know articles take some time, but this was discussed weeks ago and AMD already made an official announcement, and beta bioses hitting the streets. It's not a bad article per se, just not a very well timed article. If you wanted to wait this long, might as well pushed it another week and had factual data/reports to go by.
 
I think the real issue (and still really not an issue) at this point is going to be, when and/or is PBO going to work on 3000 series. I haven't seen anything stating that the ABBA revision is going to address PBO, and only addresses the outcry of not getting 25-100MHz that wasn't even guaranteed; though in all likelihood I think PBO is just not going to work on this series as they already seem to be getting the max frequencies just at default settings. There is still some work to be done, but I think they'll get it all sorted out in the coming weeks/months. Living on the bleeding edge of new technology always comes with issues.
 
Says article was released on the 12th... A few days after leaked bioses started coming out. Just seems this close to knowing for sure if the fix will work or not as an odd time to post this type of article. Had this been posted before AMD said they had a fix incoming would have made a lot more sense. I agree there are still a lot of unknowns and this editorial does nothing to fill those in. I will happily read the updates once the official released bioses come out (good or bad), but this editorial just felt like it was either late to the party (as in it could have been at the end of last month and made sense) or to early (no new info, knowing a 'fix' is due and leaked bioses are ready showing up) to be useful.

It was written a few days ahead of the release. I do not control publishing scheduling. I think you miss the point of an editorial.

Do... you... I mean....

You realize it takes work and effort to gather the results in a repeatable scientific manner as to make them worthwhile and not just man on the street level data. I mean... it probably takes a week or more just to correlate the data, THEN write, edit, and publish an full blown article.

If you have some magic effing wand to do this more power to you. But get over yourself otherwise. Their results were from DAYS before the leak happened and they published the article they had on hand so as to NOT waste their effort. Just because it doesn't include the beta reviews you wan't doesn't make it a bad article.

Wow... just... wow...

It does. A motherboard review is upwards of 30-35 hours of work for me. It takes someone else a couple of hours to edit the article after that. I actually completed a motherboard review using the updated BIOS which hasn't been published yet. So far, the fix didn't work for the 3900X I have, but it did alter performance slightly. That said it's still the leaked fix, not an official release. I don't have official release's for the motherboards I have on hand. I don't think comprehensive benchmarking will be something I'll do on this topic. The data will come out given enough time in the regular reviews as I do more articles with updated BIOS / AGESA code versions. You'll be able to see any changes in the results for yourself.

CPU reviews are a bit quicker, and I have more coming. So, I don't know how I'll handle it at this point. You will see the updated data, but I don't know how much more time I'll spend on this issue.
 
It was written a few days ahead of the release. I do not control publishing scheduling. I think you miss the point of an editorial.



It does. A motherboard review is upwards of 30-35 hours of work for me. It takes someone else a couple of hours to edit the article after that. I actually completed a motherboard review using the updated BIOS which hasn't been published yet. So far, the fix didn't work for the 3900X I have, but it did alter performance slightly. That said it's still the leaked fix, not an official release. I don't have official release's for the motherboards I have on hand. I don't think comprehensive benchmarking will be something I'll do on this topic. The data will come out given enough time in the regular reviews as I do more articles with updated BIOS / AGESA code versions. You'll be able to see any changes in the results for yourself.

CPU reviews are a bit quicker, and I have more coming. So, I don't know how I'll handle it at this point. You will see the updated data, but I don't know how much more time I'll spend on this issue.
I can appreciate that, sorry for coming down so hard, just seemed really odd timed since most of the news/complaints came out a few weeks back and being so close to getting the official release. I agree with your assessment and assumed (like most I think) there wouldn't be huge real world gains.
 
Thanks Dan for your hard work and effort. Good report in a timely fashion.
 
On the topic of PBO and boost clocks. On my X570 Steel Legend with the official ABBA release, I tested by turning on PBO last night and setting it to +200MHz; this resulted in going back to the previous behavior of not reaching the 4.4GHz boost clock and starting working at 4.2-4.275GHz. I had to do a UEFI default reset to get the clocks back to the proper boost clock levels at 4.4GHz on single threads and by leaving all the CPU settings at their defaults.
 
Says article was released on the 12th... A few days after leaked bioses started coming out. Just seems this close to knowing for sure if the fix will work or not as an odd time to post this type of article. Had this been posted before AMD said they had a fix incoming would have made a lot more sense. I agree there are still a lot of unknowns and this editorial does nothing to fill those in. I will happily read the updates once the official released bioses come out (good or bad), but this editorial just felt like it was either late to the party (as in it could have been at the end of last month and made sense) or to early (no new info, knowing a 'fix' is due and leaked bioses are ready showing up) to be useful.

