My single core scores kinda suck. Just got 609 in CBR20. Best I've seen was about 614. I have to wonder if it's the 3200MHz memory or just a bad chip for single core performance, or maybe my board. It is an old X470, Asus x470-i. For me, I'm less concerned about games, I'm really looking for that additional multithread boost for Visual Studio compilation times. Only thing I have that I care about. As long as I don't lose low power idle which is the majority of the time for my system, I don't mind.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
5900x, running hot
- Thread starter tribunal3117
- Start date
Dunno about the single core scores, they seem to vary quite a bit. Do run twice in a row to get representative scores as there seems to be a slight "startup" time that will take a few points away. Phase settings for power is another that can influence. Tried "ASUS Optimized", but got lower performance in general so went back to auto, which is CPU decides. Generally for single core I get the highest scores with D.O.C.P. + TRC and everything else on auto. Your memory should do fine for the test though, but there could be other factors. Bios could be another factor as well and maybe chipset. It is possible that the higher end x570 boards are just a bit faster. Being 2x16GB they are most likely dual rank which can also influence performance scores. Dual rank is preferred on zen3 (either 2 dual rank or 4 single rank which becomes 2 ranks per channel).
Noticed that your VRMs hit 84 when running with PBO on. Was that during a single run or a long duration test? If it was from a single run then it is possible that they overheat when running for extended periods. Asus has been caught underrepporting the VRM temps on a few boards by 10 degrees and I'm pretty sure mine is among them, don't know if yours is as well.
I'm not sure in that case if it was a single run or not. I try to run the multicore test twice, the single core I only run once in a while and just one time, since it takes so long. Just now, I ran a multicore CBR20 test and VRM temps went from 38C to 74C on a single run. While my CPU has a D15 on it, it does seem questionable that motherboard VRMs are supposed to get that hot. This system may be a good candidate for leaving PBO off.
Just did about 5 multicore tests in a row and pulled the voltages and saw another VRM temp reading, this one says 90C.
Here was a test done on similar VRM to mine, the Asus Prime X470-Pro. From https://www.reddit.com/r/Amd/comments/8o8tbx/asus_prime_x470pro_in_the_boxcase_vrm_temperatures/
Here was a test done on similar VRM to mine, the Asus Prime X470-Pro. From https://www.reddit.com/r/Amd/comments/8o8tbx/asus_prime_x470pro_in_the_boxcase_vrm_temperatures/
Last edited:
Under advanced benchmark you can set minimum test duration. Set it to 700 seconds or so and it will automatically do two single core runs while you do something else 
The VRMs can be difficult to cool as they are blocked by both the RAM and partially by coolers for front to back airflow. VRMs are usually good for well 125 or more degrees component wise, but I believe most motherboards throttle at 105 or less. Some might even shut down for safety.
The VRMs can be difficult to cool as they are blocked by both the RAM and partially by coolers for front to back airflow. VRMs are usually good for well 125 or more degrees component wise, but I believe most motherboards throttle at 105 or less. Some might even shut down for safety.
Disabling all the PBO, I just got my best single core score yet of 619.
PBO on-
PBO off-
It's fun to play around with. What you were saying before is true though, basically that you don't get a massive amount of performance gain out of PBO, for an awful lot of heat and power. AMD has their standard PB tuned correctly out of the box to get the most out of these chips within reason. For a sustained load like 8-thread Steam Link streaming in-home like I do everyday (also has the game on top of that), it seems best to not risk pushing for the relatively small amount of additional performance out of it, risking instability and more wear and tear.
PBO on-
PBO off-
It's fun to play around with. What you were saying before is true though, basically that you don't get a massive amount of performance gain out of PBO, for an awful lot of heat and power. AMD has their standard PB tuned correctly out of the box to get the most out of these chips within reason. For a sustained load like 8-thread Steam Link streaming in-home like I do everyday (also has the game on top of that), it seems best to not risk pushing for the relatively small amount of additional performance out of it, risking instability and more wear and tear.
Last edited:
If the hwinfo from the second screenshot is from the multicore run, then it seems like it isn't boosting multicore at the same rate as the fullsize x570 boards. Mine boosts to an average of 139 to 140 watts and about 142 on peak in cinebench. If the board limits normal boost to 125w then comparing multicore numbers to mine and others who are boosting up to 142w becomes apples and oranges.
