|
Humerus posted:So it's been a few days of me fiddling with settings in BIOS and my CPU is just constantly running at max...until I found a post about Windows power modes and the High Performance setting makes your CPU run at 100% all the time. Turned that off and now it idles correctly. Ohhh... yeah I forgot to mention that possibility, simply because you seemed to imply the problem occurred only after overclocking. High Performance just locks the minimum state to 100% and stops steedstep from working.
|
#
?
Feb 10, 2019 14:21
|
|
|
|
| # ? May 15, 2024 04:16 |
|
|
HalloKitty posted:Ohhh... yeah I forgot to mention that possibility, simply because you seemed to imply the problem occurred only after overclocking. High Performance just locks the minimum state to 100% and stops steedstep from working. Yeah I'm an idiot. I don't know if my CPU downclocked before and I guess it probably didn't. Anyway thanks again.
|
|
#
?
Feb 10, 2019 14:39
|
|
|
Are hangs and restarts symptomatic of an unstable video card OC? I thought I had mine good (probably 15 hours in three different games with no issues) then out of the blue I got a random restart last night.
|
|
#
?
Feb 10, 2019 21:46
|
|
|
Red_Fred posted:Are hangs and restarts symptomatic of an unstable video card OC? I thought I had mine good (probably 15 hours in three different games with no issues) then out of the blue I got a random restart last night. Possibly, although it could be almost anything. Did you get a bluescreen or did it just reboot? You can look at bluescreen memory dumps to try to figure out what windows thinks caused the crash.
|
|
#
?
Feb 10, 2019 22:20
|
|
|
Red_Fred posted:Are hangs and restarts symptomatic of an unstable video card OC? I thought I had mine good (probably 15 hours in three different games with no issues) then out of the blue I got a random restart last night. If your PC runs P95 and Furmark fine for 5 mins, and also succeeds in 4 passes of memtest86 block move pattern it's highly unlikely the CPU/GPU/RAM is the culprit.
|
|
#
?
Feb 11, 2019 02:18
|
|
|
Just a straight reboot. I used to get them on my old system using this same video card but never really bothered to tweak my OC and just turned it off. I’ve done memtest and got no issues, will run P95 and Furmark.
|
|
#
?
Feb 11, 2019 03:11
|
|
|
Palladium posted:If your PC runs P95 and Furmark fine for 5 mins, and also succeeds in 4 passes of memtest86 block move pattern it's highly unlikely the CPU/GPU/RAM is the culprit. Those things run the hottest, but sometimes specific applications can tickle the right part of your hardware to make it fail. Sometimes furmark, etc doesn't suss out your instabilities, back off the overclock a little and see if it continues. Also, yes, software can be a problem.
|
|
#
?
Feb 11, 2019 03:46
|
|
|
I had a 9900K running @ 5Ghz on an EVGA 134-KS-E379-KR no problem but I seem to be struggling to get a 9600K to run @ 5Ghz stable on the same exact motherboard. Right now the problems seem to be manifesting themselves as the games I'm playing crashing instead of all of Windows crashing. I think I'm at 1.385v at this point, and I'm a little worried to go much higher. Anybody familiar with EVGA's BIOS able to give me some me a rundown of the settings I should be checking? I'm not at home right now otherwise I'd upload a picture of my settings or note the settings that I've made. edit: forgot a number in the voltage 
Aexo fucked around with this message at 22:41 on Feb 11, 2019 |
|
#
?
Feb 11, 2019 22:03
|
|
|
Aexo posted:I had a 9900K running @ 5Ghz on an EVGA 134-KS-E379-KR no problem but I seem to be struggling to get a 9600K to run @ 5Ghz stable on the same exact motherboard. Right now the problems seem to be manifesting themselves as the games I'm playing crashing instead of all of Windows crashing. I think I'm at 1.85v at this point, and I'm a little worried to go much higher. 1.85v RIP your chip. Your cpu may just not be able to do 5ghz
|
|
#
?
Feb 11, 2019 22:12
|
|
|
Just got a 9700k with a Gigabyte Aorus Z390 Elite. Is there a good guide on starting to OC? IDK if I wanna go super hardcore into it. I'd be pretty happy with 5.0ghz. Edit found one. IDK if my 9700k can actually hit 5ghz. Having trouble even running it at 4.5ghz stable and cool. Maybe it's the cooling I'm using. Cooler master AIO 240mm cooler. Might get a Noctua D15. E2M2 fucked around with this message at 16:48 on Feb 12, 2019 |
|
#
?
