|
Microcenter having 5900X's readily available at MSRP got me to upgrade my 3900X. I was looking for more consistent VR performance and the 5900X is helping me do stuff like get a consistent 120 FPS in games like Project Cars 2 and No Man's Sky, that was happening before on the 3900X paired with a 3090. I'm glad to see that PBO actually works on these CPU's unlike my old 3900X, the 3900X basically never ran at the 4.6ghz clocks it was supposed to run at as a max boost, but with some quick PBO twiddling I see the 5900X able to clock as high as 5.025ghz and do like 4.5ghz all core clocks in benchmarks. It also looks like there's less of a gap between the 'good' cores and 'bad cores', at least with my CPU. What I'm a bit disappointed with so far is memory, this 5900X seems to not want to take memory clocks I was able to handle on the 3900X. I was running tightly tuned CL14 timings on the 3900X with my Crucial Ballistix 16x2 CL 16 kit but that's causing errors right away when testing memory, it also doesn't want to boot at all at 3800 CL16 with timings I used to be able to boot with on the 3900X and gave me errors quickly at 3733 CL16 as well. I was using Ryzen DRAM Calculator as that's helped me in the past get usable clocks, but nothing it's giving me now is working and I'm just rocking XMP for now. I'm using an X470 Crosshair VII Wi-Fi motherboard still as I didn't see a compelling reason to jump to X570.
|
#
?
Jun 13, 2021 08:40
|
|
|
|
| # ? May 22, 2024 17:29 |
|
|
SwissArmyDruid posted:https://www.nextplatform.com/2021/06/10/why-ibm-is-suing-globalfoundries-over-chip-roadmap-failures/ If AMD had designs for 10nm and 7nm I could see AMD jumping in on the lawsuit. Considering lead times on designs, this means if Glofo was going to make 10nm available to AMD it’d have been just after the sale with a delivery time of Q2 2016 since IBM would get first dibs in 2015. That puts Zen1 and Vega as potential suspects for ported designs having originally been intended for 10nm.
|
|
#
?
Jun 13, 2021 10:06
|
|
|
Hey y'all need your help. I'm dealing with serious system instability. Several bluescreens a day. One point I'm considering is whether my high CPU temps are causing stability issues. The crashes and blue screens are unpredictable as to whether when they will occur and what error code they throw. It may be during gaming, but it may also be while displaying the desktop with no program open. However, today I've found that I can get the system to predictably crash: by stress testing it. Below is a chart of temperatures and load during repeated Prime95 stress tests. During all of them, the computer crashed with a bluescreen*.  The CPU is a Ryzen 5 3600, stock clock. It's cooled by a NZXT M22 AIO. When building the system, I became aware that this CPU ran hot even when idle. Idling temps on a warm but not hot day are in the 60C range. On load during, temperatures tend to rise up sharply and then plateau But it was explained to me that this was how hese Ryzens were made. You also notice a see-saw patterns in the ide temps, where the CPU will heat up and cool down in cycles of 10-20 seconds. I've been told back then that this, too, was normal for Ryzens, but I'm not so sure now. What is a good next step to do or look at? * the bluescreen messages are always different; during the test it was "APC inde mixmatch", " System service exception", "Page-fault in non-paged area".
|
|
#
?
Jun 13, 2021 11:07
|
|
|
Ryzens tend to ramp up quickly which makes temperatures spike very briefly, but constant >60°C in idle is definitely not normal, but also not something that should trigger random crashes (the CPU will shut off to protect itself, but well before that it will start to throttle). I think Ryzen 3000 starts thermal throttling at 95°C but I'm not sure. How is your RAM set up? orcane fucked around with this message at 12:12 on Jun 13, 2021 |
|
#
?
Jun 13, 2021 12:07
|
|
|
Lord Stimperor posted:Hey y'all need your help. Did you remove the plastic cover protecting the bottom of the AIO before installing it? Is the pump actually working? Those idle temps seem high enough to suggest the cooler isn't working at all. Prime95 has several modes: (large, blend, small), do they all crash, or does one mode crash more than the others? If there is a difference between modes it could help narrow it down (crashing frequently on small FFT but not as often on large would indicate CPU/cache issues, the opposite would indicate memory issues).
|
|
#
?