Do keep in mind that the monitoring software takes CPU cycles so probable a few more points to be had Benchmarking is fun, but the difference between 619 and the highest single core numbers found for the 5900x and 5950x is generally not noticeable in real world use.
Do keep in mind that the monitoring software takes CPU cycles so probable a few more points to be had
Benchmarking is fun, but the difference between 619 and the highest single core numbers found for the 5900x and 5950x is generally not noticeable in real world use.Jamie Marsala
Limp Gawd
- Joined
- Mar 9, 2016
- Messages
- 293
My single cores are that way as well with PBO enabled and even using curve optimizer, but my multi core is great. If I disable PBO then I get better single core and low multi core, compared to what the reviewers got. With PBO tweaked though and curve optimizer at -10 all core I do see two cores hit over 5Ghz in HWInfo64 and all cores on CCD0 all hit over 4.875Ghz, all but one hit 4.925, and CCD1 all hit at least 4.750.
My understanding was that 125W was the default limit for these CPUs from AMD. That's odd that anything is boosting to 140W if CPU limits are enforced. Maybe it's a way to encourage people to move to X570. I would've, but counted myself lucky to not have the chipset fan. I'll upgrade once there's a mITX board with no chipset fan that supports my soon to be old DDR4 and 5900X.If the hwinfo from the second screenshot is from the multicore run, then it seems like it isn't boosting multicore at the same rate as the fullsize x570 boards. Mine boosts to an average of 139 to 140 watts and about 142 on peak in cinebench. If the board limits normal boost to 125w then comparing multicore numbers to mine and others who are boosting up to 142w becomes apples and oranges.
Do keep in mind that the monitoring software takes CPU cycles so probable a few more points to be had
For zen 3 it is 142 watts PB limit for x570 motherboards and afaik some others. Just pointed it out so you would know the reason for a lot of the multicore difference you see vs those on full size x570 boards. I do believe there are som settings for increasing max watt, without going full PBO though. Increasing to 110% would give the same watts and your boards should handle 142 just fine. It is highly unlikely that there will be a passively cooled ITX board with x570 chipset as there are very few with passive heatsinks and those heatsinks are large. Keep in mind that you would probably gain 5-7% at most on multicore and probably little or no gain on single core for the cost of an entire new board and premium ITX boards are expensive. Not really worth upgrading the board IMO as zen 3 is supposed to be the last CPU on AM4 and the strix x470-i is a really nice ITX board. I do have one in my fanless living room PC and the amount of stuff they managed to fit into such a small package is mind blowing.
The only way I'd get a newer board is if the chipset fan was 60mm so I could put or replace any dead fan with a standard Noctua. Then it wouldn't be a big deal. I just didn't want to fuss with a hard to get to, mini or proprietary fan. The 140W default on X570 must just be due to raising the floor across the board for quality standards. I don't mind being stuck at 125W. I might poke around to increase it to 140 but with most things, I lean towards using things as the engineers intended. It's like cutting the shocks on a Honda Civic so it rides lower, pretty much ruins a car. I've done my share of overclocking over the years, but PBO is just easy and designed to work as it does by AMD so you can mostly rely on it. As I've learned here and in this thread, they already got the good out of these with basic Precision Boost though. Squeezing that extra out at these costs is questionable. Fun, but questionable.
I tried increasing this but no change in reported package power consumption in hwinfo from 125W, no multicore gain in cbr20 either. Maybe it's the wrong test (uses avx I believe?). Either way, I would've expected to see something change and it didn't. I think it would require enabling PBO alongside it.
I've been doing some searching on peoples results, and 650/8869 was done here on a 5900X. https://albertherd.com/2020/12/13/my-experience-with-precision-boost-overdrive-2-on-a-5900x/
Still though, crunching the numbers, jumping up 30 singlecore and 1000 multicore points in CBR20, if that same improvement applies to all tasks, really is out of proportion with the heat/power from the PBO settings required to get there. I do like the idea of a 650 single core result though.
soulesschild
Supreme [H]ardness
- Joined
- Feb 18, 2007
- Messages
- 6,152
I have to say that I finally got my first WHEA crash after like...4 weeks of using my machine. So frustrating that it's hard/impossible to suss out instability without just waiting for it to happen. At least I can easily identify which core crashed and tweak it.
