Cooling advice for a first-timer

[fondueless]

I've recently built my first PC in ~20 years, and I'm worried about the cooling on my setup. I put a Ryzen 5 3600 with the stock Wraith cooler and pre-applied thermal paste onto an ASUS Prime X570-P motherboard and put it in a Phanteks P400S case with just the included fans (two 120 fans, one in front and one in the back) run with the case fan controller set on high. (Other hardware: 2060 Super,
Corsair Vengeance LPX 16GB RAM, Thermaltake Toughpower GF1 750W 80+ Gold power supply.) This works fine for gaming so far, but I also plan to use it for CPU-heavy data analysis, though I'm not really planning on overclocking.

 

When I run a Prime95 stress test (v29.8, blend, 12-thread), I get the following temperature readings from Ryzen Master: after 1 min - 73.2 C, 2 min - 75.6 C, 3 min - 76.5 C, 4 min - 78.7 C, 5 min - 85.7 C. I ended the test at 6 minutes when it hit 92 C. My idle CPU temp is ~38-42.

 

Am I right to think that this is a bad sign for the longevity of my machine if I use it for CPU-heavy workloads? If I did make a change, should I change the CPU cooler or the case airflow, and how? Thanks for your help!!!

[jaslion]

15 minutes ago, fondueless said:

I've recently built my first PC in ~20 years, and I'm worried about the cooling on my setup. I put a Ryzen 5 3600 with the stock Wraith cooler and pre-applied thermal paste onto an ASUS Prime X570-P motherboard and put it in a Phanteks P400S case with just the included fans (two 120 fans, one in front and one in the back) run with the case fan controller set on high. (Other hardware: 2060 Super,
Corsair Vengeance LPX 16GB RAM, Thermaltake Toughpower GF1 750W 80+ Gold power supply.) This works fine for gaming so far, but I also plan to use it for CPU-heavy data analysis, though I'm not really planning on overclocking.

 

When I run a Prime95 stress test (v29.8, blend, 12-thread), I get the following temperature readings from Ryzen Master: after 1 min - 73.2 C, 2 min - 75.6 C, 3 min - 76.5 C, 4 min - 78.7 C, 5 min - 85.7 C. I ended the test at 6 minutes when it hit 92 C. My idle CPU temp is ~38-42.

 

Am I right to think that this is a bad sign for the longevity of my machine if I use it for CPU-heavy workloads? If I did make a change, should I change the CPU cooler or the case airflow, and how? Thanks for your help!!!

Terrible case for airflow. There's barely any air that can get in so the inside gets hot and you get problems. Adding more fans isn't going to help as the intake for air is just too small. So pretty much get a different case or remove the side and or front panel and run it like that.

[Froody129]

That's not too bad. Prime95 is a torture test specifically because it tortures your CPU. It's not a very realistic test of 'real' workloads, though your specific case may be different of course.

 

Do you have very high ambient temps?

Is there even pressure across the cooler? (are all the connecting points at about the same pressure)

 

I would try taking off the front panel of the case if you haven't done so already. Then at least you know if the case is choking your airflow. Perhaps another front fan wouldn't hurt either. At very least more fans means lower rpm per fan, which then means a quieter system.

 

I wouldn't stress too much if your temps are under 80-90C in your workload. Running high overclocking voltage is worse for your CPU than high temperatures, and it's the constant cycling of temperatures which really wears computer parts. Constant running at a stable temperature is about ideal for a CPU.

 

[aDoomGuy]

Prime 95 really stresses the CPU in ways that do not occur under normal use. Your temps seem to be normal. Try testing during normal use case operations.

 

1 minute ago, jaslion said:

Terrible case for airflow. There's barely any air that can get in so the inside gets hot and you get problems. Adding more fans isn't going to help as the intake for air is just too small. So pretty much get a different case or remove the side and or front panel and run it like that.

No it's not terrible. Just not optimal. Adding an intake fan to the top helps.

[jaslion]

2 minutes ago, aDoomGuy said:

Prime 95 really stresses the CPU in ways that do not occur under normal use. Your temps seem to be normal. Try testing during normal use case operations.

 

No it's not terrible. Just not optimal. Adding an intake fan to the top helps.

You're looking at 10-15c worse temps depending on what's inside the system compared to other mid range cases that have decent or better airflow. I'd call that pretty terrible if you ask me.

[aDoomGuy]

Just now, jaslion said:

You're looking at 10-15c worse temps depending on what's inside the system compared to other mid range cases that have decent or better airflow. I'd call that pretty terrible if you ask me.

Well I am not. The stock fans aren't very good, I replaced them for Noctua and ML 120/140 and I'm seeing OK temps. In a stock configuration it's pretty bad, yes but you said it wouldn't help to add fans... when in fact it does. If you compare it to the P400A which is a great budget case for airflow then of course it can't compete but at the same time there are more expensive cases that have worse airflow than the P400. If running a P400 non-A case then it's best to put a fan in the top to suck air in, it will help stabilize the temperature.

[freeagent]

Real workloads are usually pretty intense. Home user workloads are usually tame. If you can pass a stress test like p95, linpack xtreme, real bench, then you can run whatever you want. Obviously if you cant, then you have a system with limited capabilities. If you have ever heard of Folding @ Home then you would know how much of a load that places on your system. Its why people use dedicated machines for certain tasks. But if you cant run a simple stress test, add some more fans. Its only a few bucks. And if that doesn't help much, then sure, look at a new case. You could always sell yours locally. I'm not trying to be rude, so please don't take it the wrong way. I am an avid air cooler, I've chased c's for years usually employing brute force tactics. Its the only thing that works. I am speaking mainly as an overclocker, I don't do stock speed very well.

