News Curve Optimizer Heads To Ryzen Master For Zen 3 CPUs

[Admin]

AMD has upd Ryzen Master to include the curve optimizer directly within the app and provides support for the Ryzen 7 5800X3D and Threadripper Pro 5000WX CPU lineup.

Curve Optimizer Heads To Ryzen Master For Zen 3 CPUs : Read more

 

[Makaveli]

Sweet just installed this now.

 

[spentshells]

nice more software more things to conflict

so many layers of software , so little patience to troubleshoot them all

 

[jp7189]

I've been overclocking since the days when you had to swap clock crystals, and I'm normally down for many hours or tweaking and testing, but I found the process frustrating with the 5950x. I got my curve just right for every load I could throw at it from 1 - 16 cores, but I could never get the idle states where I wanted them.. I mean it's easy to set a load and watch for stability problems. It's not so easy to not set a load to hit different idle conditions (and maintain them) long enough to prove stable operations. I thought I would be good and then would have a crash at idle once per week or so. Even PBO with no other tweaks to the curve would behave that way. Both my CCDs are bronze so maybe I'm just unlucky.. who knows. In any case I reverted back to dead stock and haven't had a crash since (6+ months).

Here's hoping this new auto curve optimizer can bring success where previously I couldn't find it.

 

[Unolocogringo]

I volunteered to be a guinea pig.
With a slight adjustment to LLC, The auto per core tuning came up with the same results I did with hours and hours of reboots and testing.
So I can say it worked for my system.
If it works for others I can not say. All in about an hour of time.

 

[Unolocogringo]

First run camu up as -29 my normal settings.
Upped llc and 2nd run came back as -30.

 

[wifiburger]

doesn't work, x570 aorus master F34 + 5900x; there's no section for curve values or the optimizer on the main screen or profiles

PBO and CO are working from the bios side just fine

 

[Unolocogringo]

My idle states are one or 2 cores maxed for feeding video cards for folding.
1.4-1.45v 24/7/365 , I hear kills them pretty quick.
She now runs 4.65 Ghz @1.21 -1.22volts and +/- 10c cooler .
Asus prime X570
R5 5600x
Patriot Viper 3600 Cas 16
FE 3060ti
970EVO plus
Pure Rock2
RM850X

doesn't work, x570 aorus master F34 + 5900x; there's no section for curve values or the optimizer on the main screen or profiles

PBO and CO are working from the bios side just fine

On the left hand side UNDER Home , click on curve optomizer.
Choose all cores or per core and click "Start Optomizing".

It took about an hour for my 5600X so may take longer for higher core counts.
Once it is done you have the option to apply,discard,save,reset,copy along the bottom of page.
Then the option shows on the home screen.

 

[noxious89123]

I've been overclocking since the days when you had to swap clock crystals, and I'm normally down for many hours or tweaking and testing, but I found the process frustrating with the 5950x. I got my curve just right for every load I could throw at it from 1 - 16 cores, but I could never get the idle states where I wanted them.. I mean it's easy to set a load and watch for stability problems. It's not so easy to not set a load to hit different idle conditions (and maintain them) long enough to prove stable operations. I thought I would be good and then would have a crash at idle once per week or so. Even PBO with no other tweaks to the curve would behave that way. Both my CCDs are bronze so maybe I'm just unlucky.. who knows. In any case I reverted back to dead stock and haven't had a crash since (6+ months).

Here's hoping this new auto curve optimizer can bring success where previously I couldn't find it.

The best tool for testing Curve Optimiser settings for low load / transient load stability is Core Cycler.

https://github.com/sp00n/corecycler

More information from its creator here; https://www.overclock.net/threads/corecycler-tool-for-testing-curve-optimizer-settings.1777398/

It's basically a tool that cycles Prime95 between single individual cores, and also intermittently pauses and unpauses the workload so that the CPU gets some idle time, and also lots of transient loads. I've used it extensively, and it's been great at detecting errors.

