Stress test cpu and gpu

[MingLee420]

Wanting to run cinebench r23 and heaven benchmark at the same time to see how hot my pc gets in 40c weather and adjust the fan curve from it, would it be safe to run both at the same time?

[RONOTHAN##]

3 minutes ago, MingLee420 said:

Wanting to run cinebench r23 and heaven benchmark at the same time to see how hot my pc gets in 40c weather and adjust the fan curve from it, would it be safe to run both at the same time?

As long as your PSU isn't complete garbage, yes it's plenty safe.

[Somerandomtechyboi]

3 minutes ago, MingLee420 said:

40c weather

Are you in death valley or something?

 

Well depends on the temps, as long as nothing exceeds 100c it should be fine, prefferably under 90c tho

[LIGISTX]

3 minutes ago, MingLee420 said:

Wanting to run cinebench r23 and heaven benchmark at the same time to see how hot my pc gets in 40c weather and adjust the fan curve from it, would it be safe to run both at the same time?

Yes, but cinebench isn’t really a super CPU intensive test. Folks use it for stability testing but it’s really not enough of a load for that. 
 

But sort of regardless, yes, to answer your question that’s fine. Your PSU should handle that no problem. 

[HQuan]

11 minutes ago, LIGISTX said:

Yes, but cinebench isn’t really a super CPU intensive test. Folks use it for stability testing but it’s really not enough of a load for that. 

I'm a folk?

[Helpful Tech Witch]

Id not use cinebench or heaven. To push the cpu, p95 all the way. Thats used to torture cpus to test overclocks.

for gpu, probably furmark.

[MingLee420]

20 minutes ago, Somerandomtechyboi said:

Are you in death valley or something?

 

Well depends on the temps, as long as nothing exceeds 100c it should be fine, prefferably under 90c tho

haha just normal weather in australia, i remember my 2070super would go straight to 75 and above in the summer and was so loud as it was a zotac mini

[MingLee420]

Ok i just did both tests and my room was apparently 28c but i know that it was def higher but temps seems fine for both cpu got max 84 and gpu temp got a max of 71.7c and a hot spot of 84.6c and memory junction was 82 

 

btw I used to have my cpu oced to the walls at 5ghz with 1.285 votls and went to like 90  but decided to go stock clocks and undervolt it a few weeks ago and just by changing the cache ratio from 43 to 45 i got nearly all the performance back at 4.7ghz and with 1.18 volts and with more testing the closer the cache was from cpu speed it performs much better 

[Somerandomtechyboi]

11 minutes ago, MingLee420 said:

 

btw I used to have my cpu oced to the walls at 5ghz with 1.285 votls and went to like 90  but decided to go stock clocks and undervolt it a few weeks ago and just by changing the cache ratio from 43 to 45 i got nearly all the performance back at 4.7ghz and with 1.18 volts and with more testing the closer the cache was from cpu speed it performs much better

I think running ~1.2v-1.25v is prefferable cause after that it seems to have diminishing returns

 

Yea  undervolting is far more worth it than oc, i think ram oc is prefferable since it still gives you extra performance without shooting power bills through the roof

[LIGISTX]

2 hours ago, HQuan said:

I'm a folk?

I guess? If you use cinebench I guess so. It’s just not enough of a load to test for true stability. 
 

2 hours ago, Helpful Tech Wiard said:

Id not use cinebench or heaven. To push the cpu, p95 all the way. Thats used to torture cpus to test overclocks.

for gpu, probably furmark.

These days those are both considered “power viruses”. P95 draws too much power due to AVX instructions, I’m not sure they ever changed that. P95 is literally toooo stressful. It forces the CPU to draw unrealistic power and cooks it. If your stable and temps are not instantly 100c, P95 would tell you if your stable. But asus real bench is a better test as it has good transient load testing as well, that’s where things typically get unstable. Same sort of thing with furmark, it can cook a GPU VRM.

[thrasher_565]

hmm if you lower the temp by lowering the voltage that can get you a boost clock?