Jun 13, 2021 12:24
|
|
|
Ryzens run hotter than Intel CPUs, but not that hot. I would definitely inspect your cooling setup.
|
|
#
?
Jun 13, 2021 12:50
|
|
|
Dr. Video Games 0031 posted:Ryzens run hotter than Intel CPUs, but not that hot. I would definitely inspect your cooling setup. Indiana_Krom posted:Did you remove the plastic cover protecting the bottom of the AIO before installing it? Is the pump actually working? Those idle temps seem high enough to suggest the cooler isn't working at all. I remember that I removed something from the cooler before installing it. I think that the pump is working, because when I turn the PC on I can hear a faint gluc gluc if I hold my ear next to it. Is the sound proof enough of a working pump? I have a bit of unopened thermal paste laying around. I'll check if it is still good tonight and try to re-seat the cooler, or even move the radiator. I have heard that bad placement of pipes and radiators can mess with the cooling performance. Does this look alright? The pipes go into the second compartment of the PC. There they go to the radiator, which as mounted at the top. The radiator fan is mounted at the bottom with the idea that it blows air through the radiator from below, transferring heat out the top of the case.  Regarding Prime 95, I ran the blend test. Once I have confirmed that the cooler fits and works as intended, I'll see if I can use prime95 to dissociate between memory and CPU issues. I think that's exactly what I need, super useful. orcane posted:Ryzens tend to ramp up quickly which makes temperatures spike very briefly, but constant >60°C in idle is definitely not normal, but also not something that should trigger random crashes (the CPU will shut off to protect itself, but well before that it will start to throttle). I think Ryzen 3000 starts thermal throttling at 95°C but I'm not sure. Yeah, when I built this thing I think I was researching whether these temps were accurate, and finding others have high temps as well sort of accepted this. Back then the system also ran rock solid. I have two 16GB memory modules, which are in slots 2 and 4 (see pic). This is what HWInfo has to say about them:  Random note: here's a bunch of idle / web browsing temps. I have the impression that messing with the pipes and pulling / pushing on the cooler can slightly influence the temperature. But I can't really decide whether that's really happening or whether that's just oscillations oscillating. 
|
|
#
?
Jun 13, 2021 15:40
|
|
|
Lord Stimperor posted:Hey y'all need your help. Try and bump voltage up a hair. Also you can try and adjust LLC. For me, my RAM memory was not actually good.
|
|
#
?
Jun 13, 2021 16:24
|
|
|
Lord Stimperor posted:I remember that I removed something from the cooler before installing it. I think that the pump is working, because when I turn the PC on I can hear a faint gluc gluc if I hold my ear next to it. Is the sound proof enough of a working pump? No - it’s possible something in the pump is broken, or that there’s an air bubble trapped inside (shouldn’t be possible with an AIO, but manufacturing defects do happen). If you have access to a different cooler, I’d try swapping it out. Also, how is your fan control set up on the radiator fans? If they’re on chassis fan headers, the system might not realize that the CPU temperature going up means it needs to spin up CHA1 and CHA2 or whatever the radiator fans are connected to. You might try just forcing everything to 100% to see if that helps things. redeyes posted:Try and bump voltage up a hair. Also you can try and adjust LLC. For me, my RAM memory was not actually good. If the system is hitting 100C at stock clocks then bumping the voltage is a bad, counterproductive idea.
|
|
#
?
Jun 13, 2021 18:01
|
|
|
I started getting several bsods a day plus random crashes only in games, never during a stress test, with my old 1700. Replacing the thermal paste and giving the ram some more voltage has seemed to stabilize it for now. .
|
|
#
?
