Overclock.net banner

1541 - 1560 of 1560 Posts

·
Registered
Joined
·
421 Posts
Sure, I did, I've got it working great, I guess my post was more of a "is everyone's chips still doing fine months later?".
Well, I can report, that my chip(3900X) is doing well - after 10-11 months or so on EDC=1 - cooled with waterloop and LM as TIM.

PPT=500
TDC=230
EDC=1
Scalar=5X
Overboost=0 MHz

Boots up to 4,65 GHz on two and up to 4,6 GHz on the other 4-core ot the first CCX.

CB20 multi is around 7700 with these settings...

Best regards,
Medizinmann
 

·
Registered
Joined
·
467 Posts
Sure, I did, I've got it working great, I guess my post was more of a "is everyone's chips still doing fine months later?".
I can't complain...after watching "der8auers" Video where he cooled his CPU down to 5-10°C and the algorithm was auto applying 1.76V, i completely lost my fear that Zen CPUs are fragile...

Running +0.05V offset with EDC now, boost voltages up to 1.6V in low load scenarios...no issues over the past months (and was using EDC=1 from first day on, before).

Did you try disabling C-States regarding your Idle reboots? I had the same issue for a long time.
 

·
Registered
Joined
·
85 Posts
I can't complain...after watching "der8auers" Video where he cooled his CPU down to 5-10°C and the algorithm was auto applying 1.76V, i completely lost my fear that Zen CPUs are fragile...

Running +0.05V offset with EDC now, boost voltages up to 1.6V in low load scenarios...no issues over the past months (and was using EDC=1 from first day on, before).

Did you try disabling C-States regarding your Idle reboots? I had the same issue for a long time.
It's not any common or even "normal" behaviour if it happened it was some serious bug or some bios issue. I can't imagine it was triggered by the cpu, heard pretty nonsense to me.
I did some low temp tests in the past and never seen anything, just partly similar, crazy voltage growth for decreasing temps.
Used the EDC bug for pretty long period, however a bit different point of view. Used it because the more aggressive boosting behaviour but not the fully max-out way. Positive offset wasn't acceptable for me as 24/7, not for such a small gain, like another 50 Mhz. It was boosting far higher and better than stock with negative offset even. That was the way I prefferred those days. My cpu lived all its life more or less above stock specs. When I ended up the EDC trick, I did that because bclk oc was corrected and enabled again. So I was moving to another way of tune but the main thing is my cpu is still working just as it did in the first moments. Nothing harmful can be seen and I still used to say, You can hardly harm your Ryzen till not cutting off from responding to really harmful situations. All the disasters what I've heard happened fixed clocks, fixed voltages. Still haven't seen posts / news about cpus died or degraded with boost settings. Ok, I'm sure if You trying real hard it can be harmful somehow but I think it's pretty not easy. As I've found my cpu even with a more aggressive EDC setting was clearly reacted to really high loads just like high temps. It's far not early and not such an extent but clearly did it and if things would be more critical, more safety reactions could be seen. Sometimes it's fully enough if You solve "just" a single serious factor and wouldn't be any trouble. Cpu degradations always need a few bad things happening the very same time. High temp, High current, high load and a higher voltage just made this even worse but without high current or high temps ( I mean really highm, dangerously high) usually nothing happens.

Cpu on cold :

 

·
Registered
Joined
·
467 Posts
It's not any common or even "normal" behaviour if it happened it was some serious bug or some bios issue. I can't imagine it was triggered by the cpu, heard pretty nonsense to me.
I did some low temp tests in the past and never seen anything, just partly similar, crazy voltage growth for decreasing temps.
Used the EDC bug for pretty long period, however a bit different point of view. Used it because the more aggressive boosting behaviour but not the fully max-out way. Positive offset wasn't acceptable for me as 24/7, not for such a small gain, like another 50 Mhz. It was boosting far higher and better than stock with negative offset even. That was the way I prefferred those days. My cpu lived all its life more or less above stock specs. When I ended up the EDC trick, I did that because bclk oc was corrected and enabled again. So I was moving to another way of tune but the main thing is my cpu is still working just as it did in the first moments. Nothing harmful can be seen and I still used to say, You can hardly harm your Ryzen till not cutting off from responding to really harmful situations. All the disasters what I've heard happened fixed clocks, fixed voltages. Still haven't seen posts / news about cpus died or degraded with boost settings. Ok, I'm sure if You trying real hard it can be harmful somehow but I think it's pretty not easy. As I've found my cpu even with a more aggressive EDC setting was clearly reacted to really high loads just like high temps. It's far not early and not such an extent but clearly did it and if things would be more critical, more safety reactions could be seen. Sometimes it's fully enough if You solve "just" a single serious factor and wouldn't be any trouble. Cpu degradations always need a few bad things happening the very same time. High temp, High current, high load and a higher voltage just made this even worse but without high current or high temps ( I mean really highm, dangerously high) usually nothing happens.