My PBO crash took a long time to occur as well. Due to the nature of how PBO works, high load testing ie. Prime95 isn't enough. 1000 +/- 300 points in CBR20 multicore it doesn't really make sense to me, people are netting ~14% multicore improvements for the chance of instability and mostly, and removing the ability to run it at 100% nonstop if the need ever arised. Mine does stabilize at 80-85C without throttling, but nearly everything in the system is pushed to the limit for the equivalent of that 1000 point gain.
soulesschild
Supreme [H]ardness
- Joined
- Feb 18, 2007
- Messages
- 6,152
Yeah for me it was an idle/low power crash. And it only happened after like...if I read my logs right, after about 11 hours of idling. Basically I turned my computer on during the day and just left it alone until I woke it up from sleep when I got back to it. But testing like just waiting an hour or two, no crash.
If the hwinfo from the second screenshot is from the multicore run, then it seems like it isn't boosting multicore at the same rate as the fullsize x570 boards. Mine boosts to an average of 139 to 140 watts and about 142 on peak in cinebench. If the board limits normal boost to 125w then comparing multicore numbers to mine and others who are boosting up to 142w becomes apples and oranges.
Do keep in mind that the monitoring software takes CPU cycles so probable a few more points to be had
The answer to the 125W issue is that EDC/TDC limits are being hit first, when not enabling PBO. With PBO enabled, my scores are basically in line with everyone else, at stock I'm lagging a bit, by about 400 points in CBR20.
Attachments
-
1615074422153.png
Cinebench is somewhat "unstable" in multicore score. I seem to be getting about 8470 with a deviance of +-100 points. Not sure if the TDC and EDC are set by motherboard manufacturerer or if it is in the CPU. My guess is that you would see scores between 8200 and 8400 with 142w limit. Scores are not 100% consistent, even a very low CPU background task is enough to offset it quite a few points. Added the max I've seen on stock for comparison, but I can not achieve the max with consistency, usually it is something like 8470 to 8540 unless I shut down a lot of background tasks and all monitoring.
Another interesting thing is that the boost clock goes higher when I keep it in the mid 50s (cooling at max) vs very low noise where it will sit at aprox 60 degrees with ambient of aprox 23 degrees. E.g. I get max boost of 4300 when in the low 60s and lower average compared to keeping it in the mid 50s.
I did run it looped to get accurate readings while still running so averages are correct, but as you can see I am hitting PPT, TDC and EDC equally. The CPU is a silver sample with quite uneeven CCDs according ryzen clock tuner, so those who lucked out and got a golden sample will probably get even better results.
Another interesting thing is that the boost clock goes higher when I keep it in the mid 50s (cooling at max) vs very low noise where it will sit at aprox 60 degrees with ambient of aprox 23 degrees. E.g. I get max boost of 4300 when in the low 60s and lower average compared to keeping it in the mid 50s.
I did run it looped to get accurate readings while still running so averages are correct, but as you can see I am hitting PPT, TDC and EDC equally. The CPU is a silver sample with quite uneeven CCDs according ryzen clock tuner, so those who lucked out and got a golden sample will probably get even better results.
Last edited:
Looked through both of our monitorings and the current is the same and we are both hitting the same 97A limit for the CPU core. TDC is long term (sustained) EDC is short term (peak) limits. The difference is that my CPU runs higher voltage, not sure why. Do you have the latest bios for your board? It came out late january with AGESA 1.2.0.0.
Btw. are you running balanced windows powerplan or higher and latest chipset drivers? Posted a ryzen master image for comparison. We have the same current limits, but for some reason my voltages are significantly higher.
Btw. are you running balanced windows powerplan or higher and latest chipset drivers? Posted a ryzen master image for comparison. We have the same current limits, but for some reason my voltages are significantly higher.
Last edited:
I'm on the AMD chipset software from 2/4 (2.13.27.501). I think it's unlikely, but I wonder if a format would make any difference at all. What's your Power Reporting Deviation in hwinfo? Mine is sky high, 200% just in normal desktop use. I think there's some sort of "performance" setting that Asus has in my board, that's actually hurting performance. I'm going to play with some BIOS settings, and will report back.
Also, the chipset software no longer installs any special power plan, which I thought they removed with Zen2 and onward. I don't get it at least.
Changing this setting allowed it to go to 142W.
