• 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.

Normal temps for threadripper

the wopster

the wopster

the wopster

Soldato
Joined
14 Oct 2007
Posts
9,379
Location
newcastle
what temps should I be seeing under realbench stress stock volts(1.35) no overclock multiplier locked to 35? My chip is hitting 68c, with my custom watercooling 1xtriple 1xdouble which can’t be right and my chip seems to be throttling at stock
 
woppy101 said:
what temps should I be seeing under realbench stress stock volts(1.35) no overclock multiplier locked to 35? My chip is hitting 68c, with my custom watercooling 1xtriple 1xdouble which can’t be right and my chip seems to be throttling at stock
Click to expand...

I have no experience with TR but 68c seems a very good temp. I remember at launch that these were a warm chip, they are packing a lot of cores after all!
 
Running stock, idle is around 29 degrees C and currently running just over 50% load across all cores and it's showing 52degrees C according to CoreTemp. Clock speed at that load is 3697 on all cores.
THis is on air cooling. Not tried realbench so dunno if stats any help.
I tend not to run CPU stress tests as not representative of my use.
 
Done a test run on my 1920X. Mine is cooled using an EK supremacy block, 45mm thick 420mm and 240mm radiators both with fans just pulling air through. Load temps are what I get running a 2 hour video clip through handbrake. Prime load is using Small FFT with FMA3 instructions. CPU is at 3.975GHz (39.75x 100) and memory is 4x 8GB 3200 C14, running at 1.41V

Ambient 25
Idle 34
load 62
Prime load 73
VRM idle 45
VRM load 58
VRM Prime load 71
Load VCore 1.38V
Idle VCore 1.32V
 
I'd echo DarrenM343's kinda thoughts. Stress testing is generally throwing an unreasonable workload at it.
Sit it through a few runs of cinebench, see what temps you get there. That'll give you a more reasonable figure for "heavy, normal use".

also, have you tried the "large single blob" of TIM as in your "should I buy" thread?
It's a pretty good sized chip. If there's an air bubble trapped under the waterblock, it's going to loose a lot of contact. I always go with a single, rounded blob so pressure from above will make it spread out evenly, no chance of air getting trapped (think, if you drew a donut with the paste, add the waterblock... there'll be an air bubble, that but less extreme).
 
Last edited:
Kei said:
Done a test run on my 1920X. Mine is cooled using an EK supremacy block, 45mm thick 420mm and 240mm radiators both with fans just pulling air through. Load temps are what I get running a 2 hour video clip through handbrake. Prime load is using Small FFT with FMA3 instructions. CPU is at 3.975GHz (39.75x 100) and memory is 4x 8GB 3200 C14, running at 1.41V

Ambient 25
Idle 34
load 62
Prime load 73
VRM idle 45
VRM load 58
VRM Prime load 71
Load VCore 1.38V
Idle VCore 1.32V
Click to expand...
Is that with just the cpu in the loop?
 
Seems that thermal paste application can play quite a big role in the performance. I watched the gamers nexus video about it not long before I bought the parts and they seemed to find that the very large single blob on the Z of the RYZEN logo on the IHS performed best. (about the size of a bean) I tried that first time using the thermal grizzly paste supplied with the EK block and the temperatures I got were terrible. (around 87 degrees peak) I took it off and reapplied the paste using the same method but using EK ectotherm paste instead and got a drop in temperature to around 77 degrees peak. I then moved my pc out from under my desk to give it more breathing room above which has lowered the temps again to what I've shown above. The drop in temp on the VRM was the biggest simply by moving my pc as it used to peak at 97 degrees, now it only gets to 71.
 
Just checked my system as your temps all seem high, mine is pretty much at stock settingd right now other than the defaults for the memory clock (2800).

Have put temps below for your refrence, these are all peak values from a 10 min period taken with Ryzen master for 1950X with a Noctua HSF.

Temp at front of mobo is 27
Room temp is 24

Idle CPU Temp - 28~32 (2.2ghz all cores)
Load (single thread) CPU Temp - 51 (4.2ghz single core)
Stress (all threads) CPU Temp - 54 (3.6ghz all cores)

Will re-apply my OC later and check again - but pretty confident it doesn't hit the highs you've found with your 1920X and im only on air.
 
Kei said:
Seems that thermal paste application can play quite a big role in the performance. I watched the gamers nexus video about it not long before I bought the parts and they seemed to find that the very large single blob on the Z of the RYZEN logo on the IHS performed best. (about the size of a bean) I tried that first time using the thermal grizzly paste supplied with the EK block and the temperatures I got were terrible. (around 87 degrees peak) I took it off and reapplied the paste using the same method but using EK ectotherm paste instead and got a drop in temperature to around 77 degrees peak. I then moved my pc out from under my desk to give it more breathing room above which has lowered the temps again to what I've shown above. The drop in temp on the VRM was the biggest simply by moving my pc as it used to peak at 97 degrees, now it only gets to 71.
Click to expand...
My VRM temps have never got above 70c
 
Following up on my ealier post, with an OC to 3.9 temps go to

Idle CPU Temp 32~33 (2.2-3.9ghz all cores) [+4 / +1 C on stock]
Load (single thread) CPU Temp - 62 (3.9ghz all cores) [+11 C on stock]
Stress (allthread) CPU Temp - 65 (3.9ghz all cores) [+11 C on stock]

Thats with 1 hour runs for load and stress
 
BongoHunter said:
Following up on my ealier post, with an OC to 3.9 temps go to

Idle CPU Temp 32~33 (2.2-3.9ghz all cores) [+4 / +1 C on stock]
Load (single thread) CPU Temp - 62 (3.9ghz all cores) [+11 C on stock]
Stress (allthread) CPU Temp - 65 (3.9ghz all cores) [+11 C on stock]

Thats with 1 hour runs for load and stress
Click to expand...
What stress test is that? And what volts? What worries me is there is no way your air cooled block should be out performing my water loop

I must have a dog of a chip, my chip will not do 4ghz on anything under 1.45v but that hits 80c rather quicky, 3.9 1.38v is ok and temps hover around 75c
 
Last edited:
woppy101 said:
What stress test is that? And what volts? What worries me is there is no way your air cooled block should be out performing my water loop

I must have a dog of a chip, my chip will not do 4ghz on anything under 1.45v but that hits 80c rather quicky, 3.9 1.38v is ok and temps hover around 75c
Click to expand...
The HardOCP test on the EK block shows how poorly it performs compared with the XSPC. I'm still running mine in the supplied config which they suggest performs worse. You are right that without knowing the voltage and the stress test being used by @BongoHunter , the results can vary quite significantly. Normal full load on my chip only gets to low 60's, even aida stress test only gets it up to mid 60's. Prime pushes it straight past 70. Temps get far worse when I fully load the cpu and gpu together as the waterloop has to dump around 550-600W of heat.

TBH, your chip seems to be very similar to mine, maybe a hair worse as mine will do 4GHz, but I don't feel comfortable with the voltage going over 1.4V for prolonged periods.
 
Last edited:
woppy101 said:
What stress test is that? And what volts? What worries me is there is no way your air cooled block should be out performing my water loop

I must have a dog of a chip, my chip will not do 4ghz on anything under 1.45v but that hits 80c rather quicky, 3.9 1.38v is ok and temps hover around 75c
Click to expand...

I use Aida for stress test, can't recall exact voltage but it's under 1.4, will check and confirm when home
 
You will have people arguing all day long about thermal paste but I use the spread method when pasting directly onto a die and the X method when pasting a heat spreader, never failed me before
 
You must log in or register to reply here.
Back
Top Bottom