5600X Overclocking

[ivanrcks95]

Have had this CPU for 3 years now, and always thought that leaving it with PBO enabled was probably the best way forward, it didnt help that I had a pretty low tier board as well and didnt want to risk it. I have since upgraded to a B550 Tomahawk, which is regarded as a pretty nice board for OC and decided to play around with the chip. These are the results.

 

I run a 280mm AIO, and I have the pump and fans set to hit 50% MAX, so its dead silent, as I prioritize quiet gaming, so if the temps seem "high" for a 280mm AIO, this is why.

 

PBO Stock - chip hits about 4600-4650mhz all core, and monitoring voltage it hovers around 1.34-1.36v, R23 score was 11787 and the max temp was 80c.

Manual Stable OC - 4700mhz all core, voltage at 1.275v, R23 score goes up to 12067 and the max temp drops down to 70c

Manual Stable OC - 4750mhz all core, voltage at 1.3v, R23 score goes up again to 12330 and the max temp is now 72c

Manual Stable OC - 4800mhz all core, voltages at 1.35v, R23 score goes up again to 12359 and the max temp is now 78c

 

The last 4.8ghz OC doesnt seem to be worth it, based on temps and voltage, and the scores not seeing that big of a boost. I really like the gains seen with the 4.7ghz all core OC, not only the 3-5% uplift in R23/TimeSpy, but also the 10c temperature drop due to the voltage. What do you guys think? Worth keeping that manual overclock? Or go back to PBO? From all the research ive done, anything  under 1.3v is considered "safe" for the longevity of 5000 series, and considering it goes over 1.34v with stock PBO, seems like its a no brainer right? Are there any background/"invisible" performance points I might be loosing going to manual overclock instead of PBO? 

[Ottoman420]

This is old school oc where you are forcing the chip to run at max clock 24/7 the scores on Cinebench look good though.

 

It will "wear" out the chip faster and cost more to run daily.

 

You should try PBO Tuning and you can reach the same boost clocks and similar scores(i think) but the chip will downclock at idle and save you money in the long run on power and life of the part

 

Set PBO to manual experiment with PPT & EDC and set curve optimiser to "per core" and set every core with a "negative" offset of up to -30 and put the pbo boost to 200mhz. you will still reach 4.850ghz

 

 

Check this out

 

[ivanrcks95]

2 minutes ago, Ottoman420 said:

This is old school oc where you are forcing the chip to run at max clock 24/7 the scores on Cinebench look good though.

 

It will "wear" out the chip faster and cost more to run daily.

 

You should try PBO Tuning and you can reach the same boost clocks and similar scores(i think) but the chip will downclock at idle and save you money in the long run on power and life of the part

 

Set PBO to manual experiment with PPT & EDC and set curve optimiser to "per core" and set every core with a "negative" offset of up to -30 and put the pbo boost to 200mhz. you will still reach 4.850ghz

 

Yeah its the system I grew up on lol 

 

My main concern would be temps, more than the scores, frankly I would be very happy if it got the same scores as it does on stock PBO, if it mean a 10c drop, I havent done much digging into PBO tuning, is there a way to limit voltage using it as well to make sure I can have the same low temps I am seeing with manual OC?

[Ottoman420]

25 minutes ago, ivanrcks95 said:

Yeah its the system I grew up on lol 

 

Same^

 

You will be pleasantly suprised i think

 

With PBO you can both overclock and undervolt at the same time.

 

PPT is closely related to CPU wattage its not exact but its close

TDP is the max tempreture allowed

EDC is something like ampage.

 

So i give you some examples on my 5600x 280mm AIO I had at the max PPT 110 TDC 90 EDC 124 with a cinebecnh score similar to yours and currently i have PPT 75 TDC 90 EDC 100 with a cinebench score above 10,600, this still could be tuned better but its fine.

 

Someone here has a PPT of 135 on their 5600x and a cinebench score of 12600 so it all depends on what you want from your chip and how long you want to tune.

 

Now both these ppts/edcs are providing the same clock speeds but the cpu temp on the PPT 75 is never going above 60c but unless im benchmarking...i cant really tell the difference in day to day and gaming fps is....near identical but the temps are great and i dont feel that i have to run fans loud.

 

So i guess have fun. Ask any questions you might have and sorry if i didnt explain it very well.

 

 

 

[ivanrcks95]

49 minutes ago, Ottoman420 said:

Same^

 

You will be pleasantly suprised i think

 

With PBO you can both overclock and undervolt at the same time.

 

PPT is closely related to CPU wattage its not exact but its close

TDP is the max tempreture allowed

EDC is something like ampage.

 

So i give you some examples on my 5600x 280mm AIO I had at the max PPT 110 TDC 90 EDC 124 with a cinebecnh score similar to yours and currently i have PPT 75 TDC 90 EDC 100 with a cinebench score above 10,600, this still could be tuned better but its fine.

 

Someone here has a PPT of 135 on their 5600x and a cinebench score of 12600 so it all depends on what you want from your chip and how long you want to tune.

 

Now both these ppts/edcs are providing the same clock speeds but the cpu temp on the PPT 75 is never going above 60c but unless im benchmarking...i cant really tell the difference in day to day and gaming fps is....near identical but the temps are great and i dont feel that i have to run fans loud.

 

So i guess have fun. Ask any questions you might have and sorry if i didnt explain it very well.

 

 

 

Okay, the actual PPT, TDC and EDC values, and how they work between one another is quite confusing. However, the curve optimizer for PBO is pretty straight forward and its what ive ended up using.

