Overclock.net banner
521 - 540 of 2,009 Posts
Windows version 21H1 OS build 19043.1165. I don't really have a "favorite" version of Windows 10, I usually just keep it up to date. I know that sometimes Windows updates can cause system issues and even loss of performance however I started keeping it up to date for the security patches. So far I have been lucky and haven't had any issues with keeping Windows up to date.
Thanks..will have a look into that build..
 
anyone noticed if the EDC is lower than 140 the boost will be higher?
I did some testing with EDC from 140 to 200 and basically all my scores/boost are same.
I tried EDC 125 - 130 and all single scores are higher and boost is also higher
I wonder if is motherboard issue or this behavior is normal ?
 
anyone noticed if the EDC is lower than 140 the boost will be higher?
I did some testing with EDC from 140 to 200 and basically all my scores/boost are same.
I tried EDC 125 - 130 and all single scores are higher and boost is also higher
I wonder if is motherboard issue or this behavior is normal ?
Normal behavior. This has been talked about since near launch of Zen 3. You have to find the EDC sweet spot. I use 170 for my 5900x as this balanced single core and mutli-core. Low EDC will hurt multi-core.
 
anyone noticed if the EDC is lower than 140 the boost will be higher?
I did some testing with EDC from 140 to 200 and basically all my scores/boost are same.
I tried EDC 125 - 130 and all single scores are higher and boost is also higher
I wonder if is motherboard issue or this behavior is normal ?
I just did some testing in r23 single. I set affinity to one thread on my best core and watched the core effective clock. At my usual 155 EDC, that core's effective clock averages 4842.5. Lowering to 125EDC raises it to 4866.1 effective clock average. Boost was 50mhz higher, but overall a difference of less than 24mhz avg doesn't make this a worthwhile tweak for everyday use. Too much is lost overall. YMMV of course.
 
Here you go. Different results at different EDC for me. PPT and TDC are set too high to be reached.
I wanted to re-run these anyway after finishing my curve optimizer tuning, so thanks for the motivation. Best EDC for my setup didn't change with changes to the PBO curve.

This bunch is with C23
Please note: Scores are for comparison to each other only. Tests were done with ryzen master & HWiNFO64 open, corsair icue, and armoury crate running my absolutely unnecessary LEDs. Having those all running is a 1200-1800 point loss.

EDC-MHZ - Temp (Tctl/Tdie) - Score (best of 3 runs, 1 minute between to allow cpu to cool.)
125 - 4330 - 74.0 - 26938
135 - 4370 - 78.9 - 27293
145 - 4401 - 82.9 - 27438
150 - 4414 - 84.9 - 27446
155 - 4418 - 87.0 - 27695 *This is where the CPU runs right up before throttling, it allows it to run a bit hotter than higher EDC.
160 - 4409 - 86.3 - 27669
165 - 4402 - 86.3 - 27391
No point in graphing higher, they just get progressively worse in all 3 categories.
2.8% difference from best to worst.

This Bunch with Prime95 Small AVX2 enabled
EDC-MHZ - Temp (Tctl/Tdie)
125 - 3686 - 65.2
135 - 3771 - 68.0
145 - 3855 - 71.5
150 - 3917 - 73.3
155 - 3955 - 74.5
160 - 3982 - 76.6
165 - 3976 - 77.2
8% difference from best to worst.

Multi core CPU-z bench
135 - 12735.4
145 - 12874.1
150 - 12880.5
155 - 12865.5
160 - 12859.5
165 - 12834.4
175 - 12808.9
1% difference from best to worst. (Essentially standard deviation/margin of error stuff here.)

Single core CPU-Z bench (set to single thread to not heat soak the cpu during multi thread test. ...Sorry, I wasn't sitting through 7 single core R23s at this point.)
115 - 659.3
125 - 658.3
135 - 656.6
145 - 656.9
155 - 655.5
165 - 654.3
175 - 653.1
<1% difference from best to worst... (Again, Standard deviation/margin of error range.)

As you can see, not worth losing multi-core for single core. I suggest you do your own testing though, so many factors determine optimal settings.
 
Here you go. Different results at different EDC for me. PPT and TDC are set too high to be reached.
I wanted to re-run these anyway after finishing my curve optimizer tuning, so thanks for the motivation. Best EDC for my setup didn't change with changes to the PBO curve.

This bunch is with C23
Please note: Scores are for comparison to each other only. Tests were done with ryzen master & HWiNFO64 open, corsair icue, and armoury crate running my absolutely unnecessary LEDs. Having those all running is a 1200-1800 point loss.

EDC-MHZ - Temp (Tctl/Tdie) - Score (best of 3 runs, 1 minute between to allow cpu to cool.)
125 - 4330 - 74.0 - 26938
135 - 4370 - 78.9 - 27293
145 - 4401 - 82.9 - 27438
150 - 4414 - 84.9 - 27446
155 - 4418 - 87.0 - 27695 *This is where the CPU runs right up before throttling, it allows it to run a bit hotter than higher EDC.
160 - 4409 - 86.3 - 27669
165 - 4402 - 86.3 - 27391
No point in graphing higher, they just get progressively worse in all 3 categories.
2.8% difference from best to worst.

This Bunch with Prime95 Small AVX2 enabled
EDC-MHZ - Temp (Tctl/Tdie)
125 - 3686 - 65.2
135 - 3771 - 68.0
145 - 3855 - 71.5
150 - 3917 - 73.3
155 - 3955 - 74.5
160 - 3982 - 76.6
165 - 3976 - 77.2
8% difference from best to worst.

Multi core CPU-z bench
135 - 12735.4
145 - 12874.1
150 - 12880.5
155 - 12865.5
160 - 12859.5
165 - 12834.4
175 - 12808.9
1% difference from best to worst. (Essentially standard deviation/margin of error stuff here.)

