Overclock.net banner

1 - 20 of 21 Posts

·
Registered
Joined
·
10 Posts
Discussion Starter · #1 ·
Hey guys I have a custom loop with D95 and 2x 360mm thick rads. The issue I am having is high temps with an optimus block on the 5800x. Idle it sits in the 40s but starting P95 with SFFT it'll jump directly to 80c and climb over 2 minutes to 90c, starts at 4.5ghz then slowly goes down to 4.2ghz when hitting 90c.

Can anyone please provide comparative results/temps on a custom block.

Here is the full specs
https://www.userbenchmark.com/UserRun/41840536

And a few pics so you can see how it is setup.
2488658


2488657
 

·
Not new to Overclock.net
Joined
·
78,944 Posts
Disable AVX in Prime95.

So yes, this is normal with AVX on in Prime95.
 

·
Registered
Joined
·
10 Posts
Discussion Starter · #4 · (Edited)
Ok so I used blender and hardware monitor for some comparative benchmark temperature data for the Ryzen 7 5800x. First pic is mid first test, second pic is end of second test and third is after 3~ min cool down. I was trying to find this info to compare against for a good while before posting.

Also, the block was not fully making contact. Against what I wanted to do, I drained my loop and checked the contact area of the plate, about 1/3 of the plate was not well sealed against the CPU. After fixing it I am starting P95 SFFT's at an immediate ramp to 78c-80c and capping around 84c-85c after 10-15 min of running.

I would like to know if others can achieve something better with a different block. Apparently I bought a firehouse of a CPU because R9 5900x are still no where to be bought without paying a scalper.

2488698



2488699



2488700
 

·
The Fabricat0r
Joined
·
5,934 Posts
I'm assuming you are using PBO, which no matter what I did over-volted and caused my 5800x to throttle with a Quantum Velocity and plenty of rad. So I just manually overclocked to 4.6 all core and 1.22V which dropped load temps from 80+ to 65 max. I'd suggest messing with voltage offsets first or be lazy like me.
 

·
Registered
Joined
·
10 Posts
Discussion Starter · #6 ·
Yeah I am using stock settings completely on the system besides OC on the 3080 for mining and gaming and the 3600 ram profile set and infinity fabric at 1800. I just tuned my pump curve better and got a little less ramp and lower overall temps by a few C. In normal operation (chrome, outlook, WoW & mining all running at the same time.) My temps are pretty low with mid 40s for both the CPU and GPU. I just like to know the benchmark/slam temps so I know what is fine and what isn't. Helps me know the cold plate is making contact etc.

I'll probably OC this thing as some point and then tune voltage, I try not to OC for benchmarks scores anymore after breaking a ton of components over the past 20 years trying to get crazy OC's for daily/benchmark operations. I really like having a resale value when I am ready to goto next gen, which is pretty much every gen. The only AMD generation I haven't owned to date from athlon 64 is the ryzen 2000, only because I had two first gens and was waiting for a third gen for a better jump up.

I'm assuming you are using PBO, which no matter what I did over-volted and caused my 5800x to throttle with a Quantum Velocity and plenty of rad. So I just manually overclocked to 4.6 all core and 1.22V which dropped load temps from 80+ to 65 max. I'd suggest messing with voltage offsets first or be lazy like me.
 

·
Registered
Joined
·
151 Posts
Mine is very similar. I have a 5800x and Corsair block, although I even kept the Corsair paste which may hurt it. Anyway, after a few minutes of P95 SFFT, it stabilized at 85C. 29C water at the time.
 

·
Registered
Joined
·
151 Posts
But that seems like a really extreme test. I've run Cinebench and encoded video quite a bit a never saw temps that high. Always stays under 80C.

For a little more info, my effective clock did drop to 4200 and it was running the cores at about 1.2v. Using PBO with +200. Loadline curve is set to Auto. Gigabyte X570 Ultra.
 

·
Not new to Overclock.net
Joined
·
78,944 Posts
With my Noctua NH-D15S (the single-fan model) and the Noctua NT-H1 high-grade thermal compound, my temperature never gets much higher than about 80°C under the most demanding workloads. I've heard the fan a few times since Upgrade Day on Aprill 11th, and each time I'd check HWiNFO64 to see what the temp is.
 
  • Rep+
Reactions: robbiemcg1226

·
Registered
Joined
·
10 Posts
Discussion Starter · #10 ·
But that seems like a really extreme test. I've run Cinebench and encoded video quite a bit a never saw temps that high. Always stays under 80C.

For a little more info, my effective clock did drop to 4200 and it was running the cores at about 1.2v. Using PBO with +200. Loadline curve is set to Auto. Gigabyte X570 Ultra.
Yeah P95 has always been the worst case test. Nothing I've ever done has ever made temps as high as that. I'm pretty sure it has a function besides slamming your CPU but idk what it is lol. No one benchmarks with it anymore because nothing used for PC tasks produces the same heat but it's always the first thing I do for heat and OC tests. If your system can live through that heat/OC for an hour nothing else will break it.
 

·
Registered
Joined
·
151 Posts
Well that's interesting. I decided to re-paste my block with kryonaut. Seem to have made it slightly worse. Now I'm getting 86C. Maybe I used too much? Obviously the Corsair paste wasn't too bad though.
 

