[metalshark]

Looks pretty old. Maybe it was a bug in an older bios version, or maybe you got that by increasing the cpu current capability setting instead of using the PBO limits.
Anyway I'll give it a try this weekend and will post here any update.

LOL - same with 3801 - don't worry you'll find many pushing beyond the default limits on the motherboard. Maybe not PPT but EDC/TDC most definitely. Checkout Buildzoid, etc

 

[Sleepycat]

My Vcore is on auto and soc@1,1V entered value in bios.

You mean I should try LLC3 on both ?

Yes, I currently use:
Vcore: Auto
Vcore LLC: 3
VSOC: 1.09V
VSOC LLC: 3

I recommend setting both Vcore LLC and VSOC LLC to 3. It doesn't seem to overshoot much when under all core loads. I also don't notice problems with single core voltage since the load is probably much lower and the effect of LLC is much less than all core.

 

[GRABibus]

It used to be, ill test it and post back
Arctic freezer II 360 with rev 3.0 offset mounting & 3x ARCTIC BioniX P120 A-RGB fans

EDIT
It's stable but hot because of the GPU... this thing at full throttle is a power hog

Spoiler: .

Yes, I currently use:
Vcore: Auto
Vcore LLC: 3
VSOC: 1.09V
VSOC LLC: 3

I recommend setting both Vcore LLC and VSOC LLC to 3. It doesn't seem to overshoot much when under all core loads. I also don't notice problems with single core voltage since the load is probably much lower and the effect of LLC is much less than all core.

Sorry but it is unstable.
I crash in Realbench in minutes.

Either CPU silicon limitation or not enough voltage delivered to CPU when it boosts high.
In this last case, a much beter colling would maybe help.

 

[Enferlain]

I made a test just now. I have a C8F and using BIOS v3801.
By setting PBO limits to Motherboard, I read on HwInfo the limits being 395W/255A/200A for PPT/TDC/EDC .
By setting PBO limits to manual 600/600/600, I still read on HwInfo the limits being 395W/255A/200A for PPT/TDC/EDC .
It doesn't seem to be able to surpass those limits, so I stand on my statement being correct in my case.

It gets overwritten when you use asus oc for pbo. That's where I set 740 480 650. Which unfortunately I'll probably have to do again. I'm not really satisfied with the scores I get but having lower temps is better than higher scores for me, I'm not trying to push for a world record or anything. Kinda annoying considering everything worked properly on a x370 board.

Anything other than the above 3 limits results in 38-41 idle temps and 60 spikes for my use case, when before I was able to run an emulator, a game, chrome, and a lot of background apps wit the cpu keeping at 44c without issues and would idle at low to mid 30s. I've even seen below 30c when the room temp was really low. Some weird **** going on. Maybe I'll try undervolting.

 

[Kelutrel]

It gets overwritten when you use asus oc for pbo. That's where I set 740 480 650. Which unfortunately I'll probably have to do again. I'm not really satisfied with the scores I get but having lower temps is better than higher scores for me, I'm not trying to push for a world record or anything. Kinda annoying considering everything worked properly on a x370 board.

Anything other than the above 3 limits results in 38-41 idle temps and 60 spikes for my use case, when before I was able to run an emulator, a game, chrome, and a lot of background apps wit the cpu keeping at 44c without issues and would idle at low to mid 30s. I've even seen below 30c when the room temp was really low. Some weird **** going on. Maybe I'll try undervolting.

Sir, the Asus PBO bios page doesn't work, it doesn't do what you tell it to do.
When you set any manual limit it leaves your cpu at stock pbo/co limits, that is why you see the cpu having lower temperature and having lower performances than expected.

 

[Enferlain]

Sir, the Asus PBO bios page doesn't work, it doesn't do what you tell it to do.
When you set any manual limit it leaves your cpu at stock pbo/co limits, that is why you see the cpu having lower temperature and having lower performances than expected.

The results with asus pbo are different than stock pbo though. Asus tab is the only one that correctly sets the limits I type in. Amd tab just defaults to motherboard values

I'm trying my best to replicate these results but it seems really hard even with ambient temps being like 10c down compared to when it was made.

