• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

***The Official 5900X \5950X owners thread***

Soldato
Joined
24 Aug 2013
Posts
4,551
Location
Lincolnshire
Been a 3 chip design you need to spread the paste over the full top of the CPU IHS. I use the Der8uaer offset kit on mine to center the block over the 2 CCD's.
https://www.caseking.de/en/der8auer...et-Custom-Mount-FSD8-034.html?tplview=desktop
Sold out in the UK at ocuk now, so would have to ship from there sister company caseking if you where interested.

Thanks for the idea but i pretty much gave up and stuck my old 10900k back in.

Repasted a few times and it didnt make much difference paste was spread to each corner of the IHS when removing the cooler and cooler was mounted dead flat. Tried a few methods including spreading manually over the entire IHS. Only thing is the IHS is a bit concave so perhaps this is the issue. like 53c idle in bios each time.
Perhaps dodgy CPU but i couldnt be bothered RMA then to try again. Perhaps just wait for 12900k or whatever it is.

Main issue is my system is in a cupboard for silence where it recirculates most of the air so the ambient can get pretty hot. Like 30c average after a few hours gaming sometimes more especially in the summer. In games i dont think it would cope very well, perhaps throttle or overheat as the 86c i got was pretty much 20c ambient. Whereas my 10900k even OC hits maybe 70c maximum.
 
Soldato
Joined
27 Jan 2003
Posts
5,780
Location
Chesterfield, UK
High end pc’s stuck in enclosures will result in high temperatures. Best thing you can do is to modify the cupboard and put some very slow big fans to extract the heat. You won’t hear the fans but it will make a massive difference to the temps.
 
Associate
Joined
28 Jul 2021
Posts
162
Location
London
Trying to optimize 5900X on a Gigabyte X570 board. Which would be better - negative offset to the core -0.130V and no negative curve or negative offset to the core -0.1V and negative curve -10 on all cores? The second variant gives me more points in Cinebench 20. Also how do you test for stability? I've tried 1 hour CB20 loops and it seems OK.
 
Soldato
Joined
28 May 2007
Posts
16,792
Trying to optimize 5900X on a Gigabyte X570 board. Which would be better - negative offset to the core -0.130V and no negative curve or negative offset to the core -0.1V and negative curve -10 on all cores? The second variant gives me more points in Cinebench 20. Also how do you test for stability? I've tried 1 hour CB20 loops and it seems OK.

Id probably lean towards holding the highest boost.

An hour CB will probably show 99% of issues. I’d call that desktop stable myself at least for the CPU.
 
Associate
Joined
28 Jul 2021
Posts
162
Location
London
Id probably lean towards holding the highest boost.

An hour CB will probably show 99% of issues. I’d call that desktop stable myself at least for the CPU.

Can you please explain a bit further what do you mean by holding the highest boost - which one of the options that I was considering would that be /I've recently moved to Ryzen so may miss some points/. Is that referring to clock boost or something else? And how can I measure these since I've calculated those 2 options to be more or less the same by counting a sinle item in curve in its highest possible value of 5 mV /as per my understanding/?
I have left the CPU power limit to its nominal /of 142W if I remeber it correctly/.
 
Last edited:
Associate
Joined
30 Jun 2006
Posts
1,049
Location
Cardiff
Trying to optimize 5900X on a Gigabyte X570 board. Which would be better - negative offset to the core -0.130V and no negative curve or negative offset to the core -0.1V and negative curve -10 on all cores? The second variant gives me more points in Cinebench 20. Also how do you test for stability? I've tried 1 hour CB20 loops and it seems OK.

I did a 0.1 voltage offset and a negative -10 per core tested overnight and then raised it in increments of 5. I used core cycler to do this. Until all were done but a bit of pain tbh. then when I was finally was happy ran CB all night and then left BFV running all night as this game makes my cpu run hot.
 
Associate
Joined
18 Oct 2011
Posts
2,135
Posted this in another thread as i had forgot about it decided to moved it over to the correct place. @LtMatt

Running my 5950x at -5 all core with PBO2 is absolutey fine, no issues but when i try -5 on my best 2 core and -5 on my second best and then the rest -10 i would boot back into windows and have my main monitor resolution around 1080-768p and there would be no way to change it (no options). Soon as i would go back to bios and revert back to all core -5 runs fine again.
 
