F5F Stay Refreshed Hardware Desktop MSI X570S Carbon Max WiFi PBO "Motherboard" problems with 5900X setup.

MSI X570S Carbon Max WiFi PBO "Motherboard" problems with 5900X setup.

MSI X570S Carbon Max WiFi PBO "Motherboard" problems with 5900X setup.

B
BestkingJoris
Member
Find
213
10-31-2018, 09:49 PM
#1
Hi folks, This is an ongoing problem I've been having since I purchased the board and it's not anything serious, it still gnaws at me. I come at it every now and then to see if I can track down the cause, and here I am now, posing the question here. When I bought my board in January 2024, I started out on BIOS 7D52v18 and my temps were always capped at 65-66 max on my 5900X. My BIOS is configured for XMP (DOCP) and PBO using Auto on most settings ("Motherboard" PBO limits) and a -25 CO. I use an AIO but even that if I run Cinebench R23 and 2024, I should be hitting the early 80s. My previous board, an ASUS ROG Strix B550-E Gaming, never had such limits with the same PBO settings but I switched to the MSI board due to other unrelated issues with connectivity. I say "limits" because something is causing the package power to cap out at ~165W, according to both HWMonitor and HWiNFO. When I flashed 7D52v19, the same problem persisted. It didn't bother me a whole lot as I was getting most of the performance and energy savings with minimal performance loss are never a terrible thing! But hey, I'm the kind of guy who would run a nuclear reactor to get the full amount of performance available if it were an option! At one point after flashing 7D52v19, I was playing around in the BIOS with the various PBO settings and I did something, but I don't know what, to get the CPU to draw more power and boost higher, and of course run hotter - with the expected results in Cinebench R23 and 2024, pulling 200-210W package power, just like on the ASUS. So it seemed that whatever was limiting the CPU was no longer doing so. The problem is, I don't know what change, if any, caused the so-called limits to go, because I don't recall doing anything special. But there it was - working as expected, and at the time I just assumed setting the PBO limits from "Motherboard" to "Auto" the fix. But when I reset the CMOS and flashed 7D52v1A, then 7D52v1B1, and now 7D52v1B, it's back with the same limits as before, and that even goes for flashing 7D52v19 again. I tried everything I could think of, from manually setting the PBO limits to changing the load-line modes, and nothing I did made a lick of difference. I tried resetting the CMOS again and going with a bare minimal set of customizations; still nada. If I slap the CPU back into my ASUS, it's completely fine, so I'm confident nothing is wrong there and I'm pretty darn sure it isn't anything in Windows, as I switched the power plan to High Performance and that also made no discernible difference. However, I did discover that using a more realistic manual EDC limit, like 190-200A, the CPU would work closer to its potential and what I had on my previous ASUS board. It's just weird because using the "Motherboard" settings has never done this before I got this board. When I came at the problem again yesterday, I took a few screenshots from Ryzen Master (which I installed just for the purposes of monitoring the PBO limits; I do not have it installed day-to-day). This is with the default "Motherboard" limits set. Notice the power draw (also confirmed in HWMonitor, etc., to be ~160W), the clock speeds lower than what the chip is capable of with the limits as high as they are, and the EDC only drawing ~103A. If I set the limits to manual and tighten up the EDC down to the 190s, I'm getting the expected results. The TDC is drawing ~130A and the power draw is ~204W. It's also the same if I dial in the "Motherboard" limit values for the TDC and PPT but dial the EDC back to 190, which is what RM indicates is being used when the explicit "Motherboard" limits are set. It just doesn't make sense that setting a higher EDC would cap the draw and performance of the CPU. Oh, and here is a screenshot of my PBO settings. Nothing fancy about them. Has anyone seen this before, and is it normal and the ASUS is the exception to the rule? Thank you in advance for any insights.
Reply
Reply
B
BestkingJoris
10-31-2018, 09:49 PM #1