 

[Kelutrel]

The results with asus pbo are different than stock pbo though. Asus tab is the only one that correctly sets the limits I type in. Amd tab just defaults to motherboard values


View attachment 2527130

I'm trying my best to replicate these results but it seems really hard even with ambient temps being like 10c down compared to when it was made.

Are all these 5 results from the same motherboard and Windows 10 ?

 

[Enferlain]

Are these results all on the same motherboard and Windows 10 ?

Same motherboard, yes, and no it's Win 11, which is pretty much my fallback excuse if I don't end up being able to get the results I want.

Edit: The screenshot outside of the quotes is on my previous motherboard, x370 taichi

 

[Kelutrel]

Same motherboard, yes, and no it's Win 11, which is pretty much my fallback excuse if I don't end up being able to get the results I want.

The 9234 result you are trying to match, was it also taken on Windows 11 ?

I ask this because there is a known L3 cache issue in Windows 11 with AMD cpus that may make that result unmatchable until that issue is fixed. You can see the problem by running the AIDA64 cache benchmark on Windows 11.

 

[sonixmon]

At least the public versions are free. I was on the free CTR 2.1 for a very long time. It got what I needed. It was lower scores in CB R20/23 compared to PBO Advanced, but the temperatures were much lower. I use it to lower temperatures as I am on air cooling and not chasing benchmark scores now.

But in the end, there are a number of different ways of increasing your performance by paying for parts or software, so as long as we have the choice to decide on where to spend our money, then that is good.

Agree, I have been supporting his work for two months and have been happy with Hydra. Since I could not find a Dark Hero version available when I was purchasing I settled for the original CH8. Hydra actually mimics the DH hybrid OC and works well for me. It provided me an improvement in scores and cooler temps (especially while gaming). I went from mid to high 50s while Gaming to low to mid 40s. My idle temp also dropped a few degrees. CTR helped me some to lower my temps some but didn't improve scores much.
He has added several features and I am still on 1.0a even though b was just released I haven't grabbed it yet. I figure another month or so and public version should be out. Or if I get a good beta version then I will stop there until official release. Worth it to me to give him a few bucks for his work.

 

[NotReadyYet]

I'm getting much better TimeSpy scores with CTR than Hydra. Is there a way to import CTR values into Hydra?

 

[Sleepycat]

I'm getting much better TimeSpy scores with CTR than Hydra. Is there a way to import CTR values into Hydra?

There are situations where CTR is easier and more straightforward than Hydra. You can think of CTR as a multi-profile static OC which changes based on load. Hydra is more of a multi-profile PBO which changes based on load, triggers and allows you to set a per core CO to slow areas down in case of instability.

For benchmark results, CTR gave me better scores and lower temperatures because I could easily just enter the clocks I wanted for benchmarks.

For normal usage, Hydra seems to give me lower power consumption, lower temperatures and an easy way to quickly make a curve compensation when something crashes. So based on what I have used for my own PC, I can't say that one is better than the other. I'm probably more comfortable with CTR, but do appreciate not having to guess if I need to adjust Px High, Px Mid, Px Low, P2 or P1 when I have a game or display driver crash.

 

[noxious89123]

If it helps, my understanding is that PBO limits will never go over the motherboard max limits, independently from any value that you may be able to input.

I made a test just now. I have a C8F and using BIOS v3801.
By setting PBO limits to Motherboard, I read on HwInfo the limits being 395W/255A/200A for PPT/TDC/EDC .
By setting PBO limits to manual 600/600/600, I still read on HwInfo the limits being 395W/255A/200A for PPT/TDC/EDC .
It doesn't seem to be able to surpass those limits, so I stand on my statement being correct in my case.

On my CH8DH, I can go above 395W PPT / 255A TDC / 200A EDC but only if I input the values in the Asus PBO section of the BIOS, but not if I set it through the AMD section of the BIOS. I can set it higher in the AMD section, but it won't actually go above those limits. It behaved this way on earlier BIOS and on 3801 which I am using now.

With that said, increasing them above the values I stated does not improve performance. I've not tested it so I can't say for certainty, but I expect higher values may actually hurt performance. Plenty of users on OCN have reported higher CB R20 multi core scores when using lower values. I think this is due to lower temperatures allowing higher sustained clock speeds.

Think you'll find the TDC/EDC are the other way round. Have for instance gone way above 300A on EDC.