It's also really good at logging what it's doing, so if you come back after running it to find your PC has crashed and rebooted, you'll be able to see in the logs which core it was testing when it failed.

I can appreciate that it's not sensible to trust software from an unknown source, and on these forums I am an unknown user. Generally I only read the articles on Tom's Hardware and don't use the forums, but I have been an active user on overclock.net since 2005, if that counts for anything.

 

[wifiburger]

The best tool for testing Curve Optimiser settings for low load / transient load stability is Core Cycler.

https://github.com/sp00n/corecycler

More information from its creator here; https://www.overclock.net/threads/corecycler-tool-for-testing-curve-optimizer-settings.1777398/

It's basically a tool that cycles Prime95 between single individual cores, and also intermittently pauses and unpauses the workload so that the CPU gets some idle time, and also lots of transient loads. I've used it extensively, and it's been great at detecting errors.

It's also really good at logging what it's doing, so if you come back after running it to find your PC has crashed and rebooted, you'll be able to see in the logs which core it was testing when it failed.

I can appreciate that it's not sensible to trust software from an unknown source, and on these forums I am an unknown user. Generally I only read the articles on Tom's Hardware and don't use the forums, but I have been an active user on overclock.net since 2005, if that counts for anything.

yeah core cycler is a must for for CO testing, forget AMD curve optimizer (from what I see it returns crazy unstable values same with OCCT cycler

I just completed 8h run pass for CO-17 for my 5900x. Mostly due to core1,core7 .

I'll prob try -30 -25 -20 for each cores to see what else is stable

 

[jp7189]

The best tool for testing Curve Optimiser settings for low load / transient load stability is Core Cycler.

https://github.com/sp00n/corecycler

More information from its creator here; https://www.overclock.net/threads/corecycler-tool-for-testing-curve-optimizer-settings.1777398/

It's basically a tool that cycles Prime95 between single individual cores, and also intermittently pauses and unpauses the workload so that the CPU gets some idle time, and also lots of transient loads. I've used it extensively, and it's been great at detecting errors.

It's also really good at logging what it's doing, so if you come back after running it to find your PC has crashed and rebooted, you'll be able to see in the logs which core it was testing when it failed.

I can appreciate that it's not sensible to trust software from an unknown source, and on these forums I am an unknown user. Generally I only read the articles on Tom's Hardware and don't use the forums, but I have been an active user on overclock.net since 2005, if that counts for anything.

Thanks for posting that. I'll check it out.

 

[jp7189]

The best tool for testing Curve Optimiser settings for low load / transient load stability is Core Cycler.

https://github.com/sp00n/corecycler

More information from its creator here; https://www.overclock.net/threads/corecycler-tool-for-testing-curve-optimizer-settings.1777398/

It's basically a tool that cycles Prime95 between single individual cores, and also intermittently pauses and unpauses the workload so that the CPU gets some idle time, and also lots of transient loads. I've used it extensively, and it's been great at detecting errors.

It's also really good at logging what it's doing, so if you come back after running it to find your PC has crashed and rebooted, you'll be able to see in the logs which core it was testing when it failed.

I can appreciate that it's not sensible to trust software from an unknown source, and on these forums I am an unknown user. Generally I only read the articles on Tom's Hardware and don't use the forums, but I have been an active user on overclock.net since 2005, if that counts for anything.

I ran per core CO in the new ryzen master and it gave me -30 on all cores except #9 which was -24. Stable under load and new highs in multithreaded benchmarks
.. but CoreCycler crashed and rebooted after 4 hours.. nuts.

So thumbs up on CoreCycler finding the instability.. thumbs down on the new ryzen master optimizer

 

[drivinfast247]

I recently picked up a 5900x and started messing with curve optimization in BIOS. Using Ryzen Master to find the core hierarchy I spent a day optimizing the CPU. I had it boosting to 5150mhz and Aida64 stable with decent temps.

