The case cooling can be tested by leaving the side panel off. It sounds like it spikes very quickly so its either the cooler or bios. What clocks is the CPU using? You said it sits at 1.4v then ends up at 1.2v or so. It may be down clocking.
It it were me I wouldn't stress test in the current climate but try an undervolt to ca. 1.25v and then check temps and then stress test. I'm not sure of the ideal volts for zen2 but zen+ is 1.25v stock for a 1600af but it will happily run at 1.15ish.
I'm interested in case fans because I like to run folding@home overnights and a 105W CPU plus a 250W graphics card makes for a lot of heat to take out of the case once I get the CPU cooler doing it's job properly.Originally Posted by watercooled
I'm on BIOS 7C37vA7, dated 2020-01-13. It's the most recent non-beta BIOS. There is a beta (vA87, 2020-07-03) but I'm dubious about a beta BIOS.Something doesn't sound right TBH, from the reviews I've seen online and the experience of friends who have ran the same CPU with stock cooler, it doesn't seem like load temps should be shooting up like that.
It could also be an incorrect temperature reading but the dropped clock suggest otherwise.
What BIOS version are you on?
AS I type I'm folding on GPU only and I'm getting CPU temps of 81-84'C with one core at 4GHz, one at 2.3GHz, and the rest jumping around between 600MHz and 1.5GHz
My case is a Fractal Design Define 7. Taking the side and top off doesn't change how fast the CPU heats up.
It idles at 1.4v which seems way too high, then drops to 1.35v one-core or 1.25v all core under load.
I'm going to have a poke through the BIOS tonight and try & cap the maximum voltage the board will supply
It can be difficult to reliably read Ryzen's idle voltage and there were a few people overly worrying about this during launch - the act of polling for a measurement can cause the voltage to spike so it's hard to get a true idle. And the 'idle' clocks reported by some software may in fact be the voltage used to enable single core boost, so even higher than full load.
And yeah that BIOS should be recent enough to eliminate any early bugs. It's a bit of a mystery then. Do you have any other compatible coolers, just to try?
Edit: Besides maybe for experimenting, I wouldn't mess too much with voltages because it may cause instability and/or limit clocks. The way modern CPUs use boost clocks, the 'max' voltage isn't a single, fixed value any more.
Sounds like some dodgy voltages (IIRC I set 1.2V on a friends rig) or a cpu with bad thermal contact either above or below the IHS
I don't have another AM4 cooler in the house unless the cooler from an old Phenom 9950 would fit? That's socket AM2+
I took the heatsink off & replaced the thermal grease with some old Arctic MX4 I had. While I was at it I found a tiny recessed switch on the side of the cooler that was marked L H. It was on the L setting. On the assumption that this was Low & High for the fan I switched it to H.
It hasn't made any difference to idle temps but under load it boosted higher (4.1GHz all core) and took a lot longer (several minutes) to reach 95'C and clock down a bit. After 10 minutes it seems to be steady at 95'C and 3.875GHz, which is ~100MHz faster than it used to settle at.
The fan is noticably louder now, but not enough to bother me.
Having looked online it does seem it's a fan speed switch. Not owning one I wasn't aware of that.
Out of interest, and obviously being careful about touching something potentially hot or sticking fingers into moving fans, how hot is the heatsink under load once it's had a chance to warm up? Reason for asking, a lukewarm/cool heatsink despite 90+ temps would concur with poor thermal contact. A hot heatsink suggests it is sinking heat away to somewhat effectively.
AMD cooler mounts haven't changed since before AM2 believe it or not! I was just curious whether something might have been wrong with the heatsink mounts - it does go on secure right? A friend of mine built a system back in the Bulldozer days and hadn't flicked the tension arm round for fear of breaking something, so it was just pressed lightly against the CPU, not doing a great job of cooling it!
As Marcvs said, it's also in the back of my mind whether there might be something wrong with the heatspreader on the CPU but I'd expect that to be a pretty rare fault.
This and similar measurements are what's making me think the stock cooler should be performing better than it is:
From https://www.techspot.com/review/1875...%20temperature.
Edit, another review and another load:
From https://lanoc.org/review/cpus/7985-a...900x?showall=1
That's using an aftermarket heatsink, and probably with only CPU load on an open testbed (so no hot air circulating in a case). So if f@h is hitting the CPU hard enough it might actually be normal now, if it's taking a while to get that hot, and the heatsink itself is sinking heat away properly. If that's the case, it might call for a heatsink upgrade if you're planning to be doing that sort of loading frequently.