Feb 12, 2019 05:20
|
|
|
Man so far I've gotten stable results with my Cooler Master AIO at 4.4ghz, but I think its being undervolted at 1.14v. Low 70Cs on every core at max load on Prime95. Played BFV and it seems stable at this speed. Guessing if I'm gonna want to push any more that I'm gonna have to shell out for a DH-15. What is a safe temperature at max load? I think at 1.3v I was hitting 100C. edit: oh I guess I was running the newest version of Prime 95 so that mighta been why it was unstable and hot? E2M2 fucked around with this message at 00:51 on Feb 13, 2019 |
|
#
?
Feb 13, 2019 00:42
|
|
|
I hate to be the bearer of bad news but you're not even overclocking at this point. The Core i7-9700K has a 4.6 GHz all-core turbo speed out of the box and on a decent Z390 board like the Aorus Elite you should be hitting that without issue, especially if Gigabyte isn't limiting PL2 power or time as it seems not to be on many Z390 boards. It sounds like your heatsink has severe problems and you should probably make sure it's mounted properly and/or functioning at all before proceeding. TheJeffers fucked around with this message at 01:20 on Feb 13, 2019 |
|
#
?
Feb 13, 2019 01:04
|
|
|
Recommendations for TIM replacement for a 8700K? Would prefer non conductive solutions over liquid metal.
|
|
#
?
Feb 13, 2019 01:17
|
|
|
nescience posted:Recommendations for TIM replacement for a 8700K? Would prefer non conductive solutions over liquid metal. Just use liquid metal. There are no surface mount components on the 8700k so there is practically zero risk.
|
|
#
?
Feb 13, 2019 01:23
|
|
|
I did a delid + liquid metal for the first time on an i7-8086K recently and like Don Lapre says there is nothing under the heatspreader to short with it. If you apply a bead of RTV to re-bond the heat spreader to the substrate I highly doubt you would ever have any issues.
|
|
#
?
Feb 13, 2019 01:28
|
|
|
nescience posted:Recommendations for TIM replacement for a 8700K? Would prefer non conductive solutions over liquid metal. If you are talking delidding, just use liquid metal, it is fairly easy to do covering just the tiny die and the matching side of the heat spreader and can be done far from the rest of the computer where you don't have to worry as much about shorting something out. But if you are just talking above the heat spreader, use thermal grizzly kryonaut. Granted kryonaut will also frequently yield a ~10c reduction if you use it for the delidding replacement as well. I did thermal grizzly conductonaut for a TIM under the heat spreader, then kryonaut above it and got a 20C reduction in load temps on a water cooled 7700k.
|
|
#
?
Feb 13, 2019 01:29
|
|
|
This is for a on the go ITX build, I was just afraid the frequent jitter/bumps from traveling would somehow cause a leak for the LM 
|
|
#
?
Feb 13, 2019 01:36
|
|
|
nescience posted:This is for a on the go ITX build, I was just afraid the frequent jitter/bumps from traveling would somehow cause a leak for the LM It would be fine, especially if you use RTV to glue the IHS back down with the help of a proper delidding/relidding tool. Liquid metal has a really strong surface tension, and its also really sticky once you finally get it spread on the die.
|
|
#
?
Feb 13, 2019 01:55
|
|
|
TheJeffers posted:I hate to be the bearer of bad news but you're not even overclocking at this point. The Core i7-9700K has a 4.6 GHz all-core turbo speed out of the box and on a decent Z390 board like the Aorus Elite you should be hitting that without issue, especially if Gigabyte isn't limiting PL2 power or time as it seems not to be on many Z390 boards. Yea, I'm worried that the cooler just isn't any good? Could it not be contacting well or something? Using this thing: https://www.amazon.com/Cooler-Master-MasterLiquid-Chamber-MLA-D24M-A18PC-R1/dp/B07CRGC899/ref=sr_1_5?keywords=cooler+master+AIO&qid=1550021963&s=gateway&sr=8-5
|
|
#
?