Jun 13, 2021 18:07
|
|
|
Lord Stimperor posted:I have heard that bad placement of pipes and radiators can mess with the cooling performance. Does this look alright? The pipes go into the second compartment of the PC. There they go to the radiator, which as mounted at the top. The radiator fan is mounted at the bottom with the idea that it blows air through the radiator from below, transferring heat out the top of the case. So to be clear here, is the radiator mounted horizontally at the top of the case with the inlet/outlet facing down (good) or vertically anywhere with the inlet/outlet facing horizontally at the top (generally quite bad)?
|
|
#
?
Jun 13, 2021 18:42
|
|
|
Lord Stimperor posted:I remember that I removed something from the cooler before installing it. I think that the pump is working, because when I turn the PC on I can hear a faint gluc gluc if I hold my ear next to it. Is the sound proof enough of a working pump? That does sound wrong. You might want check this Gamers Nexus video about watercooling orientation and air bubbles. Stop Doing It Wrong: How to Kill Your CPU Cooler (AIO Mounting Orientation)
|
|
#
?
Jun 13, 2021 18:45
|
|
|
For troubleshooting purposes, see what idle temps are like with an air cooler (if you have one handy). I have a 3600 using the stock cooler and under light usage (browsing, whatever) it runs at between 35 and 40C. That'll tell you whether it's something weird with the CPU/mobo/case airflow situation, or whether it's something wonky with your AIO.
|
|
#
?
Jun 13, 2021 20:18
|
|
|
I'd be checking the paste on your CPU. My 3700x on idle on the stock fan is about 20 degrees lower then that.
|
|
#
?
Jun 13, 2021 20:19
|
|
|
TLDR the AIO is borked, I think. I'm gonna order an air cooler tonight, I've had it with this bitch that takes up so much space and where you can't see whether it's working. Long story: I've pulled out and cleaned most of the components. I think that I used a thermal pad when I put on the CPU cooler the first time. Unfortunately it only covered about 2/3 of the heatspreader, which is a bit disappointing. I put on some fresh paste. I also changed the location of the radiator. Previously, it was located at the top of the case, as I reasoned hot air would want to naturally want to travel upwards. Downside is that the pump also had to force the water upwards. Now I've mounted the radiator to the only other possible location in the case, the back panel. The in- and outputs of the radiator are now below the CPU. If I understand correctly, this arrangement should make it better for the water to flow. The thing is, the radiator stays cold all the time. The CPU is running at 80+ degrees. The tube and cold plate casing get hot. But the radiator remains at the same temperature as the case, cool to the touch. That means its hosed, right?
|
|
#
?
Jun 13, 2021 21:53
|
|
|
Something is seriously not right. My 3600X idles in the 30s with a good air cooler, and stays under 75 when really working out. The temps you describe make it sound like the cooler is just not working.
|
|
#
?
Jun 13, 2021 22:13
|
|
|
Yeah that's what I figured in my last post, CPU being hot but radiator cold says to me that the water isn't circulating. I'm currently going back and forth between a CoolerMaster Hyper 212 with RGB bling, a be quiet! Dark Rock, and a Noctua NH-L12. For the first two it comes down to performance; what speaks for the latter is that I just don't like big towers very much. That was the reason I wanted the AIO in the first place.
|
|
#
?
Jun 13, 2021 22:18
|
|
|
I'm running a be quiet! Dark Rock Slim and it's been great, but I also went with a dark and brooding build esthetic. The only lights in my case is the mild orange glow from the thin LED strip over the I/O backplate on the motherboard itself. Everything else is black.
|
|
#
?
