The Ryzen 7 5800X has a way to high powertarget (great results with powerlimit tweaks)

[Nexium]

These two do the same thing. You can select one or the other. Ryzen doesn't need you to set voltages, it can demand it from the motherboard itself. Try to keep one and forego the other. It misses the point if you reduce 15mV and add 125 somewhere else. It only reduces base voltage and that is without the incrementally static 125 offset anyway.
We dearly miss the Stilt making things clearcut.

Thank you for your suggestion. I tried it and set CPU LLC back to AUTO setting. I did not pass my usual GTA 5 game stress test and crashed to desktop within about 30 min. So to me this should disprove PBO Scalar provides the proper voltage stability needed to replace my current setting with CPU LLC at level 3.

I was only able to reach -15 stable with the CPU with both the motherboard .125 offset and CPU LLC at level 3. Without these two, I'm back at -12 to maintain stability. Again, my theory is the -15 should allowing the CPU's algorithm to boost to max clocks more often and consistently with all cores to 5Ghz during normal or heavy use which is what I'm seeing in just by browsing the web with Chrome and a few office apps running.

 

[blu3dragon]

These two do the same thing. You can select one or the other. Ryzen doesn't need you to set voltages, it can demand it from the motherboard itself. Try to keep one and forego the other. It misses the point if you reduce 15mV and add 125 somewhere else. It only reduces base voltage and that is without the incrementally static 125 offset anyway.
We dearly miss the Stilt making things clearcut.

My understanding of PBO scaler is that it raises either the time for which the cpu will request max voltage (allowing maximum frequency), or it raises the maximum voltage itself, allowing it to sustain a higher boost clock at a given temperature. As temperature increases, the allowable voltage decreases. As requested voltage decreases, clock speed decreases. Increasing this value is effectively a way to oc the cpu, with the cost being possible reduced lifespan of the cpu.

This has a slightly different effect to LLC, which alters the voltage drop between the requested (set) level and that actually seen by the cpu. Increasing LLC will not alter boost clocks directly, but since increasing LLC will increase actual voltage at the cpu core, it will increase heat, which will the result in the cpu lowering it's max voltage band then lower the clock to match. Increasing LLC can help with stability, but also possibly at the cost of some cpu life and setting to a high level can result in voltage overshoots. Generally it is better to have this at a medium to low setting and increase stability by increasing the set (or requested) voltage level.

Setting a voltage offset and changing LLC might be more similar. A voltage offset will offset the voltage level requested by the cpu and so lowering this would have a somewhat similar effect to lowering LLC. However, with Zen3 the cpu seems to be monitoring this somehow and clock stretching when the set voltage goes too low.

So far in my experience, it seems best to leave voltage offset at 0, llc at auto, and just tune curve offset.

You can also increase PBO scaler to allow higher boosts at a given temperature as long as you are ok with potentially reduced cpu life.

 

[mtcn77]

My understanding of PBO scaler is that it raises either the time for which the cpu will request max voltage (allowing maximum frequency), or it raises the maximum voltage itself, allowing it to sustain a higher boost clock at a given temperature. As temperature increases, the allowable voltage decreases. As requested voltage decreases, clock speed decreases. Increasing this value is effectively a way to oc the cpu, with the cost being possible reduced lifespan of the cpu.

This has a slightly different effect to LLC, which alters the voltage drop between the requested (set) level and that actually seen by the cpu. Increasing LLC will not alter boost clocks directly, but since increasing LLC will increase actual voltage at the cpu core, it will increase heat, which will the result in the cpu lowering it's max voltage band then lower the clock to match. Increasing LLC can help with stability, but also possibly at the cost of some cpu life and setting to a high level can result in voltage overshoots. Generally it is better to have this at a medium to low setting and increase stability by increasing the set (or requested) voltage level.

Setting a voltage offset and changing LLC might be more similar. A voltage offset will offset the voltage level requested by the cpu and so lowering this would have a somewhat similar effect to lowering LLC. However, with Zen3 the cpu seems to be monitoring this somehow and clock stretching when the set voltage goes too low.

So far in my experience, it seems best to leave voltage offset at 0, llc at auto, and just tune curve offset.

You can also increase PBO scaler to allow higher boosts at a given temperature as long as you are ok with potentially reduced cpu life.

+1. I think you have good insight into the topic. Honestly, I wouldn't worry about scalar if that is the case precisely because right above you also have the PTTL setting which should, incidentally, determine when the scalar starts its temperature adjustment.