Cpu on cold :

Yeah, maybe it is because of the 5950X he was using...i rewatched the video tho and have to correct myself a bit...max voltage drawn was 1.675V Auto...but still pretty nice.
 

·
Registered
Joined
·
83 Posts
Been playing around with this again on a 3900XT with 330 / 220 / 16. Scalar at 10x is stable and in games all cores seem to boost to 4.375ghz. LLC auto, voltage auto.

But what I've noticed is SVI2 TFN voltage under a heavy load like using the stability test in CTR can sit at like 1.33~1.36v. From my experiencing doing manual overclocking most would say that is too high.

Scalar at 2~4x drops the voltage a bit, but clocks are worse.

If it's not recommended to allow load voltages that high with manual overclocks struggling to see how safe it can be allowing the EDC bug to do it.

edit - I stuck PBO/scalar on auto and did the CTR stability check and SVI2 TFN voltage is around the same lol. Presume its due to an AVX light workload.
 

·
Registered
Joined
·
77 Posts
mines been spiking up to 1.45v for years. doesn't stay there long though. I wouldn't worry about it.
 

·
Registered
Joined
·
83 Posts
mines been spiking up to 1.45v for years. doesn't stay there long though. I wouldn't worry about it.
Yeah I made a mistake above.

I was watching this video to get a bit more help working with the EDC bug -
One of my M2 drives doesn't like BCLK overclocking so I'm giving it a miss. Can boot fine at 101/102, but the drive went missing.

Anyway, next question, my Cinebench r20 runs with 160 / 113 / 16 seems to constantly get to 85~90 degrees and results in what looks to be throttling and a score of 7182 points.

I'm running a 360mm AIO cooler and idle temps for example tend to be around 36~40 degrees. 3900xt. LLC is at auto just now as is voltage.

Is there something else causing Cinebench to run this hot for me? I see a lot of other people at 70~80 degrees. HWiNFO seems to suggest 160w and 113a is a reasonable limit in terms of the chip coming close to maxing that. I know if I drop it lower temps should come down a bit but many people in this topic run upwards of 200.
 

·
Old crazy guy
Joined
·
2,461 Posts
Yeah I made a mistake above.

I was watching this video to get a bit more help working with the EDC bug -
One of my M2 drives doesn't like BCLK overclocking so I'm giving it a miss. Can boot fine at 101/102, but the drive went missing.

Anyway, next question, my Cinebench r20 runs with 160 / 113 / 16 seems to constantly get to 85~90 degrees and results in what looks to be throttling and a score of 7182 points.

I'm running a 360mm AIO cooler and idle temps for example tend to be around 36~40 degrees. 3900xt. LLC is at auto just now as is voltage.

Is there something else causing Cinebench to run this hot for me? I see a lot of other people at 70~80 degrees. HWiNFO seems to suggest 160w and 113a is a reasonable limit in terms of the chip coming close to maxing that. I know if I drop it lower temps should come down a bit but many people in this topic run upwards of 200.
Seems definitely too high, those are temps you should get only with Prime95...

PPT and TDC seems to be in the right spot.

Check how it goes with Scalar 1x or a slight negative vCore offset.
 

·
Registered
Joined
·
83 Posts
Seems definitely too high, those are temps you should get only with Prime95...

PPT and TDC seems to be in the right spot.

