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

[Guest2]

Might take a while, but why don't you try testing 1 core at a time if you are struggling to get stability by making mass changes.

I let Ryzen master pick my two best cores (6 as gold and 0 as second best)
Should this not be trusted?

Ideally I would like to leave all cores as they are with 6 and 0 to boost if these are the best

 

[LtMatt]

I let Ryzen master pick my two best cores (6 as gold and 0 as second best)
Should this not be trusted?

Ideally I would like to leave all cores as they are with 6 and 0 to boost if these are the best

Yes that's the right method, focus on those for now and negative them as much as you can without instability.

If you do get instability, either remove two notches from the curve optimizer or reduce the Auto OC by 25Mhz.

 

[Guest2]

Yes that's the right method, focus on those for now and negative them as much as you can without instability.

If you do get instability, either remove two notches from the curve optimizer or reduce the Auto OC by 25Mhz.

So it is negatives on the good cores, rather than negatives on all cores except good one?

There seems to be a lot of confusion on PBO / PBO2. Perhaps AMD should have not included it in BIOS updates without good (any) documentation. All I can find is youtube reviews, peoples comments on forums and a few AMD slides that say to stick it on x10 scaler with 200Mhz offset and negatives on the good cores. Others are saying negative offsets on all cores and leave the good cores with no offset

 

[LtMatt]

So it is negatives on the good cores, rather than negatives on all cores except good one?

There seems to be a lot of confusion on PBO / PBO2. Perhaps AMD should have not included it in BIOS updates without good (any) documentation. All I can find is youtube reviews, peoples comments on forums and a few AMD slides that say to stick it on x10 scaler with 200Mhz offset and negatives on the good cores. Others are saying negative offsets on all cores and leave the good cores with no offset

It can be either, no sample is the same unfortunately.

Start with negative on your two best cores, keep going until you get instability. Then focus on the other cores.

That said, try my settings and see how they work for you if you don't have the patience for the above.

PBO Enabled
PBO limits set to Motherboard
Scalar x10
Auto OC set to 100Mhz
First 6 cores, negative offset, try -5 at first, then later try -10 if stable
Second 6 cores, negative offset, try -10 at first then later try -20 if stable

This works well on my 5950x. If i put the offsets lower than -10 / -20, i get instability. If i increase Auto OC by 25Mhz more, i get instability.

This combination is what got me to the top of the Cinebench R20 and R23 ST benchmarks.

 

[xAliasx]

AMD 5800X AMD Curve Optimizer Boost set to 5050Mhz for all cores and good temperatures when PC gaming. So basically when all things are good thermal,electrical ETC all cores have the ability to boost up to 5050Mhz.

offset VF curve voltage chart
https://docs.google.com/spreadsheet...QvVSKyfD4DQghgUHb46-VIIzl3BQvX-hYt0D4/pubhtml

Timestamp and BIOS settings

Cinebench -Single core 653 - 0:01

The Dark Pictures Anthology LittleHope - 0:23

Assassin's Creed Valhalla - 0:37

Metro Exodus - 0:48

Spoiler

[/spoiler]

Thanks for posting your analysis. Was pretty helpful to use the curve optimizer with it. I managed to get -10 on the first 4 cores and -35 on the 5-12 cores with a 100mhz boost on my 5900x. Havent done much stability checks but getting 9000 in Cinebench 20 multi core.

Btw, what was the overlay you had in your games to display the stats? Assuming that is not MSI afterburner.

 

[gerardfraser]

Thats why there is spec and there is silicon lottery. This is overclockers, nothing is guaranteed.

Besides, I am yet to see anyone with a true lemon CPU. All reports of bad cases can be chalked up to motherboard, bios, bios settings, cooling. All Zen 3 are wildly overperforming their spec.
I would even go on a limb and say gerards seemingly stellar results are not so much because of silicon lottery, but because of a very good motherboard.

Btw setting more negative offset on good cores is not always a good idea, whatever the slides say. For me, two good cores I keep at 0 and experiment with worse ones, to bring their actual clocks higher.