 

[Nexium]

My understanding of PBO scaler is that it raises either the time for which the cpu will request max voltage (allowing maximum frequency), or it raises the maximum voltage itself, allowing it to sustain a higher boost clock at a given temperature. As temperature increases, the allowable voltage decreases. As requested voltage decreases, clock speed decreases. Increasing this value is effectively a way to oc the cpu, with the cost being possible reduced lifespan of the cpu.

This has a slightly different effect to LLC, which alters the voltage drop between the requested (set) level and that actually seen by the cpu. Increasing LLC will not alter boost clocks directly, but since increasing LLC will increase actual voltage at the cpu core, it will increase heat, which will the result in the cpu lowering it's max voltage band then lower the clock to match. Increasing LLC can help with stability, but also possibly at the cost of some cpu life and setting to a high level can result in voltage overshoots. Generally it is better to have this at a medium to low setting and increase stability by increasing the set (or requested) voltage level.

Setting a voltage offset and changing LLC might be more similar. A voltage offset will offset the voltage level requested by the cpu and so lowering this would have a somewhat similar effect to lowering LLC. However, with Zen3 the cpu seems to be monitoring this somehow and clock stretching when the set voltage goes too low.

So far in my experience, it seems best to leave voltage offset at 0, llc at auto, and just tune curve offset.

You can also increase PBO scaler to allow higher boosts at a given temperature as long as you are ok with potentially reduced cpu life.

Based on your in depth explanation of the PBO Scaler, it's similar to my personal understanding of how it works and aware of the consequence to the lifespan of the CPU for more performance in theory. Temperature is not an issue for me at the moment, but I'm still keeping a close eye on my CPU voltages to keep it within or below the max voltage limit of these 5000 Ryzen CPUs when boosting clocks. I also tried CPU LLC set one level down lower to 4 for instance, unfortunately games start crashing again so I believe I have found my CPUs limit or close to it while maintaining stability with my settings. Fine tuning to running less voltage as possible at the limit is always a good thing.

There are always risks with overclocking to the limit I'm used to and accepted for many years . Technology these days are outdated so fast like cellular phones, I do have less regard to keeping any computer components for no more than 4-5 years at a time anyway to satisfy this hobby. Plus, I haven't personally experienced any degradation effects with any overclocked Intel or AMD CPUs over the past 20 years doing this as long as I stay within or below the safe acceptable voltage and temperature range, but then again I'm also not running at 100% 24/7.

 

[Zach_01]

Based on your in depth explanation of the PBO Scaler, it's similar to my personal understanding of how it works and aware of the consequence to the lifespan of the CPU for more performance in theory. Temperature is not an issue for me at the moment, but I'm still keeping a close eye on my CPU voltages to keep it within or below the max voltage limit of these 5000 Ryzen CPUs when boosting clocks. I also tried CPU LLC set one level down lower to 4 for instance, unfortunately games start crashing again so I believe I have found my CPUs limit or close to it while maintaining stability with my settings. Fine tuning to running less voltage as possible at the limit is always a good thing.

There are always risks with overclocking to the limit I'm used to and accepted for many years . Technology these days are outdated so fast like cellular phones, I do have less regard to keeping any computer components for no more than 4-5 years at a time anyway to satisfy this hobby. Plus, I haven't personally experienced any degradation effects with any overclocked Intel or AMD CPUs over the past 20 years doing this as long as I stay within or below the safe acceptable voltage and temperature range, but then again I'm also not running at 100% 24/7.

Do not compare these 7nm chips to what we had 5, 10, or 20 years ago. The same or even less electomigration could potentially inflict more degradation on thinner traces and transistors, caused by voltage and high-er current. Even if temp is within limits. Taking scalar on manual, and above X1, is nevertheless an override, hence PBO-verride. Use it wisely and sparingly, I sincerely advise you.

Too much sustained voltage (+higher speed +load = higher sustained current) could worsen things quickly too, as much as too high voltage alone under load. Thats why these 7nm chiplets hit their stock target voltages and speeds shortly and as AMD declares... opportunistically.