Jun 13, 2021 22:24
|
|
|
Lord Stimperor posted:TLDR the AIO is borked, I think. I'm gonna order an air cooler tonight, I've had it with this bitch that takes up so much space and where you can't see whether it's working. There are some suboptimal ways to install an AIO, but the only outright bad/potentially damaging way is if the pump (which in AIOs is almost always on the CPU block) is on the top of the loop. That wasn't a problem in your first configuration though, and if there's a part of your radiator that's still above the pump then it's fine in your second setup too. It's seeming pretty likely that your pump is to blame here (which is a total loss if so), but maybe you can try wiggling around your tubes and stuff a bit to move any potentially obstructive air pockets that are stuck somehow. Lord Stimperor posted:Yeah that's what I figured in my last post, CPU being hot but radiator cold says to me that the water isn't circulating. I'm currently going back and forth between a CoolerMaster Hyper 212 with RGB bling, a be quiet! Dark Rock, and a Noctua NH-L12. For the first two it comes down to performance; what speaks for the latter is that I just don't like big towers very much. That was the reason I wanted the AIO in the first place. I have a big dual-tower cooler (Dark Rock Pro 4), and it is indeed a pain in the rear end to install. Or rather, it's pretty easy to mount onto your motherboard, but fitting it into your case and plugging everything in around it is what's a pain unless you have a super spacious case. Honestly, if you're not gonna be overclocking or anything, then there's nothing wrong with a good-quality low-profile cooler. If even stock coolers are often good enough in non-overclocked systems, then a Noctua low-profile cooler will certainly be. (I'd probably spend the extra $10 on the taller variant, but that's just me). Your case will need better airflow with an air cooler though, so maybe pick up an extra case fan to go in one of those empty front slots while you're at it.
|
|
#
?
Jun 13, 2021 22:27
|
|
|
Dr. Video Games 0031 posted:There are some suboptimal ways to install an AIO, but the only outright bad/potentially damaging way is if the pump (which in AIOs is almost always on the CPU block) is on the top of the loop. That wasn't a problem in your first configuration though, and if there's a part of your radiator that's still above the pump then it's fine in your second setup too. It's seeming pretty likely that your pump is to blame here (which is a total loss if so), but maybe you can try wiggling around your tubes and stuff a bit to move any potentially obstructive air pockets that are stuck somehow. CaptainSarcastic posted:I'm running a be quiet! Dark Rock Slim and it's been great, but I also went with a dark and brooding build esthetic. The only lights in my case is the mild orange glow from the thin LED strip over the I/O backplate on the motherboard itself. Everything else is black. CaptainSarcastic posted:I'm running a be quiet! Dark Rock Slim and it's been great, but I also went with a dark and brooding build esthetic. The only lights in my case is the mild orange glow from the thin LED strip over the I/O backplate on the motherboard itself. Everything else is black. Ended up with the dark rock slim. I originally wanted the cheaper cooler master 212 rgb. But it pissed me off that it came with 2 cables and a dongle thingy, so I went with dark and elegant. Reviews said that the low-profile noctua might be a bitch with placing memory and I can't be arsed with putting another complex thing in there, especially since I'll be swapping and testing things for a while now. Thanks everyone for getting me to check the cooler. with a bit of luck this saved my cpu from permanent damage.
|
|
#
?
Jun 13, 2021 22:36
|
|
|
The Dark Rock Slim is tall, but has no problem clearing my RAM, which with 4 sticks w/heat spreaders is not always an easy feat. I'd originally tried to put a black version of the Cooler Master Hyper EVO 212 or whatever it's called in, but the mounting for it was a huge pain in the rear end and one of the mounting screws broke as I tried to install it, which is what led me to the Dark Rock cooler instead. The mounting process is much more rational, and once I set the fan curves it is incredibly quiet.
|
|
#
?
Jun 13, 2021 22:46
|
|
|
Lord Stimperor posted:I also changed the location of the radiator. Previously, it was located at the top of the case, as I reasoned hot air would want to naturally want to travel upwards. Downside is that the pump also had to force the water upwards. Now I've mounted the radiator to the only other possible location in the case, the back panel. The in- and outputs of the radiator are now below the CPU. If I understand correctly, this arrangement should make it better for the water to flow. I don't think gravity would matter. All the water that needs to be pumped up will come down the other tube and help the pump. The way I see it AiO can have two possible problems. The pump can be failing, or there are air trapped in the wrong place. The latter should be solvable by making sure all the air gathers in the other end of the radioator. I would do this by setting the computer on its side and raising the radiator as high as possible. Maybe also tilt the computer on different sides in case the pump still has some spot for trapping.
|
|
#
?