Go back to Youtube.
 
On the topic of PBO and boost clocks. On my X570 Steel Legend with the official ABBA release, I tested by turning on PBO last night and setting it to +200MHz; this resulted in going back to the previous behavior of not reaching the 4.4GHz boost clock and starting working at 4.2-4.275GHz. I had to do a UEFI default reset to get the clocks back to the proper boost clock levels at 4.4GHz on single threads and by leaving all the CPU settings at their defaults.

This is what I've experienced with the 3900X on multiple boards.
 
This is what I've experienced with the 3900X on multiple boards.
So, maybe the next question is, what's wrong with PBO? Used to work well on 1k and 2k ryzen, now with pushing everything so close to the brink maybe pbo doesn't have any room to work?
 
It's very likely that, yes, PBO just has no headroom. On a single core we're already seeing 1.5v+ (my 3600X hits 1.469-1.475 to reach its 4.4GHz) to hit boost clocks (have seen a few screenshots exceeding 1.5v also), it's likely there just isn't a satisfactory trade off to get any extra frequency, and the safe voltages for all core loads are touted to be around 1.37v max.

Temps seem to be pretty limiting this gen as well; my 3600X during gaming sees 84-85C and my setup goes as such: NZXT H700i w/ EK 3.2 DDC elite pump with 140ml reservoir -> out to Radeon VII -> out to XSPC EX360 radiator with 3x Noctua NF-A12x25PWM -> out to EK supremacy block on the 3600X -> out to XSPC TX360 radiator also with the NF-A12x25's -> back to pump. My fan and pump PWM profiles are pretty aggressive as well, fans are at 2000rpm with CPU temp above 65c and pump is at max rpm (4400-4500) above 70c.

Regardless, AMD has done really well with IPC for this gen. Seeing some reviews with clocks matched, Intel can't keep up, that overclocking headroom is the only thing they have to save them, and that only really works in certain scenarios. It's a good time for CPU innovation and evolving; now we just need that to come back to the GPU front.
 
Last edited:
It's very likely that, yes, PBO just has no headroom. On a single core we're already seeing 1.5v+ (my 3600X hits 1.469-1.475 to reach its 4.4GHz) to hit boost clocks (have seen a few screenshots exceeding 1.5v also), it's likely there just isn't a satisfactory trade off to get any extra frequency, and the safe voltages for all core loads are touted to be around 1.37v max.

Temps seem to be pretty limiting this gen as well; my 3600X during gaming sees 84-85C and my setup goes as such: NZXT H700i w/ EK 3.2 DDC elite pump with 140ml reservoir -> out to Radeon VII -> out to XSPC EX360 radiator with 3x Noctua NF-A12x25PWM -> out to EK supremacy block on the 3600X -> out to XSPC TX360 radiator also with the NF-A12x25's -> back to pump. My fan and pump PWM profiles are pretty aggressive as well, fans are at 2000rpm with CPU temp above 65c and pump is at max rpm (4400-4500) above 70c.

Regardless, AMD has done really well with IPC for this gen. Seeing some reviews with clocks matched, Intel can't keep up, that overclocking headroom is the only thing they have to save them, and that only really works in certain scenarios. It's a good time for CPU innovation and evolving; now we just need that to come back to the GPU front.
I just got a great deal on a b450 gaming itx and a 1600, put it in a small case with water... Thinking it may be time to upgrade the CPU to a 3600 though (I put it together this last weekend with a fury nano, lol).
 
the safe voltages for all core loads are touted to be around 1.37v max.
Preferrably not above 1.325v if you don't want to run into very quick degradation. : - )
 
So, maybe the next question is, what's wrong with PBO? Used to work well on 1k and 2k ryzen, now with pushing everything so close to the brink maybe pbo doesn't have any room to work?

This is exactly the problem.
 
This is exactly the problem.
I mean, in my mind that's good, they are trying to wring everything out of the CPU by default. I'm seeing less reason to have a regular and X model though as there isn't much differenciating them.
 
Preferrably not above 1.325v if you don't want to run into very quick degradation. : - )

It seems this new ABBA BIOS is trying to adhere to that 1.325v number on all-core loads. Before ABBA mine would sit at 1.365-1.37v, and after ABBA it now does 1.325-1.33 on all-core loads.
 
It seems this new ABBA BIOS is trying to adhere to that 1.325v number on all-core loads. Before ABBA mine would sit at 1.365-1.37v, and after ABBA it now does 1.325-1.33 on all-core loads.
That is a good thing, as it doesn't seem like any loss of performance, so lower voltages seems like an added benefit.
 
Become a Patron!
Back
Top