Feb 13, 2019 02:41
|
|
|
E2M2 posted:Yea, I'm worried that the cooler just isn't any good? Could it not be contacting well or something? Just to clarify where are you getting your temp and voltage measurements from? Yes your cooler could be not contacting well. I wonder if you plugged the pump in actually? Maybe go back and make sure the pump is plugged in and the fans are spinning when it's on and that kind of thing.
|
|
#
?
Feb 13, 2019 03:09
|
|
|
VelociBacon posted:Just to clarify where are you getting your temp and voltage measurements from? Yes your cooler could be not contacting well. I wonder if you plugged the pump in actually? Maybe go back and make sure the pump is plugged in and the fans are spinning when it's on and that kind of thing. Using CoreTemp and CPU-Z to check Temps and vCore respectively. Right now using Prime95 26.6 I'm at 4.8ghz 1.248v, and nothing above 85C right now. I think it was maybe just the newer Prime95 which apparently is super taxing and unrealistic? Right now I'm tuning the v-core down right now to see what I can get 4.8ghz stable at.
|
|
#
?
Feb 13, 2019 03:27
|
|
|
E2M2 posted:Using CoreTemp and CPU-Z to check Temps and vCore respectively. Right now using Prime95 26.6 I'm at 4.8ghz 1.248v, and nothing above 85C right now. I think it was maybe just the newer Prime95 which apparently is super taxing and unrealistic? It has to do with AVX workloads (BF5 I think uses them too because it triggered an AVX offset I used to use). I'd push it to 4.9GHz and see what you need for that (voltage wise). I'd allow temps up to 100C for stress testing since you won't come anywhere close to that in normal workloads.
|
|
#
?
Feb 13, 2019 03:29
|
|
|
VelociBacon posted:It has to do with AVX workloads (BF5 I think uses them too because it triggered an AVX offset I used to use). Ah I was wondering about temperatures because everywhere I looked was a different number. poo poo, so I might have to do a AVX offset to run BF:V? edit: so I got cocky and tried to do 5.0ghz with the same settings I had for 4.9ghz and crashed, and I think BIOS reverted back to all factory settings, even back to before I flashed the newest update? Anyways I'm tuning 4.9ghz now, trying to see how low vcore can be stable. E2M2 fucked around with this message at 05:18 on Feb 13, 2019 |
|
#
?
Feb 13, 2019 03:34
|
|
|
E2M2 posted:Ah I was wondering about temperatures because everywhere I looked was a different number. This is the dumb problem I was having with the same mobo. I guess it's got two states and some crashes switch the state. It's the dumbest poo poo in the world and if I knew about it I probably would have bought another board. Just use the Favorites menu thing in the BIOS (I think Insert key adds things to favorites menu) to make it fast to adjust. Don't need to set up an AVX offset really unless you can't get your desired GHz without too much voltage/heat. I think you can still add voltage right? VelociBacon fucked around with this message at 09:41 on Feb 13, 2019 |
|
#
?
Feb 13, 2019 09:00
|
|
|
VelociBacon posted:This is the dumb problem I was having with the same mobo. I guess it's got two states and some crashes switch the state. It's the dumbest poo poo in the world and if I knew about it I probably would have bought another board. Just use the Favorites menu thing in the BIOS (I think Insert key adds things to favorites menu) to make it fast to adjust. Don't need to set up an AVX offset really unless you can't get your desired GHz without too much voltage/heat. I think you can still add voltage right? I lost all my settings, even saved profiles. It reverted back to Bios update 3, when I had 7D flashed on it. edit: I seem to have a stable OC at 4.9ghz @1.236v, we'll see how low it can go.
|
|
#
?
Feb 14, 2019 00:54
|
|
|
E2M2 posted:I lost all my settings, even saved profiles. It reverted back to Bios update 3, when I had 7D flashed on it. Yeah you have to update that BIOS separately. It's like having 2 motherboards and when you crash hard enough it switches to the other one. What are your temps with that?
|
|
#
?
Feb 14, 2019 01:12
|
|
|
VelociBacon posted:Yeah you have to update that BIOS separately. It's like having 2 motherboards and when you crash hard enough it switches to the other one. What are your temps with that? I think it just crashed before Prime95 was running for more than a couple of seconds so temps shouldn't have gotten up very high. Probably just way undervolted? Was stable at 4.9 for 20 mins though. It was a bluescreen.
|
|
#
?