 

Pretty much setting everything to auto, except PBO boost to +100mhz and curve optimizer to a -20 offset. This gets me 4.7-4.75ghz, whilst the voltage fluctuates between 1.28-1.31v, so not far off my manual OC settings. Temps max out at 74c and the score is at 12,119, a bit lower than manual OC but that is to be expected as PBO doesnt hold clocks, but simply fluctuates within parameters. And as you said, on idle/low load, it drops back down to 3.4/3.5ghz with 0.9v.

 

Definitely liking this system more. Going to play around to see if i can still have it stable at an even lower voltage and call it a day.

[Ottoman420]

 

Spoiler

 

19 minutes ago, ivanrcks95 said:

Okay, the actual PPT, TDC and EDC values, and how they work between one another is quite confusing. However, the curve optimizer for PBO is pretty straight forward and its what ive ended up using.

 

Pretty much setting everything to auto, except PBO boost to +100mhz and curve optimizer to a -20 offset. This gets me 4.7-4.75ghz, whilst the voltage fluctuates between 1.28-1.31v, so not far off my manual OC settings. Temps max out at 74c and the score is at 12,119, a bit lower than manual OC but that is to be expected as PBO doesnt hold clocks, but simply fluctuates within parameters. And as you said, on idle/low load, it drops back down to 3.4/3.5ghz with 0.9v.

 

Definitely liking this system more. Going to play around to see if i can still have it stable at an even lower voltage and call it a day.

I had my curve set to -30 on all cores which caused crashes and WHEA errors, now i have -25 on 4 cores and -20 on my best cores (some people suggest -10) but yes it takes playing around I would also put the boost to 200mhz.

 

Check Ryzen Master to find out your best 2 cores, Use HWinfo to check the PPT/EDC values during cinebench runs, i found that EDC needed to be higher than PPT and if EDC hit 100% in a benchmark i would raise it by 1 untill i saw 99% and this is where i saw the boost in scores.

 

TBH i found/find the ppt stuff confusing as well but i also think this is where your best pbo tune will come from.

 

See the screenshot(sorry for size) my edc only hits 86% so on this 75ppt tune i could drop edc from 100 to around 95 to get that 99% edc in hwinfo for a better score

 

 

 

[ivanrcks95]

40 minutes ago, Ottoman420 said:

 

  Reveal hidden contents

 

I had my curve set to -30 on all cores which caused crashes and WHEA errors, now i have -25 on 4 cores and -20 on my best cores (some people suggest -10) but yes it takes playing around I would also put the boost to 200mhz.

 

Check Ryzen Master to find out your best 2 cores, Use HWinfo to check the PPT/EDC values during cinebench runs, i found that EDC needed to be higher than PPT and if EDC hit 100% in a benchmark i would raise it by 1 untill i saw 99% and this is where i saw the boost in scores.

 

TBH i found/find the ppt stuff confusing as well but i also think this is where your best pbo tune will come from.

 

See the screenshot(sorry for size) my edc only hits 86% so on this 75ppt tune i could drop edc from 100 to around 95 to get that 99% edc in hwinfo for a better score

 

 

 

Seems like a +100mhz and a -20 offset is the best it can do. -25 crashes straight away at +100mhz. At +125mhz with -20, while it is stable for 10min of R23, crashed while using chrome.

 

Currently stress testing +100mhz with -20 with 10min R23, Prime 95 and then just desktop use, if its stable in all 3, thats probably what I'll end up using, as it keeps clocks above the stock 4650mhz, even when heat soaked, whilst being under 1.3v, which drops stress test temps from 83c to 75-76c at the same pump and fan curves, while having higher scores overall. 

 

If it cant be stable at +100mhz with a -20, ill go down to +50mhz or +75mhz at -20, just so that I can at least match/slightly beat stock PBO performance but do it at significantly lower temps.

[ivanrcks95]

I had to go down to +50mhz and a -20 offset, this is the only time that it passed every single stress test and normal testing. 

 

12042 as the best score at R23 with maxing out at 74c.

After a 10minute R23 test, the result dropped down to 11096, with a max temp of 75.4c.

Passed both TimeSpy and FireStrike

Passed 20minutes of Prime95 with a full "blend" load. This was by far the most demanding test, but even still, at the quiet fan and pump profile it only hit 80.8c max after 20minutes. No crashes.

 

All of the scores were a 3-5% improvements overall compared to the stock PBO results. 

 

With these settings, the voltage fluctuates between 1.275 and 1.281v, however, what I did notice is that with these settings the core clocks are literally LOCKED to 4.7ghz, they dont budge at all, even after 10min of R23 it still hold strong at 4.7ghz. Whereas with the +100mhz and -20 settings, the clocks do spike to 4.75ghz, and stay between 4.7 and 4.75 when doing a single test, however they fluctuate quite a lot when doing a stress test, especially as temperature goes up, and most of the time they were actually below 4.7ghz, hovering around 4.65-4.68.

 

This is reflective as to why the scores and temperatures between +50mhz and +100mhz runs were so similar, after the 10min R23 test the +100mhz run was at 11106, a 0.09% difference compared to the 10min +50mhz run. The +50mhz setting just manages to be more stable and lock the clock at 4.7ghz more often.

 

Ill keep testing day to day and gaming use with these settings and see how it goes, but as of right now I am happy. Small boost in performance, decent drop in temperatures over stock PBO, exactly what I wanted.