I will agree that 5000's curve optimizer is a way better OC, in terms of performance and reliability, than static values that its the only way on 3000... but still its easy to misuse and you should all be cautious as there is no background that connects 7nm ZEN3 (and ZEN2) to anything. Any reference to older gens of chips I find it to be mistargeted and/or irrelevant.
Tell me 1 CPU of the past (pre 7nm) that can be degraded with 1.35V static. Well, ZEN2 can and I'm pointing at this as an example to past references. Its uncharted waters both ZEN2/3.

What OCing users (most) seems not grasping is how and why these CPUs are regulated the way they are. Its my estimate based on what I see on attempted settings.

----------------------------------------------------

This is general and not directly refering to 7nm ZEN but its food for thought for everyone.

 

[blu3dragon]

I also discovered something interesting recently. The 5800x is unique in the way it generates heat, and may require adjustment of the positioning of the cooler and mounting. Read more here: https://www.reddit.com/r/overclocking/comments/kszoh9

Since I have an arctic liquid freezer ii rev3 with an offset mount option I decided to try this. I had originally installed it with the normal mounting position. Removed and switched to the offset position. End result? Almost no difference. After letting it sit for 24hrs, with an 85C temp limit set in the bios, p95 with fixed fft size of 84, avx disabled, I was previously at ~145-150W with a motherboard temp of 22C. Now I'm at 150-152W with a motherboard temp of 23C. That's either a very slight improvement, measurement error, or a change due to better luck with the quantity of thermal compound I used

Regarding PBO Scaler I just did a quick experiment and this appears to make no difference with a steady state load that is not power or temp limited. Maybe this makes a short term difference if a power or voltage limit is reached? I also checked all core steady state load and that was the same as well.

I tried with P95, single thread, in place fft fixed size of 84, avx disabled, and used hwinfo to report average effective clocks over >15s for each core.

First ran with all cores to let everything heat soak, then switched to a single core, and used task manager to assign it to a single core (same core each time). Let it sit for a bit for temps to settle down. Clicked reset counters, then let it run for >15s for the average to settle down and made a note of that.

BTW, this is a very good way to test if your curve setting is stable for each core. I thought I was 100% stable, but core2 failed after a few min of having a single thread locked to that core and so I need to go reduce that offset :-|

	scaler x1	scaler x1	scaler x10	scaler x10
Core	Average effective clock	temp (C)	Average effective clock	temp (C)
0	4850	70	4840	71.3
1	4827	70	4848	70.3
2	4806	69.6	4791	69.5
3	4745	69.1	4735	69.1
4	4877	71	4877	70.5
5	4861	70.1	4867	69.9
6	4724	74.6	4765	74.8
7	4750	72.1	4737	72.1
average	4805	71	4808	71

 

[mtcn77]

You know the reference two negatives does not prove a positive? You are making assumptions based on unknowns.

 

[blu3dragon]

I tried a cinebench run as well and that gave the same results too. Clean boot, quit all apps running in the task bar. Single core 264 w/PBO scaler=x10, 266 w/PBO scaler=x1. Multicore was 2593 vs 2594. I'm actually surprised it was that consistent since normally the scores change a little if I run a 2nd time.

 

[Nexium]

Do not compare these 7nm chips to what we had 5, 10, or 20 years ago. The same or even less electomigration could potentially inflict more degradation on thinner traces and transistors, caused by voltage and high-er current. Even if temp is within limits. Taking scalar on manual, and above X1, is nevertheless an override, hence PBO-verride. Use it wisely and sparingly, I sincerely advise you.

Too much sustained voltage (+higher speed +load = higher sustained current) could worsen things quickly too, as much as too high voltage alone under load. Thats why these 7nm chiplets hit their stock target voltages and speeds shortly and as AMD declares... opportunistically.

I will agree that 5000's curve optimizer is a way better OC, in terms of performance and reliability, than static values that its the only way on 3000... but still its easy to misuse and you should all be cautious as there is no background that connects 7nm ZEN3 (and ZEN2) to anything. Any reference to older gens of chips I find it to be mistargeted and/or irrelevant.
Tell me 1 CPU of the past (pre 7nm) that can be degraded with 1.35V static. Well, ZEN2 can and I'm pointing at this as an example to past references. Its uncharted waters both ZEN2/3.

What OCing users (most) seems not grasping is how and why these CPUs are regulated the way they are. Its my estimate based on what I see on attempted settings.

----------------------------------------------------

This is general and not directly refering to 7nm ZEN but its food for thought for everyone.