Feb 14, 2019 01:36
|
|
|
Just a heads-up, the next one or two times you reboot it may switch back to the other BIOS. My half-assed 9700K 5ghz attempt ended up doing the same thing on an Aorus Pro WiFi board, where it switched to the backup BIOS which I updated and tweaked for overclocking, before it then decided to switch back to the non-backup BIOS but with default settings. I took it in stride as it least it prevented me from rendering everything unbootable.
|
|
#
?
Feb 14, 2019 06:57
|
|
|
Yeah after your next lockup make sure you check that it isn't in your other BIOS at some earlier point in the tuning where you had a way higher vcore or something.
|
|
#
?
Feb 14, 2019 15:27
|
|
|
A few questions for the experts I'm doing my first real CPU overclock on an i7-8700k in an ASUS Z390-E motherboard with 16 GB of DDR4-3200 RAM and using an EVO 212 as my cooler. I've gotten the CPU to stable with a multiplier of 49 (4.9 Ghz) on 1.28 mV of power with a -2 AVX offset and on LLC Level 6. Ran XTU for 20 minutes, RealBench for 15 minutes, Cinebench (multicore score was 1590) and 3dMark/Heaven Benchmark for CPU/GPU baselining with no crashes. Planning on running x264 overnight tonight to verify that the system is stable for all reasonable workloads. Going down to 1.27 mV gave me crashes with Cinebench. The problem is my max temp through all this were just a hair over 80 with one core at 82. I was able to get the system to boot and complete XTU without crashing at 1.35 mV on 5 Ghz but the heat was worrying me. I saw temps creep well into the upper 80s and I think one core even went above 90 and this was only running XTU which I understand is not that severe a stress test. 1) Does my performance seem reasonable for this set up? 2) I don't fully understand the power control stuff. I watched the below video and the guy on it seemed to say that LLC Level 6 is fine but that feels really high to me. Should I change my power settings further? 3) Should I try for 5 Ghz or is that not advisable given my temperature situation? 4) Any additional tests I should consider? Prime95 scares me.  https://www.youtube.com/watch?v=pbRauc8gpQM edit: Cinebench scores 
axeil fucked around with this message at 23:42 on Feb 19, 2019 |
|
#
?
Feb 19, 2019 23:40
|
|
|
Regarding voltage: use hwinfo64 and look for vcore and VID. The Vcore is what your CPU is getting and the difference between the VID and the Vcore is what is affected by the LLC calibration. VID is going to be quite similar to what you set in BIOS. We adjust the LLC calibration to be as near to flat as possible (VID=Vcore under load is ideal) so that we know what we set in BIOS is what the CPU receives. Those temps are absolutely not a problem. If you're not crashing absolutely add some clock speed but if you aren't using prime95 know that it might be stable only for awhile. You can always add voltage or reduce clock speed when you have crashes.
|
|
#
?
Feb 20, 2019 00:19
|
|
|
Your temps are about what I would expect from an 8700k overclock and a EVO 212. That CPU runs hot even on much larger coolers unless you delid it, and the 212 isn't exactly a top performing cooler; its mainly just good enough as long as you don't care about it running either hot or noisy.
|
|
#
?
Feb 20, 2019 00:48
|
|
|
VelociBacon posted:Regarding voltage: use hwinfo64 and look for vcore and VID. The Vcore is what your CPU is getting and the difference between the VID and the Vcore is what is affected by the LLC calibration. VID is going to be quite similar to what you set in BIOS. We adjust the LLC calibration to be as near to flat as possible (VID=Vcore under load is ideal) so that we know what we set in BIOS is what the CPU receives. Thanks so much for your help! This is what I'm seeing in HWiNFO64 (not under load), and I don't see VID reported, is it somewhere else?  And good to know that the temps aren't a problem. If I go up a notch to 5 Ghz, what's the max acceptable temp during stress testing? Opinions seem to really differ. Indiana_Krom posted:Your temps are about what I would expect from an 8700k overclock and a EVO 212. That CPU runs hot even on much larger coolers unless you delid it, and the 212 isn't exactly a top performing cooler; its mainly just good enough as long as you don't care about it running either hot or noisy. Thanks for the info. Do you think it's even worth trying for a 5 Ghz overclock or am I too likely to run into thermal issues?
|
|
#
?