Hi folks, This is an ongoing problem I've been having since I purchased the board and it's not anything serious, it still gnaws at me. I come at it every now and then to see if I can track down the cause, and here I am now, posing the question here. When I bought my board in January 2024, I started out on BIOS 7D52v18 and my temps were always capped at 65-66 max on my 5900X. My BIOS is configured for XMP (DOCP) and PBO using Auto on most settings ("Motherboard" PBO limits) and a -25 CO. I use an AIO but even that if I run Cinebench R23 and 2024, I should be hitting the early 80s. My previous board, an ASUS ROG Strix B550-E Gaming, never had such limits with the same PBO settings but I switched to the MSI board due to other unrelated issues with connectivity. I say "limits" because something is causing the package power to cap out at ~165W, according to both HWMonitor and HWiNFO. When I flashed 7D52v19, the same problem persisted. It didn't bother me a whole lot as I was getting most of the performance and energy savings with minimal performance loss are never a terrible thing! But hey, I'm the kind of guy who would run a nuclear reactor to get the full amount of performance available if it were an option! At one point after flashing 7D52v19, I was playing around in the BIOS with the various PBO settings and I did something, but I don't know what, to get the CPU to draw more power and boost higher, and of course run hotter - with the expected results in Cinebench R23 and 2024, pulling 200-210W package power, just like on the ASUS. So it seemed that whatever was limiting the CPU was no longer doing so. The problem is, I don't know what change, if any, caused the so-called limits to go, because I don't recall doing anything special. But there it was - working as expected, and at the time I just assumed setting the PBO limits from "Motherboard" to "Auto" the fix. But when I reset the CMOS and flashed 7D52v1A, then 7D52v1B1, and now 7D52v1B, it's back with the same limits as before, and that even goes for flashing 7D52v19 again. I tried everything I could think of, from manually setting the PBO limits to changing the load-line modes, and nothing I did made a lick of difference. I tried resetting the CMOS again and going with a bare minimal set of customizations; still nada. If I slap the CPU back into my ASUS, it's completely fine, so I'm confident nothing is wrong there and I'm pretty darn sure it isn't anything in Windows, as I switched the power plan to High Performance and that also made no discernible difference. However, I did discover that using a more realistic manual EDC limit, like 190-200A, the CPU would work closer to its potential and what I had on my previous ASUS board. It's just weird because using the "Motherboard" settings has never done this before I got this board. When I came at the problem again yesterday, I took a few screenshots from Ryzen Master (which I installed just for the purposes of monitoring the PBO limits; I do not have it installed day-to-day). This is with the default "Motherboard" limits set. Notice the power draw (also confirmed in HWMonitor, etc., to be ~160W), the clock speeds lower than what the chip is capable of with the limits as high as they are, and the EDC only drawing ~103A. If I set the limits to manual and tighten up the EDC down to the 190s, I'm getting the expected results. The TDC is drawing ~130A and the power draw is ~204W. It's also the same if I dial in the "Motherboard" limit values for the TDC and PPT but dial the EDC back to 190, which is what RM indicates is being used when the explicit "Motherboard" limits are set. It just doesn't make sense that setting a higher EDC would cap the draw and performance of the CPU. Oh, and here is a screenshot of my PBO settings. Nothing fancy about them. Has anyone seen this before, and is it normal and the ASUS is the exception to the rule? Thank you in advance for any insights.

Reply
Reply
A
AthenasLight
Posting Freak
Find
781
10-31-2018, 09:49 PM
#2
It's common across many boards, including ASUS ones. Some handle PBO better than others—my X570 Master does worse when PBO is enabled on defaults compared to others, though all boards can show these issues. Certain AGESA versions seem to perform better, which could explain what you're experiencing, but most boards tend to have overly strict limits. The problem lies with PBO itself, not the motherboard. From what I understand, AMD hasn't addressed it even in AM5, and since I haven't tuned AM5 PBO as much as AM4, my perspective is limited.
Reply
Reply
A
AthenasLight
10-31-2018, 09:49 PM #2

It's common across many boards, including ASUS ones. Some handle PBO better than others—my X570 Master does worse when PBO is enabled on defaults compared to others, though all boards can show these issues. Certain AGESA versions seem to perform better, which could explain what you're experiencing, but most boards tend to have overly strict limits. The problem lies with PBO itself, not the motherboard. From what I understand, AMD hasn't addressed it even in AM5, and since I haven't tuned AM5 PBO as much as AM4, my perspective is limited.

Reply
Reply