Single core CPU-Z bench (set to single thread to not heat soak the cpu during multi thread test. ...Sorry, I wasn't sitting through 7 single core R23s at this point.)
115 - 659.3
125 - 658.3
135 - 656.6
145 - 656.9
155 - 655.5
165 - 654.3
175 - 653.1
<1% difference from best to worst... (Again, Standard deviation/margin of error range.)

As you can see, not worth losing multi-core for single core. I suggest you do your own testing though, so many factors determine optimal settings.
Here is what I used, optimized for multicore, but I have a really decent chip for Curve Optimizer, and it's Core Cycler stable as well.

2522381




2522382



2522383


2522384




2522385


2522386
 
  • Rep+
Reactions: TimeDrapery
Here you go. Different results at different EDC for me. PPT and TDC are set too high to be reached.
I wanted to re-run these anyway after finishing my curve optimizer tuning, so thanks for the motivation. Best EDC for my setup didn't change with changes to the PBO curve.

This bunch is with C23
Please note: Scores are for comparison to each other only. Tests were done with ryzen master & HWiNFO64 open, corsair icue, and armoury crate running my absolutely unnecessary LEDs. Having those all running is a 1200-1800 point loss.

EDC-MHZ - Temp (Tctl/Tdie) - Score (best of 3 runs, 1 minute between to allow cpu to cool.)
125 - 4330 - 74.0 - 26938
135 - 4370 - 78.9 - 27293
145 - 4401 - 82.9 - 27438
150 - 4414 - 84.9 - 27446
155 - 4418 - 87.0 - 27695 *This is where the CPU runs right up before throttling, it allows it to run a bit hotter than higher EDC.
160 - 4409 - 86.3 - 27669
165 - 4402 - 86.3 - 27391
No point in graphing higher, they just get progressively worse in all 3 categories.
2.8% difference from best to worst.

This Bunch with Prime95 Small AVX2 enabled
EDC-MHZ - Temp (Tctl/Tdie)
125 - 3686 - 65.2
135 - 3771 - 68.0
145 - 3855 - 71.5
150 - 3917 - 73.3
155 - 3955 - 74.5
160 - 3982 - 76.6
165 - 3976 - 77.2
8% difference from best to worst.

Multi core CPU-z bench
135 - 12735.4
145 - 12874.1
150 - 12880.5
155 - 12865.5
160 - 12859.5
165 - 12834.4
175 - 12808.9
1% difference from best to worst. (Essentially standard deviation/margin of error stuff here.)

Single core CPU-Z bench (set to single thread to not heat soak the cpu during multi thread test. ...Sorry, I wasn't sitting through 7 single core R23s at this point.)
115 - 659.3
125 - 658.3
135 - 656.6
145 - 656.9
155 - 655.5
165 - 654.3
175 - 653.1
<1% difference from best to worst... (Again, Standard deviation/margin of error range.)

As you can see, not worth losing multi-core for single core. I suggest you do your own testing though, so many factors determine optimal settings.
even so, with how tight amd auto regulation of clocks with zen. those differences can be just run time variations. it is impossible to determine which is best edc settings from such tests.
 
Putting my new Arctic LF II 420mm to the test! Not bad for just PBO and +50 AutoOC.
2522440
2522441
 
EDC 130A worked best for me, I get the highest MT in CB23 compared to 150A. This is with 170W PPT 120W EDC
There is something getting me mad is the single thread boost is not fixed, like sometimes CB23 ST can run the bench with boost up to 5GHz, sometimes 4.95GHz, sometimes 4.9GHz, sometimes 4.8GHz. Same for CPUZ some times my single score bumps to 700 ( I never was able to get 700 again),sometimes 690, sometimes 680.
Even in games sometimes boost to 4.9, sometimes 4.8 in the same game in same areas. (That's all with same settings)
Something in this CPU is playing around, and I am sure it is not from background processes.
And yes I made sure it is the same core being benched in all single core tests.


2522483
2522484
2522485
2522486
 
Need help for first time AMD user. I tried to oc ram and pc does not post at any setting I use. Flck 1800, ram 3600. Dram voltage 1.55v soc voltage is 1.25v.
Your DRAM and SOC voltages are too high. You might want to ask here . . .


At 3600MHz just lv flck alone. May have to loosen the timings a bit to work. Like if the RAM is designed to run at 16,18,18, 21, for example, try 16,19,19, 39.

BTW, if you are the one who sold me an Asus AM3 motherboard years ago, i like to tell you that it still works paired with a Phenom II 840T used as my HTPC :)
 
Someone shared in another thread what PBO settings they are using.

This nets me the below. Passes Core Cycler too. BRB, gonna boot into BIOS, get screenshots.

The Curve offsets are what I know works Core Cycler stable and so far with new settings, stable as well

2522975


2522967


2522970


2522971


2522972


2522973


2522974
 
Someone shared in another thread what PBO settings they are using.

This nets me the below. Passes Core Cycler too. BRB, gonna boot into BIOS, get screenshots.

The Curve offsets are what I know works Core Cycler stable and so far with new settings, stable as well

View attachment 2522966

View attachment 2522967

View attachment 2522970

View attachment 2522971

View attachment 2522972

View attachment 2522973

View attachment 2522974
what's your ambient temps? lowering it helps a lot too
 
what's your ambient temps? lowering it helps a lot too
I'm maybe at 68F/20C, I forgot to turn my A/C off when I left for work.
 
When i get whea error in event viewer APIC ID 10,okay its core 5 id,but which value in bios under curve optimizer is to change, direct number 5 core id as i see (0-11 are values msi mobo) or Core 4 since core 4 in bios is core 5 in ryzen master
 
521 - 540 of 2,009 Posts