[MingLee420]

3 hours ago, thrasher_565 said:

hmm if you lower the temp by lowering the voltage that can get you a boost clock?

intel cpu's only boost highest on 2 cores like 5.1ghz but windows and maybe some other os's spreads workloads so single core overclocking is kinda useless only all core is useful and to get from 4.7 to 4.8 or higher is not worth it in terms of voltage/watts for my 10700k

[Somerandomtechyboi]

5 hours ago, LIGISTX said:

These days those are both considered “power viruses”. P95 draws too much power due to AVX instructions, I’m not sure they ever changed that. P95 is literally toooo stressful. It forces the CPU to draw unrealistic power and cooks it. If your stable and temps are not instantly 100c, P95 would tell you if your stable. But asus real bench is a better test as it has good transient load testing as well, that’s where things typically get unstable. Same sort of thing with furmark, it can cook a GPU VRM.

Uhhh any other alternatives to p95 and furmark?

 

Cause p95 is really good at finding ram errors for me, alot better than memtest64, and ive never really had issues with furmark burning gpus, my gts 250 iirc only gets upto mid 70s and my moms laptop gt720 only gets to mid 90s which i consider pretty normal since it is a laptop after all and not really used for any gpu intensive gaming or anything like that

 

Ive looked at occt but idk how it detects errors and 1 hour limit is stupid, i was interested in using it for stressing 1 core at a time but doesnt seem to do that

 

 

I am on 775 so no sht i dont have issues since its ancient hardware but i still wanna know if theres any better testing software thats as good or better than p95 or furmark

[LIGISTX]

1 minute ago, Somerandomtechyboi said:

Uhhh any other alternatives to p95 and furmark?

 

Cause p95 is really good at finding ram errors for me, alot better than memtest64, and ive never really had issues with furmark burning gpus, my gts 250 iirc only gets upto mid 70s and my moms laptop gt720 only gets to mid 90s which i consider pretty normal since it is a laptop after all and not really used for any gpu intensive haming or anything like that

 

Ive looked at occt but idk how it detects errors and 1 hour limit is stupid, i was interested in using it for stressing 1 core at a time but doesnt seem to do that

 

 

I am on 775 so no sht i dont have issues since its ancient hardware but i still wanna know if theres any better testing software thats as good or better than p95 or furmark

For 775 p95 is fine. For RAM testing I followed the DDR4 overclocking guide found here and ran Google’s RAM test software under Ubuntu via linux subsystem for windows: https://github.com/integralfx/MemTestHelper/blob/oc-guide/DDR4 OC Guide.md

 

That would apply to any system, I am fairly certain it would work on 775 without issue, and it’s one hell of a RAM test. 
 

For CPU I run asus real bench: https://rog.asus.com/rog-pro/realbench-v2-leaderboard/

[Ralfi]

I’m also in Australia & I’m curious to see how my Winter-built PC will handle the warmer weather.

 

During the cooler months, I ran FurMark’s GPU & CPU test at the same time for up to an hour (mainly 20-30 minutes) & that was enough for me…& my PC (iirc, GPU hot spot was at high 80°s-low 90°s & CPU high 60°s-mid 70°s).

 

Idle in winter, my 6700XT & 5600X were at low 30°s. In the warmer days recently, they were 10° warmer, but I’m yet to stress test them

[MingLee420]

18 hours ago, Ralfi said:

I’m also in Australia & I’m curious to see how my Winter-built PC will handle the warmer weather.

 

During the cooler months, I ran FurMark’s GPU & CPU test at the same time for up to an hour (mainly 20-30 minutes) & that was enough for me…& my PC (iirc, GPU hot spot was at high 80°s-low 90°s & CPU high 60°s-mid 70°s).

 

Idle in winter, my 6700XT & 5600X were at low 30°s. In the warmer days recently, they were 10° warmer, but I’m yet to stress test them

Yh same my gpu clocks down as it is alot warmer than before