But then my CBR20 performance overall was cut in half. I think Asus has some bugs or did too little testing on my board. I did a reset to optimized defaults and tested, still the 125% overreport / 125W limit.
I do think a combination of setting CPU Core Voltage to manual and disabling (or enabling) other Asus options may get it to where it's supposed to be. I may try an older BIOS.
But then my CBR20 performance overall was cut in half. I think Asus has some bugs or did too little testing on my board. I did a reset to optimized defaults and tested, still the 125% overreport / 125W limit.
I do think a combination of setting CPU Core Voltage to manual and disabling (or enabling) other Asus options may get it to where it's supposed to be. I may try an older BIOS.
That being your stock result, is what I'm getting with PBO enabled. Last two runs were 7588 (PBO off), and 8456 (PBO on). Yes I am on AGESA 1.2.0.0. The only other non-beta bios available to me was 1.0.8.0 which worked great, in fact it seemed better in some ways, but I don't think I ever ran CBR20 on it.
What I think I need to do is adjust the "power telemetry" option in my BIOS but I can't find it. Leaving everything relevant to stock settings from Asus-optimized and changing that value should bring me up to where I need to be. Hard to say where the bug is.
I just got 7588 again now and took a screenshot of hwinfo.. same old 125W/130% over report on power.
Seems kindo of strange. My bios is pretty much defaults, except for memory, fan curves and boot drives. It does show 1.42-1.47 or something like that when I am in bios and looking at CPU voltage with it set to auto. Afaik you wouldn't get full PBO unless your motherboard was considered powerful enough to handle 142 watts. You might be able to get an answer in the Asus ROG forums, they have a separate section for 400 and 500 series motherboards there.
It is strange. But I'm hitting over 200W with PBO, and with PBO disabled I'm hitting 142W with the manual setting (just halves my CBR20 score). 200W has to indicate the board is "powerful enough"? My bios changes are also as light touch (mem/fans/wake on LAN), I set defaults to be sure I never hit a setting that restricted anything. I'm posting around about it, but you don't find nearly as many x470 (and even less Asus x470-i users) that upgraded to Zen3.
I think you're right on the Asus ROG forums, that would be a small enough community to get a good answer from.
Hope you figure it out, really interested to see what kind of scores you get when running full "stock".
I did some tests to figure out what kind of impact cooling has when running PBO. Hadn't run PBO on after adjusting TRC either. It seems like it will boost quite a bit higher when keeping the temps down. E.g. the current is fixed, but the voltage can go higher. Score is after pre-chilling the coolant so it was running at around 71 degrees during the scored run. Temps below is after about 7 minutes of running in a loop so probably 1 degree or so away from stable state. It boosts quite a bit higher when I am able to keep it in the low to mid 70s than before I made changes to my cooling and maxed out at around 78 degrees with full fan speed. Still not seeing amazing gains though with aprox 6% score for 50% increase in average watts.
I did some tests to figure out what kind of impact cooling has when running PBO. Hadn't run PBO on after adjusting TRC either. It seems like it will boost quite a bit higher when keeping the temps down. E.g. the current is fixed, but the voltage can go higher. Score is after pre-chilling the coolant so it was running at around 71 degrees during the scored run. Temps below is after about 7 minutes of running in a loop so probably 1 degree or so away from stable state. It boosts quite a bit higher when I am able to keep it in the low to mid 70s than before I made changes to my cooling and maxed out at around 78 degrees with full fan speed. Still not seeing amazing gains though with aprox 6% score for 50% increase in average watts.
Btw. probably forgot to mention that I did have Corsair h150i PRO XT as a temporary solution while I tried to decide what to do for cooling. That is the non-rbg equivalent of the Corsair 360mm capellix AIO. That would average 195w and around 86-87 degrees with PBO on so your air cooler is doing very well, even with PBO on
I'd go big air even if it were significantly worse. I want to be able to never touch it again. With a graphite thermal pad, all I'll have to do is check if the screws are tight on the heatsink.. some year down the road when I have it apart and it comes to mind. I don't think I've seen mine go above 85C, ever. That's with PBO on. PBO off, of course my board isn't running "full stock", but it would clearly be overkill for it.Btw. probably forgot to mention that I did have Corsair h150i PRO XT as a temporary solution while I tried to decide what to do for cooling. That is the non-rbg equivalent of the Corsair 360mm capellix AIO. That would average 195w and around 86-87 degrees with PBO on so your air cooler is doing very well, even with PBO on
I'm going to keep posting around seeing if anyone knows anything. I'm the only person posting publicly anywhere that's using the x470-i on Zen3. It's in tier D in LTT's (famous?) motherboard VRM list. Which says that it'll support 105W TDP CPUs at stock with good cooling. https://linustechtips.com/topic/113...k2YDc5dRY0YLN63ig1qc1T9r7FnHRhHvTHHrQ5PPNveRY
Your SOC is 20W vs mine is 4.5W as well.