Jun 13, 2021 23:53
|
|
|
Lord Stimperor posted:I also changed the location of the radiator. Previously, it was located at the top of the case, as I reasoned hot air would want to naturally want to travel upwards. Downside is that the pump also had to force the water upwards. Now I've mounted the radiator to the only other possible location in the case, the back panel. The in- and outputs of the radiator are now below the CPU. If I understand correctly, this arrangement should make it better for the water to flow. Lord Stimperor posted:The thing is, the radiator stays cold all the time. The CPU is running at 80+ degrees. The tube and cold plate casing get hot. But the radiator remains at the same temperature as the case, cool to the touch. That means its hosed, right? I almost was going to ask if there's a chance it doesn't even have fluid in it, but you mentioned some glug glug noise. Out of curiosity, what motherboard header are you plugging the pump into? Since the LED/etc. is via a separate header it appears, I wonder if the header you plugged it into could be the issue, such as the BIOS not properly managing the pump. When I used AIOs, I usually just disabled any pump control options in the BIOS - the pump ran at 100% all the time, but I never really noticed it.
|
|
#
?
Jun 14, 2021 07:25
|
|
|
While idle temps should be lower and there certainly may be something wrong with it, I'd also caution against expecting too much out of a 120mm AIO to begin with.
|
|
#
?
Jun 14, 2021 19:04
|
|
|
just imo but personally I think pumps should be configured to always run at 100%. if you're in a situation where you're sensitive to pump noise, at least do a quick ramp from like 40% at an "idle" temp (let's say 30C) to bring the pump up to full speed by the time there is any load on the system (let's say start ramping at 50C and full speed at 60C). full pump speed absolutely will make a big difference in thermals on the system.Helter Skelter posted:While idle temps should be lower and there certainly may be something wrong with it, I'd also caution against expecting too much out of a 120mm AIO to begin with. I disagree with this. People expect far too little out of a 120mm rad. the AMD Radeon 295x2 pushed at least 400W core power through a 120mm rad at stock and could maintain package temps under 60C. Now obviously that's using dual waterblocks on a pair of bare dies (and I think the memory and VRM was aircooled) but that's the point. The radiator itself will push a lot of heat, even if it's "only" 120mm, if the coldplate isn't limiting it. 125W is absolutely no problem though a 120mm radiator, it's nowhere near the thermal bottleneck people think it is, you're barely warming the fluid up at that point so a bigger rad barely helps. What is limiting is how fast you can get the heat into the loop (so get that pump spinning), and a bigger radiator barely changes how fast heat moves across the IHS/coldplate (only a very slight increase from a slightly lower temperature in the working fluid). Yeah AMD 7nm processors have thermal density that makes that even trickier than usual, but that means the thermal bottleneck is even less about radiator size and even more about something you (effectively) can do nothing about. Going to a bigger radiator very rarely produces the gains people think it does, because the radiator just isn't the thermal bottleneck. It's how fast you can get the heat into the loop. People just buy big radiators and then talk themselves into believing that it must have been necessary, because gosh, thermals are already kinda mediocre, think how much worse it would have been with a rad that's only half the size! but that big giant radiator only decreases the temperature of the working fluid a little bit. Remember that processor package temps are not the same thing as the temp of the working fluid inside the radiator/AIO. Almost nobody actually measures this (at least without a custom loop). It takes a big increase in radiator area to produce a small decrease in working fluid temp and a big decrease in working fluid temp to produce a small decrease in package temps. logically, if a 295X2 is getting 60C package temps, the fluid is cooler than that, because otherwise the fluid would be heating the