Duly noted. I understand the curve optimization overclocking should not be approached with the same the old conventional overclocking ways. But this was just one configuration I was able to get running and stabilized but may not be my final daily setting just yet. I'm still learning and playing with the curve optimizer to compare my current most stable results. I'm may also be closing in to finding a stable curve optimized negative setting that should net me the similar boost clock behavior and frequency without touching and CPU voltage offsets or PBO Scalar by identifying the 2 best cores first, then configuring the curve by individual cores from there. It's taking a long time by even more meticulous trial and error, but should be better optimized this way if I can get it stable somehow.

 

[darrenj]

I have a 5800x on a GA X570i Aorus Pro WIFI costum water loop dual rad in a tiny Ghost Louqe MKIII.
Will try these PBO settings

 

[Kamilz34]

I have mine set on a B550-F to just straight core VID 1.275 and CCX0 Ratio of 46.75. So basically all my cores run to 4.675 GHz and no more. But my Cinebench R23 score is quite good for that at 15780 repeatable, got above 16100 average when I could do -15 undervolt.

It runs at like 66/67C max temps during the run usually with a 280mm Arctic AIO with fans spinning at less than 50%.


I have my PBO at motherboard, I didn't touch anything else as any curve optimizer settings at all, even -5 make my computer not even want to boot up. It used to run at -15 but idk why it won't accept that anymore

Is 1.275v too low for a CPU?

 

[thesmokingman]

I have a 5800x on a GA X570i Aorus Pro WIFI costum water loop dual rad in a tiny Ghost Louqe MKIII.
Will try these PBO settings

You're better off just leaving it at auto or a simple pbo + manual limit aka PPT limit of whatever floats your boat for power draw. Like for ex. 160w on a 5800x is really humming along. That's doable since you're on a full loop.

 

[Kamilz34]

Btw for most I see in this thread, your (perf) core clock is not your actual clock, your effective clock is what the program sees. My Cinebench scores are directly related to the Effective clock and could care less that my (perf) clock is at 5Ghz when the effective is at 4.3GHz. I am no expert, but that is what I see from my testing. Correct me if I'm wrong.

 

[thesmokingman]

Btw for most I see in this thread, your (perf) core clock is not your actual clock, your effective clock is what the program sees. My Cinebench scores are directly related to the Effective clock and could care less that my (perf) clock is at 5Ghz when the effective is at 4.3GHz. I am no expert, but that is what I see from my testing. Correct me if I'm wrong.

Effective clock vs instant (discrete) clock | HWiNFO Forum

Core clocks are used for epeen. Effective are what's actually doing work as you found out.

 

[Kamilz34]

Effective clock vs instant (discrete) clock | HWiNFO Forum

Core clocks are used for epeen. Effective are what's actually doing work as you found out.

Eeeeeee haha I saw some people show their HWiNFO and quote their core clock speeds not the effective. Glad I found something new

 

[blu3dragon]

I have mine set on a B550-F to just straight core VID 1.275 and CCX0 Ratio of 46.75. So basically all my cores run to 4.675 GHz and no more. But my Cinebench R23 score is quite good for that at 15780 repeatable, got above 16100 average when I could do -15 undervolt.

It runs at like 66/67C max temps during the run usually with a 280mm Arctic AIO with fans spinning at less than 50%.


I have my PBO at motherboard, I didn't touch anything else as any curve optimizer settings at all, even -5 make my computer not even want to boot up. It used to run at -15 but idk why it won't accept that anymore

Is 1.275v too low for a CPU?

Since you have manually set VID you have already undervolted. You don't need to use curve offset in this case. Interesting that had an effect though, I would have thought it would be ignored.

Btw for most I see in this thread, your (perf) core clock is not your actual clock, your effective clock is what the program sees. My Cinebench scores are directly related to the Effective clock and could care less that my (perf) clock is at 5Ghz when the effective is at 4.3GHz. I am no expert, but that is what I see from my testing. Correct me if I'm wrong.

As thesmokingman already mentioned, this is true. But now you need to look at why your effective clock is so much lower than actual

Two things I am aware of that can cause this with Zen3:

1) ASUS Fmax enhancer, if enabled try disabling this.
2) Setting a voltage offset that is too low for the cpu.