The main difference with regards to the CPU is that mine is pulling about 7 watts more. Not sure why you or how you get that low SOC reading. My readings are about normal for the SOC. Only wondering if ASUS did put in place voltage limiters at certain current limits on some boards at stock.
A tier D rating should be fine for stock, but probably not enough for sustained PBO. I am running a board from the S tier, but am still about uncertain if I am comfortable with PBO on at all times. More worried about long term CPU wear with AVX etc. than the motherboard though. If the gains had been better then I might have run a PB and manual OC hybrid (PB until certain amp limit). But at most I could pretty much keep my single core and gain 10% in multicore which I would rarely benefit from as I almost never fully load 12 cores over long periods of time.
You do create kind of a wind tunnel in your case which amplifies the cooling capabilities. If we put your CPU, motherboard and cooler in my case (fairly large fulltower) and put 3 140mm fans on inntake and 1 on exhaust then the temps would prabably be 3-7 degrees worse for the CPU due to distances. I did pick the case due to having room for a lot of cooling and easy to install huge graphics cards.
Do keep in mind that I almost always pre-soak my system when doing thermal tests, e.g. run for 5-15 minutes before reading thermals depending on the cooler. I can keep the CPU at 71 or so degrees on a single cinebench PBO run but temps go up around 4 degrees after reaching stable state when on a loop. Would be same for the AIO. On a single cinebench PBO run the AIO temps would be 78 or so, but pre-soaking for 10 minutes would increase temps about 8 degrees. Every CPU is different (even if same model) so comparing coolers can only be done on the exact same parts. A great sample might run 10-15 degrees cooler than a bad sample on the same cooler.
A tier D rating should be fine for stock, but probably not enough for sustained PBO. I am running a board from the S tier, but am still about uncertain if I am comfortable with PBO on at all times. More worried about long term CPU wear with AVX etc. than the motherboard though. If the gains had been better then I might have run a PB and manual OC hybrid (PB until certain amp limit). But at most I could pretty much keep my single core and gain 10% in multicore which I would rarely benefit from as I almost never fully load 12 cores over long periods of time.
You do create kind of a wind tunnel in your case which amplifies the cooling capabilities. If we put your CPU, motherboard and cooler in my case (fairly large fulltower) and put 3 140mm fans on inntake and 1 on exhaust then the temps would prabably be 3-7 degrees worse for the CPU due to distances. I did pick the case due to having room for a lot of cooling and easy to install huge graphics cards.
Do keep in mind that I almost always pre-soak my system when doing thermal tests, e.g. run for 5-15 minutes before reading thermals depending on the cooler. I can keep the CPU at 71 or so degrees on a single cinebench PBO run but temps go up around 4 degrees after reaching stable state when on a loop. Would be same for the AIO. On a single cinebench PBO run the AIO temps would be 78 or so, but pre-soaking for 10 minutes would increase temps about 8 degrees. Every CPU is different (even if same model) so comparing coolers can only be done on the exact same parts. A great sample might run 10-15 degrees cooler than a bad sample on the same cooler.
I agree, my board should be fine for matching yours with standard PB, but should fall down in comparison with PBO. I'll never hit that 9000 score you have.
I think the answer is that Asus is playing with settings / bad defaults. My control mode is "auto overclocking" while yours is just default. I never changed any of this stuff. I'm going to have to see what is required to get that to default. edit- Actually I think that's because I set the 200MHz boost override at a point, but it'd be good to remove that.
I think the answer is that Asus is playing with settings / bad defaults. My control mode is "auto overclocking" while yours is just default. I never changed any of this stuff. I'm going to have to see what is required to get that to default. edit- Actually I think that's because I set the 200MHz boost override at a point, but it'd be good to remove that.
