• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

5800X Boost set to 5050Mhz all cores with AMD Curve Optimizer

Yes that is different from what I see. If i remove all the negative voltage my all core frequency goes down to 4.6-4.7Ghz. That said, i suspect it's a combination of both, but for me the negative voltage provided more benefit.


Interesting. When you say car, is it more like a land rover? If so i think i know where it is. I will have to try and remember to try that out next time.

If you can ever grab a screenshot of the spot and the place where you aim, let me know.

Not sure if the map / car has been optimised now after numerous patches but it used to be this one. Crouching inside the front of the red bus you can go inside (lower right below) and looking towards it also tanked my fps

Off topic still but what GPU and CPU time in ms are you getting generally? For me im getting 6-8ms GPU, 3-5ms CPU

UKaN3Pq.png
 
Not sure if the map / car has been optimised now after numerous patches but it used to be this one. Crouching inside the front of the red bus you can go inside (lower right below) and looking towards it also tanked my fps

Off topic still but what GPU and CPU time in ms are you getting generally? For me im getting 6-8ms GPU, 3-5ms CPU

UKaN3Pq.png
Got it, i know that bus and have been in it before. I'll check that out next time and see if i can grab a screenshot - assuming i can stay alive long enough. :p

I have no idea what the ms i am getting, given the high FPS i will assume they are very low.

This game appears to work very well with the RX 6800 XT and there is no stuttering at all. It uses a lot of video memory too this game.
 
Got it, i know that bus and have been in it before. I'll check that out next time and see if i can grab a screenshot - assuming i can stay alive long enough. :p

I have no idea what the ms i am getting, given the high FPS i will assume they are very low.

This game appears to work very well with the RX 6800 XT and there is no stuttering at all. It uses a lot of video memory too this game.
Not much at your measly 1440p :p

Back on topic.
Should setting negative offset on all cores to say -5 yield a higher all core boost than currently (4.55Ghz)
I'm wanting to keep the PBO single core boost on two good cores to 4.95Ghz, whilst pushing all cores up to 4.7Ghz
 
Not much at your measly 1440p :p

Back on topic.
Should setting negative offset on all cores to say -5 yield a higher all core boost than currently (4.55Ghz)
I'm wanting to keep the PBO single core boost on two good cores to 4.95Ghz, whilst pushing all cores up to 4.7Ghz
My native resolution is actually 5120x1440 @120HZ, but i recorded this video at lower resolution to flex the 5950X with Smart Access Memory muscles. :)

Optimize each individual core for best results. A global setting will reduce your potential frequency.
 
So I've settled on my PBO settings now, ended up going back to LLC Auto, disabling curve optimizer because it wasn't doing anything for me.

Now getting 11k cinebench r20 multithread and 630 single thread.

Multithread max power draw is 250w with voltage between 1.26 and 1.28 (clocks settle between 4.35ghz to 4.4ghz)

Single thread max power draw is 80w with voltage between 1.4 and 1.45 (clocks between 4.8ghz and 5.1ghz depending on which core is loaded)

Gaming power draw is 80-130w depending on how multithreaded the game is and voltages generally sit around 1.40 (clock generally between 4.7ghz and 4.8ghz on any core that is loaded)

The most importantly thing is its stable, havent had a crash in 15 hours of stability tests, benchmarking, games etc.

I don't know why curve optimizer doesn't do anything for me, but I know that tons of people say the current Agesa is very buggy so I'll put it down to a bad BIOS. I will probably update the BIOS and re-tune PBO when a more stable BIOS is released in January
 
So I've settled on my PBO settings now, ended up going back to LLC Auto, disabling curve optimizer because it wasn't doing anything for me.

Now getting 11k cinebench r20 multithread and 630 single thread.