Last edited by watercooled; 19-07-2020 at 12:51 AM.
It's small enought that I missed it completely on first inspection and I thought I'd looked quite closely! :-D
The heatpipes and heatsink are hot to the touch. They feel about the same as the backplate of the graphics card under load (GPU temp 80'C), and noticably warmer than the passive heatsinks on the VRMsOut of interest, and obviously being careful about touching something potentially hot or sticking fingers into moving fans, how hot is the heatsink under load once it's had a chance to warm up? Reason for asking, a lukewarm/cool heatsink despite 90+ temps would concur with poor thermal contact. A hot heatsink suggests it is sinking heat away to somewhat effectively.
I remember that feeling about the tension arm on AMD heatsinks from my Athlon64x2 days and I've definitely locked it corrrectly this time!AMD cooler mounts haven't changed since before AM2 believe it or not! I was just curious whether something might have been wrong with the heatsink mounts - it does go on secure right? A friend of mine built a system back in the Bulldozer days and hadn't flicked the tension arm round for fear of breaking something, so it was just pressed lightly against the CPU, not doing a great job of cooling it!
After running f@h on both CPU and CPU for a couple of hours I've noticed that the CPU eventually drops to base clocks (3.8GHz)So if f@h is hitting the CPU hard enough it might actually be normal now, if it's taking a while to get that hot, and the heatsink itself is sinking heat away properly. If that's the case, it might call for a heatsink upgrade if you're planning to be doing that sort of loading frequently.
I'm guessing that's because the GPU (980Ti) is putting out enough heat to warm the case as a whole making the CPU cooler less efficient.
My normal "heavy" load is overnight video transcoding on the CPU with Handbrake and f@h on the GPU, so I think it's worth looking at an aftermarket cooler. From Hexus reviews I was liking the look of either the Noctua NH-D15, or maybe trying an AIO with the Corsair iCue H115i RGB Pro XT. Unfortunately the only place I can fit a 280mm radiator is in the front of my case and I'm not sure I want a radiator on my air intake.
Perhaps it's 'normal' now then. It's positive that the heatsink itself is hot, showing it's doing its job properly.
And yeah, having a few hundred watts of extra heat pumped into the case from the GPU isn't going to help the CPU heatsinks's cooling ability either.
The Wraith Prism is ostensibly a fairly good cooler itself, somewhere around the venerable Hyper 212 from what I've seen online, so it takes some beating on a budget.
Is there any room to change the fan profile or is it already quite fast/loud?
I always knew I'd need to add extra case fans at some point...
I'm getting base clocks all-core without exceeding maximum safe temperature so it's adequate for normal use :-)The Wraith Prism is ostensibly a fairly good cooler itself, somewhere around the venerable Hyper 212 from what I've seen online, so it takes some beating on a budget.
Assuming the BIOS fan curve is being implemented properly, the fan hits 100% when the CPU reaches 70'C, and it's quite loudIs there any room to change the fan profile or is it already quite fast/loud?
Yeah that's fair enough, was just a thought - free solutions are usually better than having to spend money!
If you do choose to upgrade, I'd be interested to know how it goes. It should be useful for anyone in the same boat, too.
My next step is going to be a more detailed look at the BIOS settings to make sure there isn't some voltage boosting going on by default that isn't really needed.
Can you suggest software that would let me monitor or maybe even control fan speeds from the desktop? I looked at speedfan but it can't see any fans on my system.
I'm not too sure sorry, Speedfan is what I've used in the past and many alternatives seem to be subscription-based software. Maybe a silly question but did you make sure to run as admin? It wouldn't really surprise me if it's not supported though, I don't think it's been updated in some time.
What's the reason for wanting more control? Idle fan speeds? Modern BIOSes tend to have tools to tweak the fan curve, something I had to do with my i7 7700 system as the default curve was ridiculous.
As for just monitoring fan speeds, a few utilities can do that, including HWMonitor and HWiNFO.
Now I think of it, HWMonitor Pro (paid) can control fan speeds I think, so might be worth trying the trial of that?
Mostly I want realtime info on what the fans are doing. Being able to tweak fan curves without constantly having to restart & go into the BIOS would be a nice bonus though.
Have you installed the Dragon Centre software, you should be able to set a custom profile in there
Jon
There are currently 2 users browsing this thread. (0 members and 2 guests)
Posting Permissions
LinkBack URL
About LinkBacks
Reply With Quote