 

[Illusion]

i am not a native english speaker.


i started investigating many complaints about the 5800x running very hot.

in the last 14 days i tried over 200 different PBO tweaks to get the most out of my Chip.
the 5800X sits on a B550 STRIX-F and is cooled by a Kraken X73.

even with 100% Pump Speed i get like 78-85°C in Cinebench R20... and the reason is pretty clear.
the 5800X has the same power/current limits as a 5950X. (142W PPT, 140A EDC, 95A TDC)
that means the cores consume around twice the power on the 5800X.


after AMD said "this is normal and expected" i started living with it and tweaked it with PBO.

after a specific "benchmark" from me in Rainbow six siege (a custom map with the indentical paths, destroying specific walls with the same operator)

i ran every benchmark 15 times with stock, and the two most successful PBO tweaks.

Results for Rainbow Six Siege:

Stock Mainboard Settings with DOCP on.
all Cores were between 4.35 and 4.5 Ghz with an average all core clockspeed of around 4.43 Ghz. (average VCore 1.403 and 63°C) Average FPS 309

PBO PPT 180 / TDC 140 / EDC 180. 2X Scalar, 100 Mhz Offset.
all cores were between 4.45 and 4.55 Ghz with an average all core clockspeed of around 4.48 Ghz. (average VCore 1.437V and 67°C) Average FPS 313

PBO PPT 200 / TDC 165 / EDC 200. 4X Scalar, 150 Mhz Offset.
all cores were between 4.45 Ghz and 4.58 Ghz with an average all core clockspeed of around 4.52 Ghz. (average VCore 1.465V and 72°C) Average FPS 314


Results with Cinebench R23:
Stock:
1609 SC / 15533 MC (78.8°C 1.331V @ 139W)
4850 Mhz / 4500 Mhz

PBO PPT 180 / TDC 140 / EDC 180. 2X Scalar, 100 Mhz Offset:
1623 SC / 15549 MC (81.5°C 1.34V @ 144W)
4950 Mhz / 4525 Mhz

PBO PPT 200 / TDC 165 / EDC 200. 4X Scalar, 150 Mhz Offset:
1641 SC / 15602 MC (86.1°C 1.365V @ 153W)
4975-5000 Mhz / 4550 Mhz


then i started taking some Power off the CPU and let it do it's thing. (Using PBO with the following settings reduced the performance as much as -200 MHz)

DOCP + Changing Powerlimits in PBO without changing any Scalar or Clockspeed Offset.

PBO PPT 105 / TDC 70 / EDC 90

Rainbow Six Siege:

all cores were between 4.58 Ghz and 4.72 Ghz with an average all core clockspeed of around 4.67 Ghz. (average VCore 1.38V and 61°C) Average FPS 324

Cinebench R20:
1611 SC / 15091 MC (64.8°C 1.2V @ 105W)
4850 Mhz / 4425 Mhz



The results are the biggest improvement so far.
Gaming and other lighter Tasks are much faster than PBO while consuming way less power.
Only the Multicore performance is around 3.5% lower while consuming around 33% less power.

i am currently trying this in other applications like Blender and other/heavier Videogames like Battlefield V and Cyberpunk 2077.

i highly recommend testing this for yourself

So, what is your VCore voltage set to? Auto? Also, what is the VCore set to when doing curve optimizer? Thanks!

 

[Waldorf]

Got pretty good gains using the clock tuner from 1usmus.
no numbers yet, was just messing with the tool prior to clean win install..

go thru the guide, even if you used prior CTRs.
CTR 2.0

 

[brandofriva]

Got pretty good gains using the clock tuner from 1usmus.
no numbers yet, was just messing with the tool prior to clean win install..

go thru the guide, even if you used prior CTRs.
CTR 2.0

Ye this gave me vastly improved numbers too.
You can however still get a pretty good improvement just from PBO etc as in the rest of this thread.

 

[Waldorf]

it makes use of pbo. just switch to enabled prior to tuning.

 

[blu3dragon]

PBO + manual tuning of curve offset gives me higher clocks than CTR 2.0. CTR uses lower voltages so less heat and power though. I need to play with it some more to figure the right balance and also understand how it works when there is a heavy avx load. It is nicely written and a lot less effort than manual tuning.

 

[Waldorf]

@bluefrisky
unless you have a really crappy block, doubt the offset bracket will make any impact,
as almost all decent aio/blocks will cover the full HS.

when i look at my 30$ block (clear), it shows the channels cover whole (cpu) HS,
and when i took a friends pump apart (240 eisbaer) for cleaning, same thing.
water channels are covering the whole HS, so the bracket wont do anything.