My experience aligns with Kelutrels, although I'm using a CH8DH not the CH8F. For some reason the standard motherboard limit for TDC (255A) is higher than the limit for EDC (200A), which does seem backwards.

As above though, this can be surpassed by setting the limits in Asus' own PBO section of the BIOS instead of the AMD Overclocking section.

 

[Enferlain]

The 9234 result you are trying to match, was it also taken on Windows 11 ?

I ask this because there is a known L3 cache issue in Windows 11 with AMD cpus that may make that result unmatchable until that issue is fixed. You can see the problem by running the AIDA64 cache benchmark on Windows 11.

No, it was win 10 from a year ago. The L3 cache issue I have been thinking about, but some people are saying it's only an aida readout issue so I wasn't sure. Either way I don't really want to go through another reinstall at this point so I'll just wait until it gets fixed and see if it's better then and maybe try out a - offset and some other options meanwhile. Still haven't found the best pbo limits either and have to test ram so got things to do.

 

[NotReadyYet]

There are situations where CTR is easier and more straightforward than Hydra. You can think of CTR as a multi-profile static OC which changes based on load. Hydra is more of a multi-profile PBO which changes based on load, triggers and allows you to set a per core CO to slow areas down in case of instability.

For benchmark results, CTR gave me better scores and lower temperatures because I could easily just enter the clocks I wanted for benchmarks.

For normal usage, Hydra seems to give me lower power consumption, lower temperatures and an easy way to quickly make a curve compensation when something crashes. So based on what I have used for my own PC, I can't say that one is better than the other. I'm probably more comfortable with CTR, but do appreciate not having to guess if I need to adjust Px High, Px Mid, Px Low, P2 or P1 when I have a game or display driver crash.

This is a great summary - thank you. Do you have an idea as to how I could possibly put in my Game Profiles clocks and voltages into Hydra from CTR? Even maybe manually?

 

[Nizzen]

There are situations where CTR is easier and more straightforward than Hydra. You can think of CTR as a multi-profile static OC which changes based on load. Hydra is more of a multi-profile PBO which changes based on load, triggers and allows you to set a per core CO to slow areas down in case of instability.

For benchmark results, CTR gave me better scores and lower temperatures because I could easily just enter the clocks I wanted for benchmarks.

For normal usage, Hydra seems to give me lower power consumption, lower temperatures and an easy way to quickly make a curve compensation when something crashes. So based on what I have used for my own PC, I can't say that one is better than the other. I'm probably more comfortable with CTR, but do appreciate not having to guess if I need to adjust Px High, Px Mid, Px Low, P2 or P1 when I have a game or display driver crash.

So Hydra doesn't help for better performance over CTR?
Looks like Hydra is messing up read and write numbers on Aida64. Numbers are WAY higher than theoretical possible. What does Hydra actual do?

 

[pantsoftime]

I found some value in using Hydra to identify usable CO settings. I have one of those CPUs that doesn't really run well with a negative offset on hardly any cores at all, and it was quite useful for dialing things in. If anyone decides to try this you have to take the numbers that Hydra gives you for CO and divide by 6 for BIOS values.

 

[SpeedyIV]

I found some value in using Hydra to identify usable CO settings. I have one of those CPUs that doesn't really run well with a negative offset on hardly any cores at all, and it was quite useful for dialing things in. If anyone decides to try this you have to take the numbers that Hydra gives you for CO and divide by 6 for BIOS values.

I don't recall the Hydra PDF manual saying to divide Hydra CO numbers by 6 to calculate what to type into the BIOS. Is that "official" or something that you discovered?

 

[pantsoftime]

I don't recall the Hydra PDF manual saying to divide Hydra CO numbers by 6 to calculate what to type into the BIOS. Is that "official" or something that you discovered?

I didn't discover it myself. One of the senior members of the 1usmus Discord mentions it frequently when people ask. I tried it and it did seem to work.

 

[Sleepycat]

Sorry but it is unstable.
I crash in Realbench in minutes.

Either CPU silicon limitation or not enough voltage delivered to CPU when it boosts high.
In this last case, a much beter colling would maybe help.

That's odd since LLC increases voltage to the CPU, not decrease. So it shouldn't crash if you increase LLC for the Vcore.