[fondueless]

Thanks to you all for your advice. I tried two tests that are more representative of what my typical CPU-heavy workloads will look like.

 

Blender render: 87-90 C

Custom python analysis: maxes at 80-85 C in the most intense part of the workload. 

 

Regarding the case, I haven't noticed a big difference with the front panel off v. on. In my workloads, it might make a degree or two of difference, but it would require somewhat careful measurement to nail down reliably.

 

I'm considering adding a better CPU cooler. If I were to swap out the Wraith stock cooler for something like a be quiet! Dark Rock Slim, CPU Cooler, 180W TDP (https://www.newegg.com/be-quiet-dark-rock-slim/p/13C-001F-00030), would you expect that to improve my situation materially? Any other recommendations you think might fit my situation better? Thanks again.

[fondueless]

7 minutes ago, fondueless said:

Thanks to you all for your advice. I tried two tests that are more representative of what my typical CPU-heavy workloads will look like.

 

Blender render: 87-90 C

Forgot to mention that this Blender rendering test is using the CPU exclusively. I obviously don't anticipate rendering without my GPU in real life, but some of the other custom analysis scripts I will be running will look more like Blender CPU rendering from a computational workload perspective.

[WoodenMarker]

6 hours ago, fondueless said:

I'm considering adding a better CPU cooler. If I were to swap out the Wraith stock cooler for something like a be quiet! Dark Rock Slim, CPU Cooler, 180W TDP (https://www.newegg.com/be-quiet-dark-rock-slim/p/13C-001F-00030), would you expect that to improve my situation materially? Any other recommendations you think might fit my situation better? Thanks again.

Sure, that would be better. What ram are you using?

What kind of features are you looking for in a cooler?

[LjjlijJ]

15 hours ago, fondueless said:

I've recently built my first PC in ~20 years, and I'm worried about the cooling on my setup. I put a Ryzen 5 3600 with the stock Wraith cooler and pre-applied thermal paste onto an ASUS Prime X570-P motherboard and put it in a Phanteks P400S case with just the included fans (two 120 fans, one in front and one in the back) run with the case fan controller set on high. (Other hardware: 2060 Super,
Corsair Vengeance LPX 16GB RAM, Thermaltake Toughpower GF1 750W 80+ Gold power supply.) This works fine for gaming so far, but I also plan to use it for CPU-heavy data analysis, though I'm not really planning on overclocking.

 

When I run a Prime95 stress test (v29.8, blend, 12-thread), I get the following temperature readings from Ryzen Master: after 1 min - 73.2 C, 2 min - 75.6 C, 3 min - 76.5 C, 4 min - 78.7 C, 5 min - 85.7 C. I ended the test at 6 minutes when it hit 92 C. My idle CPU temp is ~38-42.

 

Am I right to think that this is a bad sign for the longevity of my machine if I use it for CPU-heavy workloads? If I did make a change, should I change the CPU cooler or the case airflow, and how? Thanks for your help!!!

If you plan to run cpu heavy analysis job, getting a better cpu cooler is the higher priority here. check out the cooler tier list in this forum (it's not 100% accurate but a good reference). Just get a hyper 212 level cooler (or the tier above) and with proper mounting it should immediately make a day and night  difference. 

 

As others have already pointed out, prime95 is a torture test for testing oc stability / cooler performance in (unrealistically) extreme conditions and doesn't reflect real world senarios.

 

Since you don't oc, instead just run cinebench for maybe twice as long as how much time your average analysis jobs take. Cinebench will keep all cores running at the highest allowed clock (not to be confused with single core boost clock). If the stabilised temps look ok to you then you should be good.

 

See how you go after upgrading the cpu cooler. if more is to be desired, add another intake fan and start from there. PS I'm not familiar with the case but it looks like high pressure fans may be best for mounting at the front.)

Edited May 19, 2020 by LjjlijJ
typo

[LjjlijJ]

13 hours ago, fondueless said:

something like a be quiet! Dark Rock Slim, CPU Cooler, 180W TDP

overkill. ryzen 3600 is rated 65W.

But if you have the budget and can fit it in sure go ahead.

 

I use a deepcool gammaxx gt (i think only one tier better than hyper 212 in the tier list? got it on special for ~20 bucks) on my ryzen 3600 and here are my temps. feel free to use as reference.

 

[LjjlijJ]

Out of curiosity I tested with prime95. Also for your reference

 

Blend test ~20mins had my cpu peak & stabilise @ ~86.1C

 

Immediayely after, small ffts ~10mins had it peak & stabilise @ ~88.7C

 

[fondueless]

For anyone stumbling on this thread later, I wanted to update with what I ended up doing. Even though I knew it was overkill, I swapped out the stock cooler for a be quiet Dark Rock Slim cooler, and it works beautifully. My CPU temps reach a steady state of 65 degrees in Prime95 and 70 in Cinebench R20, so clearly I have more cooling in there than I need. Still, it wasn’t much more expensive than my other options, and now I know know I can push the hardware to the limit without worrying about temps. I made no changes to the case, so case airflow was not limiting in this particular instance. Thanks to you all for your help!

[Jeppes]

24 minutes ago, fondueless said:

For anyone stumbling on this thread later, I wanted to update with what I ended up doing. Even though I knew it was overkill, I swapped out the stock cooler for a be quiet Dark Rock Slim cooler, and it works beautifully. My CPU temps reach a steady state of 65 degrees in Prime95 and 70 in Cinebench R20, so clearly I have more cooling in there than I need. Still, it wasn’t much more expensive than my other options, and now I know know I can push the hardware to the limit without worrying about temps. I made no changes to the case, so case airflow was not limiting in this particular instance. Thanks to you all for your help!

Case-airflow will show up more in cpu-temps if you have a +200W gpu dumping its heat into the case.