chip, not the other way around. So 400W in a 120mm AIO = less than a 60C working fluid temperature. That is really not very hot at all, and a CPU that is dumping even less power will be even cooler - a 150W CPU is probably like 40C working fluid temperature. The rest of the thermal differential is just how long it takes heat to move across the IHS and coldplate. and again AMD 7nm processors just have high thermals / very spiky thermals in general. It happens on air too, even with a nice heatpipe of fluid to evaporate to absorb the heat the heat just can't move out of the die quickly enough (and evaporation moves much more heat than warming up liquid - heatpipes are effectively "phase change cooling"). The die is hot but the surface of the IHS is still cold because the heat can't move out quickly enough, the only solution is to wait for the heat to flow out to the surface of the IHS. Just generally don't worry about it, it's not a big deal and doesn't seem to produce any real change in actual performance, it's a weird phenomenon coming from older processors but there are (technically-)straightforward reasons for it and if it doesn't affect performance then it really doesn't matter. Saukkis posted:I don't think gravity would matter. All the water that needs to be pumped up will come down the other tube and help the pump. mounted on top is generally fine. if you want to be paranoid throw a couple washers on the side with the tubes, so that they're forced down lower and the fluid will fall down to the side with the tubes and the air bubble will float up to the top. what you don't want to do is front/rear mounted with the tubes on top. If you're going to front/rear mount a radiator, the tubes need to be on the "down" side so that fluid falls down into the tubes. because the "up" configuration will mean that as soon as there's any air in the radiator, it's going to start sucking air, because the fluid will be trapped in a "pocket" in the bottom of the radiator. it will work for a while but over the years the fluid will evaporate out of the AIO (it's very slow but it does happen even with "zero maintenance" tubing) and you don't want the pump to suck air. A little burbling is normal but if you hear it making loud gushing or sucking noises you need to clean it and refill it just like a custom loop (or send it back if it's still under warranty). I've actually personally observed this, I have one of the very last non-Asetek AIOs (a Cooler Master Nepton 140XL) and it's almost 7 years old now, I really need to clean and refill it because it is almost out of fluid. At one point I had it mounted tubes up and it started sucking air, I flipped it and it ran fine for another couple years. Paul MaudDib fucked around with this message at 00:03 on Jun 15, 2021 |
|
#
?
Jun 14, 2021 21:07
|
|
|
Paul MaudDib posted:just imo but personally I think pumps should be configured to always run at 100%. if you're in a situation where you're sensitive to pump noise, at least do a quick ramp from like 40% at an "idle" temp (let's say 30C) to bring the pump up to full speed by the time there is any load on the system (let's say start ramping at 50C and full speed at 60C). full pump speed absolutely will make a big difference in thermals on the system. A Paul pro-post, but especially this bit and I'll give one more reason for it. Pumps and fans with cheap-rear end bearings, especially junk journal bearings are far more likely to wear out from changes than static conditions. Even a garbage bearing will usually be fine for a very long time if you just ask it to do the same thing forever. This is often why you often see 100 year old machinery that still runs fine - it's spent 100 years only spinning up once and down once per day, or sometimes it's just been spinning 24/7 forever in a static load condition, happily suspended on its lubricant and experiencing literally zero physical wear.
|
|
#
?
Jun 14, 2021 21:34
|
|
|
For the little toy pumps in AIOs, yeah run them at 100% all the time. For a custom loop with a massive D5 pushing it though, go ahead and set it down to like 30-40% and keep it there forever.
|
|
#
?
Jun 14, 2021 22:18
|
|
|
NVM I'm posting dumb
Lord Stimperor fucked around with this message at 00:21 on Jun 15, 2021 |
|
#
?