·
Registered
Joined
·
10 Posts
Discussion Starter · #12 ·
Well that's interesting. I decided to re-paste my block with kryonaut. Seem to have made it slightly worse. Now I'm getting 86C. Maybe I used too much? Obviously the Corsair paste wasn't too bad though.
To much thermal paste isn't a thing, unless you're using paste that is electrically conductive then your fear of over pasting, is it seeping over the cpu into the socket and blowing your board/chip. Almost all of the industry has moved to ceramic now though, so that's not a concern besides liquid metal or you can still find ones that actually use silver.

Make sure there's enough mounting pressure on the block and besides that it could take time to cure/setup enough to improve thermal conductivity, artic silver seemed to worked better after a day or two then fresh. I had artic mx-5 for the first tests, then I ran out and had a tiny tube of cooler master nanogel sitting around that I used as my last option. Pretty much any thermal paste that is decent will all perform within the same margin really.
 

·
Registered
Joined
·
4,317 Posts
yeah stay away from p95 idk why ppl still recommending that and ykruncher for stress testing this days.. use rogbench stress test and blender loops for you heavy load avx needs instead..
 

·
Registered
Joined
·
151 Posts
To much thermal paste isn't a thing, unless you're using paste that is electrically conductive then your fear of over pasting, is it seeping over the cpu into the socket and blowing your board/chip. Almost all of the industry has moved to ceramic now though, so that's not a concern besides liquid metal or you can still find ones that actually use silver.

Make sure there's enough mounting pressure on the block and besides that it could take time to cure/setup enough to improve thermal conductivity, artic silver seemed to worked better after a day or two then fresh. I had artic mx-5 for the first tests, then I ran out and had a tiny tube of cooler master nanogel sitting around that I used as my last option. Pretty much any thermal paste that is decent will all perform within the same margin really.
That was my understanding as well, just surprised it seems slightly worse. I partly did it as a test though just to see if the Corsair paste was working well though as I also just used the Corsair paste in my GPU block as well. Although the results might come out different, doesn't make me feel like I should re-paste it either.
 

·
Registered
Joined
·
10 Posts
Discussion Starter · #15 ·
yeah stay away from p95 idk why ppl still recommending that and ykruncher for stress testing this days.. use rogbench stress test and blender loops for you heavy load avx needs instead..
I like it because it lets me know the worst case thats ever going to happen to temps. I like to run it as one of the first things I do after building a performance PC for that reason alone. It's an old habit but it'll point out really quick if you didn't mount a block right or something is off in a system. Besides that it's not great to use for anything else. Even with extreme overclocks p95 might crash the system when nothing else you do will, so even using it for stability needs to be taken with a grain of salt.
 

·
Registered
Joined
·
4,317 Posts
you aint doing nothing that taxing on a pc in a day to day basics.. using p95 in this type of cpus you are just creating yourself headaches. specially on this small nodes.

and if you going to use p95 without avx enable "not that i condone it btw" if we get into that, might as well just use rogbench & blender and call it a day
 

·
Registered
Joined
·
143 Posts
Sounds like reseating the block helped.

Any chance you can post a screenshot of HWInfo while running p95 showing cpu power draw and temps? Also, what are your ambient and water temps?

I posted some temps here using an arctic cooler II 240 with ~25C intake temps.
 

·
Registered
Joined
·
10 Posts
Discussion Starter · #18 ·
you aint doing nothing that taxing on a pc in a day to day basics.. using p95 in this type of cpus you are just creating yourself headaches. specially on this small nodes.

and if you going to use p95 without avx enable "not that i condone it btw" if we get into that, might as well just use rogbench & blender and call it a day
I never disabled AVX, the point I am trying to make it it covers worst case temps. So a customers PC that won't clean it for 2-3 and let the heatsink get clogged with pet hair. If your CPU block can handle prime 95 for an hour where it plateaus as clean system, then a basic customer that never slams a PC with 3 years of pet hair will equate to the same thing lol. Or you not cleaning your system more then yearly and smoke or have pets or an open system etc. It has a purpose in my eyes even if it's not a realistic workload.
 

·
Laptop Enthusiast
Joined
·
9,397 Posts
Is the pump powerful enough to handle moving enough water to draw away the excess heat? Sounds to me like something is either impeding waterflow (bad/poor bend, too many right angle turns) or pump is the issue. Can't imagine just judging from those pictures that it'd be too many turns in the loop...

Gorgeous PC by the way! Work of art. :) @robbiemcg1226
 
  • Rep+
Reactions: robbiemcg1226

·
Registered
Joined
·
10 Posts
Discussion Starter · #20 ·
Sounds like reseating the block helped.

Any chance you can post a screenshot of HWInfo while running p95 showing cpu power draw and temps? Also, what are your ambient and water temps?

I posted some temps here using an arctic cooler II 240 with ~25C intake temps.
2489216


I would happily but my pump top decided to spring a leak yesterday so I am currently gimping along until a new pump/ress comes in tomorrow from amazon. My normal water temp is anywhere from 30-40c depending on how hot it gets here. Bykski dosn't sell just the pump top to replace the cracking acrylic and I can't really wait 2-3 weeks for shipping this time, so I am replacing it with a EK-Quantum Kinetic, at least i'll have a spare D5 pump at the end of this, I guess :/, I'll try to get you the comparative data when I get it all sorted.
 
1 - 20 of 21 Posts
Top