This is 100% correct. I have mentioned this a few times in this thread and I do not have a golden sample on any Ryzen CPU I owned and I owned 6/7 of them in the last 1-2 years but always been suggested I have a golden sample on everyone of them ,which is pure BS and impossible.JMO

Thanks for posting your analysis. Was pretty helpful to use the curve optimizer with it. I managed to get -10 on the first 4 cores and -35 on the 5-12 cores with a 100mhz boost on my 5900x. Havent done much stability checks but getting 9000 in Cinebench 20 multi core.

Btw, what was the overlay you had in your games to display the stats? Assuming that is not MSI afterburner.

Your welcome and it is MSI Afterburner and RTSS Overlay Editor, I done a quick start up guide for the OSD and uploaded it to google drive. Included in the read me a link to the creators tutorials of MSI Afterburner and RTSS Overlay Editor. IF you want to start from scratch his name is Alexey Nicolaychuk just search youtube . https://drive.google.com/file/d/1Flt2EeznPd2HO438678AbjUHnwWf6E6D/view?usp=sharing

 

[Combat Fighter]

It can be either, no sample is the same unfortunately.

Start with negative on your two best cores, keep going until you get instability. Then focus on the other cores.

That said, try my settings and see how they work for you if you don't have the patience for the above.

PBO Enabled
PBO limits set to Motherboard
Scalar x10
Auto OC set to 100Mhz
First 6 cores, negative offset, try -5 at first, then later try -10 if stable
Second 6 cores, negative offset, try -10 at first then later try -20 if stable

This works well on my 5950x. If i put the offsets lower than -10 / -20, i get instability. If i increase Auto OC by 25Mhz more, i get instability.

This combination is what got me to the top of the Cinebench R20 and R23 ST benchmarks.

Ive got a 5950 as well.

One thing that’s confusing about your post you say first 6 cores, second 6 cores.

What happened to the other 4 cores?

 

[LtMatt]

Ive got a 5950 as well.

One thing that’s confusing about your post you say first 6 cores, second 6 cores.

What happened to the other 4 cores?

Guest has a 5900x so i was talking about 12 cores total.

For 5950x try the first 8 cores, etc.

This is just what worked well for my sample, no guarantee it will work well on others. Also, as @gerardfraser says above it may not be the best way to do it, but worth trying for some quick plug and play settings.

I think it works well with my sample as it gives a bit more power headroom which improves scores.

 

[Combat Fighter]

Guest has a 5900x so i was talking about 12 cores total.

For 5950x try the first 8 cores, etc.

This is just what worked well for my sample, no guarantee it will work well on others. Also, as @gerardfraser says above it may not be the best way to do it, but worth trying for some quick plug and play settings.

I think it works well with my sample as it gives a bit more power headroom which improves scores.

What sort of clocks are you getting with them settings?

 

[UKDTweak]

LTMatt, What were you using to test for stability after each change you made?

I got my highest R20 and CPU-Z score so far with -20 on all cores but never see a any cores go over 4467 on multi test or 4967 on single... though HWINFO history does show three of my cores hitting a Max of 5040 and 5015.
I have PBO set to Motherboard, everything else on Auto, ive tried various Scaler settings and from 25mhz to 200 and got worse scores, so I just dont use those two settings and leave them on auto for now.

 

[Guest2]

Guest has a 5900x so i was talking about 12 cores total.

For 5950x try the first 8 cores, etc.

This is just what worked well for my sample, no guarantee it will work well on others. Also, as @gerardfraser says above it may not be the best way to do it, but worth trying for some quick plug and play settings.

I think it works well with my sample as it gives a bit more power headroom which improves scores.

I have a 5800X, so 8 cores (or 0 - 7) core 6 is the strongest, core 0 is second strongest according to Ryzen master

 

[LtMatt]

I have a 5800X, so 8 cores (or 0 - 7) core 6 is the strongest, core 0 is second strongest according to Ryzen master

Oh sorry, I'd go with my initial suggestion then.

 

[LtMatt]

LTMatt, What were you using to test for stability after each change you made?

I got my highest R20 and CPU-Z score so far with -20 on all cores but never see a any cores go over 4467 on multi test or 4967 on single... though HWINFO history does show three of my cores hitting a Max of 5040 and 5015.
I have PBO set to Motherboard, everything else on Auto, ive tried various Scaler settings and from 25mhz to 200 and got worse scores, so I just dont use those two settings and leave them on auto for now.