Associate
Joined
28 Jul 2021
Posts
162
Location
London
I did a 0.1 voltage offset and a negative -10 per core tested overnight and then raised it in increments of 5. I used core cycler to do this. Until all were done but a bit of pain tbh. then when I was finally was happy ran CB all night and then left BFV running all night as this game makes my cpu run hot.
I went with a 0.1 voltage offset and a negative -10 per core for the best ranked 6 cores and -15 for the other 6. That gave me best CB20 score. Run CB20 for 4 hours and then prime. So far so good.
Now I am playing with Geardown Mode (GDM). Currently I've turned it off and it seems to me that the system is less responsive under load /CB20/ so I assume that means that memory is more engaged, i.e. should be good but memory tests show slightly lower results.
 
Associate
Joined
20 Sep 2008
Posts
16
r7miaFv

is it ok if cpu power hits 200w and its red?
also let me know if something looks bit off :p
and maybe i can do something better
i use curve optimizer first 4 cores negative 12 and everything else negative 20

https://imgur.com/a/wgVMt20
 
Associate
Joined
30 Jun 2006
Posts
1,049
Location
Cardiff
pcuser>I had one that could only run at -5 and that was it and the rest after I tested it over multiple nights were like -20 -25 and I think a few were -30. So now instead of running close to 85c in BFV it runs at like 71/72C which I am happy with. But then again I had the same kinda experience with the 5900X so it maybe down to the motherboard.
 
Associate
Joined
28 Jul 2021
Posts
162
Location
London
pcuser>I had one that could only run at -5 and that was it and the rest after I tested it over multiple nights were like -20 -25 and I think a few were -30. So now instead of running close to 85c in BFV it runs at like 71/72C which I am happy with. But then again I had the same kinda experience with the 5900X so it maybe down to the motherboard.

I guess that must have been your best core. People claim that you can not lower the best and second best cores too much as they aare alegedly the most used. Lowering the voltage helped with temperatires but I read that as a result single core performance suffers so I guess it's a compromise. I don't know whether playing further with curves will give much difference. I will probably play with tweaking the memory and that's going to be it. Overclocking seems to be almost gone I'm afraid.
 
Last edited:
Associate
Joined
30 Jun 2006
Posts
1,049
Location
Cardiff
That's just the thing though isn't it. Whether you can accept the stock temp values. I could never accept it and understood that I would lose performance. That might explain why my 5900/5950x seem to much slower than everyone else when I look at their R23 or 3D mark cpu scores mine is nowhere close to them. Like someone said in R23 the avg is 30k Whereas mine is only 26k.
 
Associate
Joined
18 Oct 2011
Posts
2,135
That's just the thing though isn't it. Whether you can accept the stock temp values. I could never accept it and understood that I would lose performance. That might explain why my 5900/5950x seem to much slower than everyone else when I look at their R23 or 3D mark cpu scores mine is nowhere close to them. Like someone said in R23 the avg is 30k Whereas mine is only 26k.

I would say the average with PBO enabled is 28-29k, you gotta have a good chip to just be 30k with only PBO enabled & no CO.

Still looking if anyone knows what this is all about -->

**Running my 5950x at -5 all core with PBO2 is absolutey fine, no issues but when i try -5 on my best 2 core and -5 on my second best and then the rest -10 i would boot back into windows and have my main monitor resolution around 1080-768p and there would be no way to change it (no options). Soon as i would go back to bios and revert back to all core -5 runs fine again.**
 
Associate
Joined
28 Jul 2021
Posts
162
Location
London
That's just the thing though isn't it. Whether you can accept the stock temp values. I could never accept it and understood that I would lose performance. That might explain why my 5900/5950x seem to much slower than everyone else when I look at their R23 or 3D mark cpu scores mine is nowhere close to them. Like someone said in R23 the avg is 30k Whereas mine is only 26k.

Stock temps are fine at idle and not so much under load - I wish they would be lower under load and there is not much I can do about it apart from getting a top of the line water cooling and I don't think that is going to change things much. I do understand that I'm getting lower results because of this but on the other side I can barely tolerate the 78-79 degrees under load that I'm getting under load so it seems that I don't have much choice regarding this. So I guess that I'm more or less stuck at stock. I can play with curve per core but do not know whether that is going to give me much. Other thing is to tighten timings which are loose at the moment but I guess I need to finish testing the processor first before moving to memory optimization.

@Loxa What happens if you try -9 or -8 or other values? Did you try splitting the cores in 2 /best half and the other half/? Also did you try changing video drivers /sweeping clean the former ones/?
 
Last edited:
Associate
Joined
30 Jun 2006
Posts
1,049
Location
Cardiff
do you use a program like core cycler? That will do most of the dirty work for you.

Loxa>that sounds so bizare I wouldn;t even know where to begin to try and troubleshoot that particular issue
 
Top