Jun 15, 2021 00:16
|
|
|
Lord Stimperor posted:Hey y'all need your help. despite all the stuff I just said about how Ryzen has spiky thermals, I would say idling in the mid-60s or peaking over 90C is not normal even for 7nm ryzen. And the chip really should be throttling before it gets above 90C. Start by redoing your thermal paste, and use a real thin layer. Given the chiplet placement I dunno if a pea-style or spread is better, but either way you need to keep it real thin and try to avoid air bubbles. Don't use something fancy like Kryonaut (as that is tougher to work with and can go bad over time), just some NT-H1 or Protronix series 7. Bringing your temps down will generally promote stability and those temps in the 90s may actually be the direct cause of the crashes. if it's still running hot, try reducing your PPT target and ramping up your cooling to max. if you are using an AIO (as per the previous discussion) ramp the pump up to a fixed 100% speed (make sure to enable PWM mode if appropriate), and regardless of air cooled/water cooled punch the fans to a constant 100% and just try to keep it as cool as possible for now. The PPT reduction will help keep it cool too (although this will reduce clocks, so it's changing a couple variables at one time). the next thing you need to do is disable XMP / AMD AMP (actually just reset your whole BIOS) and just run the stock AMD memory spec (or even JEDEC 2133/2400) for a while. all of those errors are potentially caused by hardware faults, and specifically memory is a real possibility given the page fault error. I’ve seen the “system service exception” on a dying memory controller as well. there's no guarantee that disabling XMP/AMP helps, If you've started to really cook the memory controller/infinity fabric - disabling XMP/AMP will reduce the voltage the mobo applies, which may mean the memory controller is no longer stable even at stock clocks on the stock VSOC voltages. If you're still getting the errors you could try bumping up VSOC (but keep the speeds/timings of RAM/infinity fabric/etc at stock settings - do not enable AMP/XMP) and see if that resolves it. Use your current vSOC readings (get them from a monitoring tool if they’re on “auto”) as a guideline of what’s currently working, just run them at stock clocks or JEDEC clocks. I wouldn't necessarily run it like that long term but if it resolves the crashes it tells you what the problem is, and really if that is the problem it's all just a matter of time for the chip so you might as well get some more stable life out of it. hopefully it's just the temperatures getting out of hand, hot chips are less stable, but motherboards can often be "helpful" about pushing up vSOC and other internal voltages to get AMP/XMP or manually-entered clocks stable and over time that can destroy the memory controller/infinity fabric. overclocking the memory controller is still overclocking even if people widely recommend it, and you have to be really really careful about what the motherboard is doing in the background with voltages. Paul MaudDib fucked around with this message at 00:39 on Jun 15, 2021 |
|
#
?
Jun 15, 2021 00:31
|
|
|
Paul MaudDib posted:despite all the stuff I just said about how Ryzen has spiky thermals, I would say idling in the mid-60s is not normal even for 7nm ryzen. And the chip really should be throttling before it gets to 90C, even with Prime95. This is a helpful list of things to do. I've already kicked out the AIO. Wasn't trusting it to be functional anymore. A small tower air cooler and fresh paste should arrive tomorrow. That should hopefully resolve the temperatures. After that I have a bunch of memory suggestions from this thread to work through. Debugging random crashes has been a lost cause in the past but this time I'm kinda optimistic, what with all the useful ideas in this thread.
|
|
#
?
Jun 15, 2021 00:41
|
|
|
Paul MaudDib posted:despite all the stuff I just said about how Ryzen has spiky thermals, I would say idling in the mid-60s or peaking over 90C is not normal even for 7nm ryzen. And the chip really should be throttling before it gets above 90C. https://www.youtube.com/watch?v=EUWVVTY63hc&t=1s Xaris fucked around with this message at 03:17 on Jun 15, 2021 |
|
#
?
Jun 15, 2021 03:13
|
|
|
Paul MaudDib posted:Start by redoing your thermal paste, and use a real thin layer. Given the chiplet placement I dunno if a pea-style or spread is better, but either way you need to keep it real thin and try to avoid air bubbles. Don't use something fancy like Kryonaut (as that is tougher to work with and can go bad over time), just some NT-H1 or Protronix series 7. Bringing your temps down will generally promote stability and those temps in the 90s may actually be the direct cause of the crashes. You're right that Kryonaut is a pain to work with. It frequently scratches the heck out of your cooler and/or ihs when you remove it a year or three later assuming you get three years out of it -- especially with am4 where you need to twist the heatsink to remove it. And it just sucks to spread and is easy to mess up. It doesn't even perform that well vs other pastes that are easy to work with. It seems fueled by hype more than anything. I'd just do an X but with a disconnected pea in the middle with modern CPUs. AMD especially & especially if you're using something non-conductive. It's easy to screw up by putting too little. It's hard to screw up by putting too much of a non-conductive tim given any excess will just ooze out if you apply sufficient pressure. I mean hopefully you don't put too much and have it ooze out, but if you do the thermal performance should be within 1c-2c of a perfect application which is nothing compared to the people who have throttling problems. I used to be a "spread a thin layer using a cc/paint spreader" person. All it takes is your tim drying out once, watching and reading a few paste application review videos, and then deliberately putting too much just to see what happens thermally to be converted into the school of "just do it the lazier way the results are the same or maybe even better." Khorne fucked around with this message at 05:03 on Jun 15, 2021 |
|
#
?