Check how it goes with Scalar 1x or a slight negative vCore offset.
Still hitting 88~89 degrees in HWiNFO (CPU package temp)





Scores



It's weird because gaming with the EDC bug is like 50~65 degrees depending on the game. Idle temps tend to be 38~43 degrees. Using the ryzen balanced power profile, but I have tried the 1USMUS as well (its said its OK with EDC bug).

edit - LOL just noticed AI Suite had buggered up my AIO pump fan curve and it was capped at 50%. Should have been able to hear the fans weren't totally ramping up. Still, even at 100% I'm at 84~85 degrees. Still seems a bit hot, but at least its not 89~90. Cinebench score now 7279.
 

·
Old crazy guy
Joined
·
2,461 Posts
Still hitting 88~89 degrees in HWiNFO (CPU package temp)





Scores



It's weird because gaming with the EDC bug is like 50~65 degrees depending on the game. Idle temps tend to be 38~43 degrees. Using the ryzen balanced power profile, but I have tried the 1USMUS as well (its said its OK with EDC bug).

edit - LOL just noticed AI Suite had buggered up my AIO pump fan curve and it was capped at 50%. Should have been able to hear the fans weren't totally ramping up. Still, even at 100% I'm at 84~85 degrees. Still seems a bit hot, but at least its not 89~90. Cinebench score now 7279.
What about with a high EDC settings?
Not sure I remember correctly but I think mine was around 75-78c with the 3800x.
With air cooling but I remember the 3800xt runs much hotter, it could fit.
 

·
Registered
Joined
·
83 Posts
What about with a high EDC settings?
Not sure I remember correctly but I think mine was around 75-78c with the 3800x.
With air cooling but I remember the 3800xt runs much hotter, it could fit.
I'm OK with 83~84 degrees in Cinebench, as long as its away from 89~90. I guess anyone getting 77~78 degrees with a 3900XT just has better cooling than me.

Playing intensive games seems to go 60~65 degrees which is also acceptable. As you said it seems the XT chips can run a bit hotter.

160 / 113 / 16 for settings seems good enough for me, it still allows c-states to be used :)
 

·
Old crazy guy
Joined
·
2,461 Posts
I'm OK with 83~84 degrees in Cinebench, as long as its away from 89~90. I guess anyone getting 77~78 degrees with a 3900XT just has better cooling than me.

Playing intensive games seems to go 60~65 degrees which is also acceptable. As you said it seems the XT chips can run a bit hotter.

160 / 113 / 16 for settings seems good enough for me, it still allows c-states to be used :)
Out of curiosity, which AIO it is?

I used 135/90/1 on the 3800x so the settings seems well balanced against mine considering the number of cores.
You could probably go a bit higher if you weren't constrained by the AIO.
 

·
Registered
Joined
·
83 Posts
Out of curiosity, which AIO it is?

I used 135/90/1 on the 3800x so the settings seems well balanced against mine considering the number of cores.
You could probably go a bit higher if you weren't constrained by the AIO.
Fractal Design Celsius S36, so maybe not the best, but I bought it when getting this Fractal Meshify S2.

I was lazy though and didn't change built in pump paste lol, I could probably knock a few degrees off switching over to kryonaut.

On a side note I actually found out today how to get Fmax Enhancer stable on my Asus board with this 3900xt. Just needed some positive voltage offset. That video I posted earlier made my dumb arse understand LLC is more about heavy load stability, vcore offsets tend to be more about idle/lower workload.

I quite like Fmax Enhancer, bit less "hacky" feeling from EDC and it still gets better boost clocks and multi-core clock speeds. Not quite as high as EDC, but a wee bit cooler.

But it is an Asus board exclusive from The Stilt, so obviously if you don't have an Asus mobo you aren't getting Fmax Enhancer.
 

·
Old crazy guy
Joined
·
2,461 Posts
Fractal Design Celsius S36, so maybe not the best, but I bought it when getting this Fractal Meshify S2.

I was lazy though and didn't change built in pump paste lol, I could probably knock a few degrees off switching over to kryonaut.

On a side note I actually found out today how to get Fmax Enhancer stable on my Asus board with this 3900xt. Just needed some positive voltage offset. That video I posted earlier made my dumb arse understand LLC is more about heavy load stability, vcore offsets tend to be more about idle/lower workload.

I quite like Fmax Enhancer, bit less "hacky" feeling from EDC and it still gets better boost clocks and multi-core clock speeds. Not quite as high as EDC, but a wee bit cooler.