Multithread max power draw is 250w with voltage between 1.26 and 1.28 (clocks settle between 4.35ghz to 4.4ghz)

Single thread max power draw is 80w with voltage between 1.4 and 1.45 (clocks between 4.8ghz and 5.1ghz depending on which core is loaded)

Gaming power draw is 80-130w depending on how multithreaded the game is and voltages generally sit around 1.40 (clock generally between 4.7ghz and 4.8ghz on any core that is loaded)

The most importantly thing is its stable, havent had a crash in 15 hours of stability tests, benchmarking, games etc.

I don't know why curve optimizer doesn't do anything for me, but I know that tons of people say the current Agesa is very buggy so I'll put it down to a bad BIOS. I will probably update the BIOS and re-tune PBO when a more stable BIOS is released in January
Not bad.

Once PBO is properly tuned with Curve Optimizer you can look at up to +5-7% on single thread and +6-10% on multi thread scores.

Tv6mawo.png


7JINqgW.png
 
I too went back to PBO auto. Despite my re attempts yesterday and getting R20 and R23 stable with slightly higher single and multi thread scores when playing CoD Warzone there were parts it went a bit ‘jittery’ in 3 separate games. No issues with latency, server or gpu. Changed back to auto and it stopped.

Maybe it was boosting certain cores then when they dropped was causing the jitter, not sure. Either way, auto pbo and curve optimiser disabled again too for me until new bios is released.
 
Hi all.
Just built my system and have been playing about with the curve optimizer and pbo.

System is as follows...
5800x
B550 Gaming Carbon WiFi
16gb Crucial Ballistix 3600 C16
Superflower Leadex Gold 850w

Most settings I try seem to result in a hard reset about 2 mins into a cinebench r23 run, any ideas what's happening?
 
Played around with curve optimizer in combination while also using manually set lower PPT TDC EDC values and I've manage to increase multicore performance over just using the curve optimizer while also lowering temps by 10c and power down from 142 at stock to 117 with just curve optimizer to just 100w with both curve + lower PPT TDC EDC values.

Stock
Screenshot-19.png


Curve optimizer -20 + PPT 125 TDC 88 EDC 125 + core offset +50

Increasing the negative curve further results in single core instability but by reducing the PPT/TDC/EDC instead the single core is not effected but multi core sees additional lower power usage and lower temps which further boosts the score.

Screenshot-44.png


CPU is using just 100w

Screenshot-45.png
 
Last edited:
Running a 5800X on Strix X470 board.
PBO on all Auto +225mhz
Put my curve at:
Core0 -20
Core1 -10** 2nd best
Core2 -10* Best core
Core3 -20
Core4 -30 Worst core
Core5 -15
Core6 -30 2nd worse
Core7 -15
Gets me 643/6061 in CB20 and 685/6820 in CPU-Z

Still trying to figure out the curve/PBO best settings... I have seen people getting +650/6100 in CB20

Update:
PBO Enabled Auto
PBO Limits Motherboard
Scalar x10
Auto OC 175Mhz
641/6142 in CB20 and 685/6820 in CPU-Z 683/6849
 
Last edited:
Anyone have luck with the 5800X? Whats you setting or process? Looked thru the post and everyone seems to be trying their own thing. Here what has worked and not worked for me:

5800X on Strix X470 with 32GB(4x8) @3800
PBO w/Auto
PBO Limits Motherboard
Scalar x10
Auto OC 175Mhz

Put my curve at:
Rank/Core
6th/Core0 -20
2nd/Core1 -5
1st/Core2 -5
5th/Core3 -20
8th/Core4 -30
3rd/Core5 -15
7th/Core6 -30
4th/Core7 -15
Gets me 640s/6100s in CB20 and 680s/6800s in CPU-Z

So far this is best I can do.... I've tried what other have said as far as decrease curve ex.(-30 to -20)on bad cores but that dropped performance. I also found that if I increase curve ex.(-5 to -10) on my best cores that CB20 crashes on single thread test. If I drop the Auto OC from 175 to 150 is fixes crashing from increasing curve but I lose alittle performance.
 