Jun 15, 2021 04:51
|
|
|
Protronix Series 7 is just great because it's cheap. If it's not going to be super heavily overclocked then it doesn't really matter and you can get a 20g tube for less than the price of a 3.5g NT-H1 tube. https://www.amazon.com/Protronix-Thermal-Performance-Heatsink-Compound/dp/B01MU2FLVJ but I will grant that the conductivity is not as good, it looks like NT-H1 is over twice as conductive and so given that we're debugging thermals you should probably just get the NT-H1 here. But I use Protronix on anything that I don't intend to keep together for very long, or that won't be overclocked and won't see benefit from a higher end paste.
|
|
#
?
Jun 15, 2021 06:28
|
|
|
My 3600 is weird. It idles at around 40c but never goes over 70ish when under load.
|
|
#
?
Jun 15, 2021 07:23
|
|
|
ChazTurbo posted:My 3600 is weird. It idles at around 40c but never goes over 70ish when under load. Fan curve also has a lot to do with this. It's definitely possible to have a fan curve designed to be very quiet during idle that ramps up quickly under load.
|
|
#
?
Jun 15, 2021 07:33
|
|
|
My 3600X idles around 35 or so, but I have the fan curve set a little high to avoid the obnoxious ramping it would do at stock. My CPU cooler is pretty much silent until it hits max RPMs, so it isn't a problem.
|
|
#
?
Jun 15, 2021 07:39
|
|
|
My 5800X idles at 50, because I want my fans to be silent when web browsing and playing movies and videos. As others have noted, it's all relative based on fan speeds and surface area, ambient temperature, preferred volume, background noise, etc etc etc These Ryzen chips seem to have reliable thermal management systems built in that hold the chips at their rated 90c, throttling as needed. Perfectly safe for the chip, though you will lose some amount of performance depending on how much it slows itself down. Try not to worry! Hard disagree on buying expensive thermal compound. It's what, a 3 degree difference between the best and worst thermal compound?
|
|
#
?
Jun 15, 2021 09:50
|
|
|
LRADIKAL posted:Hard disagree on buying expensive thermal compound. It's what, a 3 degree difference between the best and worst thermal compound? it's overall not a big expense, you're already dropping between $350 to $800 on a cpu. $10 for a few g's of anything of reasonable quality (which can really be anything, there's a ton of just decent quality from kryo to arctic mx to noctua to whatever) is fine, you aren't going to be saving a whole lot getting "worst compound". certainly don't buy a $50 tube of boutique nanocarbon xxxx diamond powder gold crypto-powered electron rgb thermal paste, but i have not seen any real substantial price difference between "best or top 5 best" thermal pastes and "worst" thermal pastes. Xaris fucked around with this message at 10:08 on Jun 15, 2021 |
|
#
?
Jun 15, 2021 10:05
|
|
|
Agreed, I wouldn't pick the worst compound either. Something in the middle from some brand.
|
|
#
?
Jun 15, 2021 10:09
|
|
|
|
| # ? May 22, 2024 17:29 |
|
|
i get an "expensive" compound because it's $15/tube instead of $5/tube that lasts me years anyway and i have experience with it being the one i can leave on the cpu/gpu for over 5 years and it's not gonna get all dry and brittle and stop doing its job less than halfway through $10 is worth that to me e;fb
|
|
#
?
Jun 15, 2021 10:12
|
|