Then AMD releases the CO update to RM and I decided to give it a shot. In less than 2 hours it did what took me a day and ended up with boosts to 5050mhz(max I think for RM) and lower temps throughout different loads. FPS for multiple games were unchanged from what I can see.

The nice thing is I can take what RM did and tweak it further in BIOS to see what else I can gain.

 

[jp7189]

I recently picked up a 5900x and started messing with curve optimization in BIOS. Using Ryzen Master to find the core hierarchy I spent a day optimizing the CPU. I had it boosting to 5150mhz and Aida64 stable with decent temps.

Then AMD releases the CO update to RM and I decided to give it a shot. In less than 2 hours it did what took me a day and ended up with boosts to 5050mhz(max I think for RM) and lower temps throughout different loads. FPS for multiple games were unchanged from what I can see.

The nice thing is I can take what RM did and tweak it further in BIOS to see what else I can gain.

Check out corecycler for stability testing. Im a believe now that noxious turned me on to it. I too had a curve that was solid under load but then I would have random reboots while the computer was idle at night. The problem was the cores were going too low in idle states. Corecycler mixes in idle states and does a good job reporting problems even before a crash that I likely wouldn't have noticed otherwise.

 

[drivinfast247]

Check out corecycler for stability testing. Im a believe now that noxious turned me on to it. I too had a curve that was solid under load but then I would have random reboots while the computer was idle at night. The problem was the cores were going too low in idle states. Corecycler mixes in idle states and does a good job reporting problems even before a crash that I likely wouldn't have noticed otherwise.

It's been on and under load and idle for two days, so far no reboots.

Will still give that corecycler a shot, though.

 

[Unolocogringo]

For me and my 5600x I had already taken the time to optimize everything and knew what was stable.
I ran it to see what the results were.
Could it be better than what I came up with.????
Our results were the same.
I ran all core and individual core tests with same results.
Could be I got a good chip . 1.21-1.23v for 3-4 core 4.65 Ghz boost.
But up to 1.32-1.36 for 6 core boost @4.5GHz.
Core 5 is my worst core topping out stable @4.65 @1.36 anything higher requires more voltage and reduces overall clock speeds.
She runs hottest and has the highest voltage of all cores.

Edit
Higher core count chips take a lot longer to dial in, so this would give an approximate starting point of reference.
Always set your over clocks in BIOS. I Never trusted software overclocking.

 

[drivinfast247]

For me and my 5600x I had already taken the time to optimize everything and knew what was stable.
I ran it to see what the results were.
Could it be better than what I came up with.????
Our results were the same.
I ran all core and individual core tests with same results.
Could be I got a good chip . 1.21-1.23v for 3-4 core 4.65 Ghz boost.
But up to 1.32-1.36 for 6 core boost @4.5GHz.
Core 5 is my worst core topping out stable @4.65 @1.36 anything higher requires more voltage and reduces overall clock speeds.
She runs hottest and has the highest voltage of all cores.

Edit
Higher core count chips take a lot longer to dial in, so this would give an approximate starting point of reference.
Always set your over clocks in BIOS. I Never trusted software overclocking.

The Ryzen Master software changes the BIOS settings. Whatever RM does can be seen under the "SETTINGS", "ADVANCED", "AMD OVERCLOCK" tabs in the BIOS. So the software is not used after you apply a setting and restart.

At least this current version seems to work that way.

 

[jp7189]

For me and my 5600x I had already taken the time to optimize everything and knew what was stable.
I ran it to see what the results were.
Could it be better than what I came up with.????
Our results were the same.
I ran all core and individual core tests with same results.
Could be I got a good chip . 1.21-1.23v for 3-4 core 4.65 Ghz boost.
But up to 1.32-1.36 for 6 core boost @4.5GHz.
Core 5 is my worst core topping out stable @4.65 @1.36 anything higher requires more voltage and reduces overall clock speeds.
She runs hottest and has the highest voltage of all cores.

