• PC Building Thread V6 - "running six RGB controller utilities at once" edition
    999 replies, posted
It's a shame that overclocking is so neutered on Pascal cards, because you could theoretically push a fuckin nasty overclock with cooling that good.
everything look good? I've been out of the loop for a while. like a decade type of while. ddr4 and m.2 are alien concepts to me. are ryzen heat sinks as ridiculously good as phenom II heat sinks were or should I pick up a hyper 212 or whatever https://imgkk.com/i/2_kt.png
Except everything apart from the core is very hot, when I had my kraken bracket my core was 40c while my VRM's even with heat sinks and a fan, were 80 - 90c regardless of what I did.
If you're using stock paste, new stuff like arctic silver 5 or cryonaut with always help a bit. Also (and I know this was kinda hit on already,) HWinfo64 is the only monitoring tool I know of that even gets into the ballpark of real cpu package thermals. Everything else is inaccurate. You could not hit 117C without blowing past tjmax and your cpu either throttling to a standstill or shutting down completely. If you are having concerns about your current thermals, remove dust filters and/or front panel and retest. See what combos work best thermally. More fans are good, but make sure they are set up in such a way to where you are not creating a vortex in your case (i.e front fans pull air in, back fans shoot air out.)
With how I have the radiator and pull fan set up on the EVGA 120 CLC, it's blowing air straight down onto the board, so when I checked out the temp readings in HWInfo, it was showing the VRM Mosfets at around 68C-72C, so my subtle hope of cooling the VRM and etc on the board came to be true.
Wait, so I shouldn't do that? (Front/bottom fans pull in, back/top fans pushing out)
gotcha on the overclocking. that's a bit of a shame. ive got a raid setup that I'm putting in as well, I figure an extra 30 watts per disk plus a bit in case I stick more disks in should keep me good and happy. as far as the case goes, buttons and ports on top is very inconvenient for me, side or front is a lot better, with side being somewhat preferred. I chose the fans because the case I had chosen only came with one. yeah the 580 and 660p are bundled
they're all 2.5 wd blacks, I'll use the 2.5 bays it has and dig out some drive bay converters for the rest
Woah, maybe old 5.25 10K RPM drives. Modern 3.5" drives use pretty much max 10W on full-tilt, and more like 3W idle.
the drives in question arent really modern. 2.2 amps at full tilt so 24 watts or so
Alright, thank you. I'm glad I got the fans right, at least. Got a new questions. Now that I've been refred to using HWiNFO, I've got it monitoring my idle, all that I've been doing is sitting on the desktop and using firefox to ask this question. My avg is 40c, which seems fine, but I'm getting these spikes to almost 50c when sitting on desktop, and 55 when having firefox going. This doesn't seem normal? Or is it, since this is the first time I'm taking a closely monitored look on how a cpu works? https://files.facepunch.com/forum/upload/527/819cc3c9-d991-44fe-bcbf-097b14017990/Untitled.png https://files.facepunch.com/forum/upload/527/d76300cb-1b1d-4d79-8af7-e83a764cd123/Capture.PNG
It's probably using the 5.5v rails, so 2.2*5.5 = 12.1.
maybe, i'd have to dig the drives out of my case to confirm
Looking at idle temps is pointless because you always get short bursts of activity from background programs which will cause temperatures to jump around sporadically from second to second. What really matters is that the max temperature stays within reasonable limits, so monitor your temps when running a stress test that pushes all cores to 100%, if you can do that without going crazy high, then you have absolutely nothing to worry about.
So I downloaded OCCT and commenced the 1 hr test, while keeping HWiNFO running in the background. OCCT was showing normal, 40c temps for 100% cores, which I thought was odd, so around 9 minuets into the test I double checked HWiNFO and that was saying 95c. I immediately stopped the test. I'm a little bit frustrated because I'm not sure what I'm supposed to interpret from this. Was I reading the OCCT graph right? What should I be looking for? https://files.facepunch.com/forum/upload/527/237d9428-4982-45d8-99bd-d4ea3ba3d45f/2019-03-30-13h39-CpuUsage-CPU Usage.png https://files.facepunch.com/forum/upload/527/2f905dea-b2b5-4a56-900f-d06382b13199/2019-03-30-13h39-Frequency-Bus.png https://files.facepunch.com/forum/upload/527/c44d9527-9c0d-4df6-92d6-fe563e92a42c/2019-03-30-13h39-Frequency-CPU0.png https://files.facepunch.com/forum/upload/527/c3b49665-6cac-4758-be8f-f12f90758c77/2019-03-30-13h39-Memory Usage-Memory Used.png https://files.facepunch.com/forum/upload/527/6ca89b96-ab5b-4b5e-a6f4-30409a318477/2019-03-30-13h39-Temperature-GPU.png I apologize again for the last 5 pages being basically me doing babys first computer assembly and troubleshooting.
So I should maybe clean the stock paste and apply some on my own? Any recommendations on a paste?
Yeah looks like it's hitting the thermal limit, you can see that the frequency drops over time to prevent the temperature from going even higher. This means that your CPU is automatically keeping itself safe so that it won't get damaged in any way, but the performance is also a little worse than it could be. That graph is showing GPU temperatures for some reason though? You said you were messing around with the cooler and remounted it without changing the paste, right? It's possible that some air bubbles or something got in there. I'd just clean off the old paste and replace it. Then just lay down the cooler flat and screw it in firmly (don't force it too hard though) so that it's making real good contact with the CPU.
I like Noctua NT-H1, it's a cheap and safe choice. Many classic options, like Arctic Silver and MX-4 aren't really that good anymore. Kryonaut is just about the best, but it's a little pricier and won't help you too much. It's mostly for hardcore overclocking. Have you changed your fan curve yet? I recommend making your cooler hit 100% RPM at 60C to take maximum advantage of frequency boosting. Oh yeah I forgot, while you're in the BIOS, double check that Precision Boost Overdrive and XFR2 is enabled, I think ASRock disables it by default.
The temps are fine, and I don't see anything out of the ordinary for a stock cooler. If you want 100% boost all the time, you are going to need a beefier cooler. I still recommend the Dark Rock Pro 4. It's expensive, but probably the last cooler you'll ever need.
Doesn't fit in every case or with all RAM, not by a long shot
warranty is a good point, I was just looking at mtbf for longevity. figured 100,000 hours is like 10 or 12 years and should be plenty, but I'd be SOL on a crib death. the seasonic souunds good.
I disagree because all of my stuff uses MX-4 and I'd say it's great, though in fairness I've not tried Noctua's TIM.
Yeah MX-4 is fine, it just isn't really the king anymore and isn't priced compellingly. Arctic Silver is the main one that's outdated.
Which reminds me, I needed to buy some more. Thanks.
So how much better can NT-H1 really be? Looking at paste reviews is usually kinda boring cause the differences are super small unless you're going for super high-end stuff. Over here you can get MX4 dirt cheap, less than 5€ for a small tube, while noctua costs a decent bit more. Maybe it depends on the region, but I'd say that's a pretty compelling deal.
If I remember correctly, MX4 has uncomfortably high electrical conductivity, and NT-H1 has very low conductivity.
I thought that was only a problem with AS and similar metal-based stuff? MX is marketed as non-conductive.
dude I don't fuckin know I'm basically talking out of my ass and half-remembering things I read a year ago
Hi yes I'd like to make a formal thank you for posting this picture as I have never seen one that accurately and succinctly depicts how fans should be set up. I love you.
Cryorig H7 beats its ass at that price point I hate to be so much of a contrarian in this thread, I'll tone it down if it's an issue
Sorry, you need to Log In to post a reply to this thread.