That might be a lead. I did try ASUS optimized VRMs etc. and it took away 0.5-1% performance. Having everything on "auto", outside of setting D.O.C.P. and TRC, is what seems to give me the best stock performance. I think the settings change quite a bit as soon as you touch any of the ASUS or AMD provided options. There is a lot of menus to go through though, unless you remember the exact settings. My SOC power is always around 18-19 watts, even on idle so really makes me wonder why yours shows up so low.
My impression is that the default setup from AMD, with the CPU controlling everything, is very good and close to optimal. Did removing the 200 boost override change anything?
I set Ryzen Master back to default for both the advanced and basic view. Back to the same old 125W / 7600 score. I did run CBR using AMD's autoOC profile untouched and back up to 8300. It doesn't seem to have a problem running 200W / 120A, but with that it only matches your base PB performance. Pretty weird stuff. I'm still leaning towards Asus just not caring about this board as I don't think there's anything wrong with my hardware. When I get a chance I'm going to try going back to Asus's (very stable) AGESA 1.0.8.0 release that I used previously and see what I get with that. It shouldn't change anything. I wouldn't upgrade my board until there's an ITX x570 board without the chipset fan.
Your PBO clocks are about the same or slightly below my PB clocks, which explains the score difference. It definitely seems strange though, but it is likely that it is something that AMD or Asus would have to solve with a correction in bios. I do think there is AGESA 1.2.0.1 or similar coming sometime within the next two months in a non-beta version.
Overall though your single core scores are normal and the multi core performance is still more than adequate for pretty much every normal task.
Overall though your single core scores are normal and the multi core performance is still more than adequate for pretty much every normal task.
I agree, I'm not fretting over the score but I do want it to be right. If it is right, I'd like to hear an explanation as to why they are. Preferably from Asus, who I think is the guilty party here. Going from 7700 (PB) to 8400 (PBO) on this board isn't a change, but I'd prefer 8400 at base PB settings. I'm definitely going to try 1.0.8.0 as it worked really well for me and I was hesitant to even go to 1.2.0.0. A lot of people were complaining about restrictions/changes in 1.2.
I also see your power deviation is 108% during CBR20 runs, which is probably near the margin of error, as I'm nearly exactly at 20% under 142W, and getting 130%. I'm out of ideas other than going to an older BIOS or waiting for a newer one. There's no reason under default BIOS settings why the CPU wouldn't be in full control and dictating power consumption to the same conditions as any other stock X470/X570 board (given my PBO results and what the board is capable of).
Thanks for the help and those screenshots.. they've really helped me figure out what's going on here, as much as I can. Being able to see what another board on the same AGESA is doing on power and readouts is pretty important.
Thanks for the help and those screenshots.. they've really helped me figure out what's going on here, as much as I can. Being able to see what another board on the same AGESA is doing on power and readouts is pretty important.
I spent the last day or so trying to flash back my bios to that older version to test.. I've flashed to older versions before on this board using the AMI DOS utility, and it's not a fun feeling to do.. worked out of course, but this time it failed a security check that wasn't a problem before, so Asus has put some additional safeguards in from flashing back. Maybe they saved me here, but who knows. I guess I'll have to keep playing with settings or wait for another update. There's definitely not a lot of x470-i users out there that moved to Zen3 and are posting.. I can't find a single person to compare with.
Afaik flashing back from AGESA 1.2.0.0 can only be done using biosflashback, not from ezflash. Your motherboard does support biosflashback (reserved USB slot). Haven't done a bios flash like that since I got the x470-i (almost 3 years ago) so don't remember the procedure for that board, but it is listed in the manual and it should work.
I'm not using either of those, ezflash or biosflashback. AMI's utility is done off a flash drive, it always worked for me in the past. I'm not going to tempt fate though. I'm sticking with what I have and whatever they put out going forward..
https://rog.asus.com/forum/showthre...-in-DOS-with-USB-tutorial-Intel-AMD-roll-back
https://rog.asus.com/forum/showthre...-in-DOS-with-USB-tutorial-Intel-AMD-roll-back
It was marked on the beta version that biosflashback was the only way to roll back the AGESA 1.2.0.0 version. AGESA 1.2.0.1 final is out for x570 so probably close for your board as well.