Edit
Higher core count chips take a lot longer to dial in, so this would give an approximate starting point of reference.
Always set your over clocks in BIOS. I Never trusted software overclocking.

For me ryzen master per core optimizer maxed all cores at -30 except #9 -24. It took at little over 2 hours on 5950x. It wasn't even remotely stable. What was great about core cycler is it called out the cores with problems (all but 3). I used that info to lower those cores to -25. Core cycler then found only 2 cores still had a problem so I lowered those to -20. Just completed an overnight run and all looks good. I'll reserve final judgment until after I have a few weeks of no problems under normal usage.

 

[Unolocogringo]

The Ryzen Master software changes the BIOS settings. Whatever RM does can be seen under the "SETTINGS", "ADVANCED", "AMD OVERCLOCK" tabs in the BIOS. So the software is not used after you apply a setting and restart.

At least this current version seems to work that way.

I never trusted software overclock programs .
So I use RM to change settings within windows for testing and monitoring. If I mess up a setting and it crashes ,on reboot I am back to my last stable bios settings.
I would never apply settings from there.

I am stating my experience with the program.
As an avid overclocker I can push it harder with PBO and get better benchmark scores with more voltage. That couple hundred mhz is not worth the required voltage to get there. AND still run 24/7/365 with 1 or 2 cores @1.38-1.41 which would shorten its life considerably.
My R5 3600 on the other hand runs 1.28v @4.4GHz all core no problem.
Every chip is different
😲
My first overclock was a massive improvement at the time. Pentium 75mhz @ 90mhz. just the step up from 50mhz FSB to 60mhz FSB ran everything faster.
I was hooked from then on.

 

[ZeroCool22]

I never trusted software overclock programs .
So I use RM to change settings within windows for testing and monitoring. If I mess up a setting and it crashes ,on reboot I am back to my last stable bios settings.
I would never apply settings from there.

I am stating my experience with the program.
As an avid overclocker I can push it harder with PBO and get better benchmark scores with more voltage. That couple hundred mhz is not worth the required voltage to get there. AND still run 24/7/365 with 1 or 2 cores @1.38-1.41 which would shorten its life considerably.
My R5 3600 on the other hand runs 1.28v @4.4GHz all core no problem.
Every chip is different
😲
My first overclock was a massive improvement at the time. Pentium 75mhz @ 90mhz. just the step up from 50mhz FSB to 60mhz FSB ran everything faster.
I was hooked from then on.

So tell me if i understand correctly, the RM Curve Optimzer changes will only be active after i open RM and hit Apply or it will be a permanent change in the BIOS itself?

Because i use RM since some time (not the last version with CO) and what i like about it is, that doesn't make permanent changes on the BIOS and changes are only applied after you open the program and click Apply, but i don't know if this is the same case with the new CO feature...

 

[hotaru.hino]

So tell me if i understand correctly, the RM Curve Optimzer changes will only be active after i open RM and hit Apply or it will be a permanent change in the BIOS itself?

Because i use RM since some time (not the last version with CO) and what i like about it is, that doesn't make permanent changes on the BIOS and changes are only applied after you open the program and click Apply, but i don't know if this is the same case with the new CO feature...

My money's on it won't save anything to BIOS if it doesn't do so already. However, for the Curve Optimizer feature, I don't see a reason why you'd want it "on demand" unless Ryzen Master has the ability to revert to a safe state without restarting the OS and you're dialing in the numbers.

 

[drivinfast247]

So tell me if i understand correctly, the RM Curve Optimzer changes will only be active after i open RM and hit Apply or it will be a permanent change in the BIOS itself?

Because i use RM since some time (not the last version with CO) and what i like about it is, that doesn't make permanent changes on the BIOS and changes are only applied after you open the program and click Apply, but i don't know if this is the same case with the new CO feature...

It directly changes the BIOS and you do not need to open RM upon boot for changes to take affect.