But it is an Asus board exclusive from The Stilt, so obviously if you don't have an Asus mobo you aren't getting Fmax Enhancer.
It's not an exceptional AIO, the water pump is slow, but probably the biggest issue is that it's old and looking at CPU block it has the water flow only in the middle.

I'll have check the Fmax Enchancer, got an ASUS board for my niece.
But it's going to take at least 2 months before I can go back home...
 

·
Registered
Joined
·
83 Posts
It's not an exceptional AIO, the water pump is slow, but probably the biggest issue is that it's old and looking at CPU block it has the water flow only in the middle.

I'll have check the Fmax Enchancer, got an ASUS board for my niece.
But it's going to take at least 2 months before I can go back home...
I'll probably replace it at some point, it runs fine in general, like if I just left this chip alone to do things as built :p

Fmax is really good for more frequent single core boosts but more importantly a higher sustained multicore boost. All without any messing around needed for anyone with decent silicon. In my case it just needed a fairly small positive voltage offset to help in idle.

More explanation here if you haven't seen it already ASUS ROG X570 Crosshair VIII Overclocking &amp...
 

·
Old crazy guy
Joined
·
2,461 Posts
I'll probably replace it at some point, it runs fine in general, like if I just left this chip alone to do things as built :p

Fmax is really good for more frequent single core boosts but more importantly a higher sustained multicore boost. All without any messing around needed for anyone with decent silicon. In my case it just needed a fairly small positive voltage offset to help in idle.

More explanation here if you haven't seen it already ASUS ROG X570 Crosshair VIII Overclocking &amp...
Yes it's not strictly needed... but at least you know you have an option for improvement :)

I've read a bit the whole thread but not having the board here made it a bit pointless, I've already forgot most of it.
 

·
Registered
Joined
·
421 Posts
Fractal Design Celsius S36, so maybe not the best, but I bought it when getting this Fractal Meshify S2.

I was lazy though and didn't change built in pump paste lol, I could probably knock a few degrees off switching over to kryonaut.
Well you really should try a repaste/remount...something is pretty fishy...or this AIO is really bad.

I got similar up to 88°C and a similar CB20-Score multi with my 3900X - when I was using the stock cooler with kryonaut as TIM - of course ramped up to 100%.

With watercooling and LM as TIM...I don't see any temps higher than 78°C - with EDC=1 (500/230/1) that is...and not that much throtteling in CB20 of course with scores around 7600.

Best regards,
Medizinmann
 

·
Registered
Joined
·
83 Posts
Well you really should try a repaste/remount...something is pretty fishy...or this AIO is really bad.

I got similar up to 88°C and a similar CB20-Score multi with my 3900X - when I was using the stock cooler with kryonaut as TIM - of course ramped up to 100%.

With watercooling and LM as TIM...I don't see any temps higher than 78°C - with EDC=1 (500/230/1) that is...and not that much throtteling in CB20 of course with scores around 7600.

Best regards,
Medizinmann
Probably a combo of a poor AIO and its paste being bad. I'll redo the paste at some point and maybe replace the AIO one day. I got it new for pretty cheap due to store selling it phasing it out as old stock (2017 model).

As I said its fine with normal use and runs quiet on my fan curve for keeping gaming at 50-65 degrees depending on title.

It's an asetek 6 pump I believe but the construction of the overall AIO is clearly on the cheaper side than some.

I've settled on using Fmax Enhancer over the EDC trick anyway, due to me finally getting it stable.
 

·
Registered
Joined
·
85 Posts
Does this work on Zen 3?

I'm trying to get around the hardcap on override introduced with AGESA 1.2.0.0 on some boards.
 

·
Registered
Joined
·
1,497 Posts
Discussion Starter #1,560
Does this work on Zen 3?

I'm trying to get around the hardcap on override introduced with AGESA 1.2.0.0 on some boards.
As far as I've understood it's only for Ryzen 3000 series this works on, they fixed the issue that was causing the override with later CPU's.

Even Ryzen 4000 APU can't use the EDC bug, they adhere and keep the limit, Ryzen 5000 should be similar.
 
1541 - 1560 of 1560 Posts
Top