The best stability test of them all. General system usage and extended gaming sessions. They always sniff out any unstable overclock.

 

[schoenzy]

Interesting. A larger negative offset on the weaker cores seems counter intuitive to what AMD is suggesting in their presentation but it's obviously working for you so I may give it a try to see if the multi-core performance improves.

My memory is next on the list. I have it running at 1900 on an untuned XMP profile, was dithering over whether to tune further incase the next AGESA would better support 2000 FCLK (crashes my machine right now), but might do it anyway.

Yeah this seems counter intuitive, but the logic from what I've gathered is windows will hit your 2 best cores the hardest. The others are more likely to boost as part of all core to a lower level in effective clock, and lower curve helps get these a bit higher, but because these won't be pegged to sit at 5GHz effective, stability should be okay. While your weaker cores will record clock highs above 5GHz, they are not likely to go to that level in effective clocks at all, let alone for extended periods. When I tried any lower than -5 on my 2 best, instant crash in R20, even launching it, so this seems to align with the theory.

 

[gerardfraser]

My CPU is not broken is it,because I seem to be able to hit effective clock 5000Mhz just running some benches and whatever else. Then there is post#1 which is about having a more sustained boost clock with AMD Curve Optimizer over PBO boost.
IF you want all cores to boost to 5000Mhz ,maybe try
Assassin's Creed Valhalla 5800X 5000Mhz all core overclock for some gaming.This is a real all core overclock and I am J/K with this post I quoted but there some crazy things in this thread overall.

 

[Saunupe1911]

Can someone share their curve and CPU settings for a Gigabyte X570 Aorus Master and a 5900X?

I rarely see a single core go ver 4.2Ghz via PBO.

 

[gerardfraser]

@bot
With PBO disabled a 5900X defaults to 4950Mhz.

 

[Guest2]

It can be either, no sample is the same unfortunately.

Start with negative on your two best cores, keep going until you get instability. Then focus on the other cores.

That said, try my settings and see how they work for you if you don't have the patience for the above.

PBO Enabled
PBO limits set to Motherboard
Scalar x10
Auto OC set to 100Mhz
First 6 cores, negative offset, try -5 at first, then later try -10 if stable
Second 6 cores, negative offset, try -10 at first then later try -20 if stable

This works well on my 5950x. If i put the offsets lower than -10 / -20, i get instability. If i increase Auto OC by 25Mhz more, i get instability.

This combination is what got me to the top of the Cinebench R20 and R23 ST benchmarks.

I tried these settings.

First as above with first 4 cores on -5 and next 4 cores on -10
Tested and ran a few benchmarks, all seems good. R20, R23, Realbench, CPUZ, memtest, DRAM calc tests
Then set first 4 cores on -10 and next 4 cores on -20
Tested and ran a few benchmarks, all seems good. R20, R23, Realbench, CPUZ, memtest, DRAM calc tests

Had an hour or so on Warzone, perfect all stable.
Then decided to give R20 a run. Crashed as soon as I started the bench.

This seems quite random to me - That it can pass rounds of benchmarking and gaming then bomb out on the same test 3 hours later.

The same happened with 200Mhz and some different settings, all seemed well and good then a random reboot in CPUZ. Could it be something like a spike or not enough voltage at the correct time?

Back to auto PBO again for now.

 

[Guest2]

My CPU is not broken is it,because I seem to be able to hit effective clock 5000Mhz just running some benches and whatever else. Then there is post#1 which is about having a more sustained boost clock with AMD Curve Optimizer over PBO boost.
IF you want all cores to boost to 5000Mhz ,maybe try
Assassin's Creed Valhalla 5800X 5000Mhz all core overclock for some gaming.This is a real all core overclock and I am J/K with this post I quoted but there some crazy things in this thread overall.

All core 4.95Ghz and only 67.5c max cpu temp Nice. Didnt you say you were only using a cheap AIO cooler too?

 

[gerardfraser]

Yes cheap CPU cooler and the video was 5000Mhz all core.Now even you can do the same thing,add CPU voltage and try if you want. Not recommended for less than crazy people.I am uploading a video with ALL core 5000Mhz with HWinfo64 on,only good for PC gaming and not P95