So I've settled on my PBO settings now, ended up going back to LLC Auto, disabling curve optimizer because it wasn't doing anything for me.

Now getting 11k cinebench r20 multithread and 630 single thread.

Multithread max power draw is 250w with voltage between 1.26 and 1.28 (clocks settle between 4.35ghz to 4.4ghz)

Single thread max power draw is 80w with voltage between 1.4 and 1.45 (clocks between 4.8ghz and 5.1ghz depending on which core is loaded)

Gaming power draw is 80-130w depending on how multithreaded the game is and voltages generally sit around 1.40 (clock generally between 4.7ghz and 4.8ghz on any core that is loaded)

The most importantly thing is its stable, havent had a crash in 15 hours of stability tests, benchmarking, games etc.

I don't know why curve optimizer doesn't do anything for me, but I know that tons of people say the current Agesa is very buggy so I'll put it down to a bad BIOS. I will probably update the BIOS and re-tune PBO when a more stable BIOS is released in January

I am having the same problem at the moment. Until recently CO massively effected boost clock. Now no matter what PBO setting or CO setting I use, my single core boost is not going any higher than stock. I can set PBO + 200 CO -10 on all cores and the single threads aren't pushing past 5000. Previously they would shoot up to 5200+ on some cores. I have done a BIOS reset which did nothing and currently putting on a fresh install of Windows to see if it was Ryzen Master screwing with it. Failing that, I will try lower Ram speed with lower VSOC voltages. I think it could just be buggy with the current BIOS at the moment.

You can see this image of my heavily under optimised CO when I had no idea what I was doing I got this result. I think PBO was +100 and I set the 2 fastest cores to -10 and the rest to -5. I managed to get all cores going over 5000 on a better optimised CO, but didn't take a pic of that.

ppq7L4nh.jpg

Now this is with PBO + 200 and CO -10 on weaker cores CCD 0 and -5 on best and -15 on CCD 1

TULgOCNh.jpg

It is lower than my stock scores. Either my software is reporting incorrect info, or something is screwing with the boost.
 
Last edited:
PBO would be the one for max single clock and curve to boost multi, I think. Fine balance for the two of them when upping pbo boost. I've been working the opposite way, all core curve - 30, scaler auto, pbo +0 then upping pbo 25 at a time and adjusting curve. 640/9.1k 5900x. Stopped at +75 as sold the chip. Still had another 125mhz to play with to boost single and vcore offset.

Definitely interesting chips to play with and sooner the better for bios update.
 
Sorted the PBO problem out with a fresh install. But you can see HWInfo recording the wrong speeds. Ryzen Master shows every core reaching over 5ghz using Boost tester with an optimised curve, however HWInfo shows a huge discrepancy. You can see Core 6 reaching 5135 in Ryzen Master, but only 4950 on HWInfo. Reducing the polling in HWInfo makes no difference either.

DyAiO4uh.jpg
 
Wake me up guys when you can do this with the curve lol
Mass Effect Andromeda Boost All Cores 5000Mhz

Xuan-Yuan Sword VII Boost All Cores 5050Mhz

Assassin's Creed Valhalla Boost All Cores 5075Mhz

Observer System Redux -Palatium CLUB 5100Mhz All cores

SEKIRO SHADOWS DIE TWICE -Diamond CLUB 5125Mhz All cores

Your spot here -Titanium CLUB 5150Mhz




Xuan-Yuan Sword VII Boost -Prestige CLUB 5200Mhz

 
Had a second play with curve optimizer but it's still useless just like Derbauer said.

I can do -10 on all cores, I can put -25 on my two best cores, heck I can even do -15 on all cores and yet nothing changes performance. I do see clock speed increase very slightly (saw multi core go from 4.38 to 4.42 and saw single core go from 4.9 to 5 HOWEVERE the Cinebench scores are exactly the same, the voltages are the same, the power draw is the same, the temps are the same

Curve optimizer is either broken on Asus x570 boards or it's just a useles placebo effect
 
Back
Top Bottom