Feb 20, 2019 01:21
|
|
|
VID is the voltage the CPU is requesting, Vcore is the actual voltage supplied by the mobo that goes into your CPU.
|
|
#
?
Feb 20, 2019 01:55
|
|
|
There is not really any harm in trying for 5 GHz, if your chip hits its thermal limits it will just throttle. If it runs 5 GHz but crashes after a while but only when it gets hot, it is possible upgraded cooling could stabilize it, but most likely you would have to delid it to really get a solid improvement even with a much larger cooler. Before I upgraded last year I was running a 7700k which is basically the same silicon less two cores, and I didn't do anything to it other than the asus multicore enhancement which takes it to 4.5 GHz all core turbo instead of just 1 core. I had it on a massive custom water cooling loop and with the stock thermal paste it hit the high 80s temperature and 120w load in Intel's extreme tuning utility's torture test. I decided I wanted it running cooler so I bought a delidding kit, popped the IHS off and replaced the paste with Thermal Grizzly Conductanaut. Same test, same clocks, same voltage after that reported high 60s temp and 115w load. So by lowering the load temps 20C, I reduced leakage current by ~5w which is a fairly significant savings. If you have a CPU that is just barely over the edge of its stability, it is possible upgraded cooling/delidding could pull it back, but in most cases it would take sub-ambient cooling (chillers/phase change, LN2, etc) to actually get it working past the normal limit. So basically just try it and see, if it doesn't work just back off and call it even at 4.9. The 8700k and newer CPUs just don't have that much clock headroom in the first place, so you aren't going to get massive performance improvements over stock like was possible with old CPUs that started at ~3 GHz and could run at 4.5+.
|
|
#
?
Feb 20, 2019 01:59
|
|
|
axeil posted:Thanks so much for your help! VID should be there somewhere but honestly it's fine just go by your Vcore so long as it's a reliable reading (can double check with cpu-z). Just don't change your LLC calibration after you're at the limits of acceptable temp or you might increase your vcore without meaning to. There seems maybe a fundamental error in your thinking about clock speeds and heat. Heat is relative only to the voltage you're applying. You could set your clock speed to 3GHz and if you didn't change your Vcore your temps would be the same. If you increase your clock speed to 5GHz it will not affect your temps, you might however get instability that is only remedied by increasing the Vcore, which in turn will increase temps.
|
|
#
?
Feb 20, 2019 02:03
|
|
|
VelociBacon posted:VID should be there somewhere but honestly it's fine just go by your Vcore so long as it's a reliable reading (can double check with cpu-z). Just don't change your LLC calibration after you're at the limits of acceptable temp or you might increase your vcore without meaning to. Higher clock speeds do cause chips to run hotter. A chip is doing 50% more work at 5 GHz than it is at 3 GHz even if the voltage is the same. Voltage generally has a bigger impact on temperature, but clock speed is very much there and not a trivial difference. Test it yourself, set a fixed vcore and change your multiplier, then watch your power consumption in torture tests at different clock speeds. Same reason the 9700k overclocked to the same speeds runs cooler and draws less power than 9900k, the 9900k is doing 20-30% more actual work thanks to HT. Power consumption/heat is significantly impacted by the amount of work the CPU is doing, and higher clock speed means the CPU can do more work.
|
|
#
?
Feb 20, 2019 02:29
|
|
|
VelociBacon posted:VID should be there somewhere but honestly it's fine just go by your Vcore so long as it's a reliable reading (can double check with cpu-z). Just don't change your LLC calibration after you're at the limits of acceptable temp or you might increase your vcore without meaning to. Ah, okay understood. I previously was working with 5 Ghz and had vcore set to 1.35 (anything less was unstable) but the temps were all 90+ when I was doing the XTU stress test so I'm concerned on two fronts: 1) temps will go even higher on more strenuous tests and 2) 1.35 might not be stable and an even higher vcore is going to generate even more heat. If I'm running a stress test and seeing peak temps over 90 that means I should dial it back, right?
|
|
#
?
Feb 20, 2019 03:21
|
|
|
XTU is probably the hottest stress test there is on an Intel CPU, its highly unlikely any real world workload will even reach 80% of that heat output. If it can handle running that, it should be fine running any other real application. The 8700k tjmax is 100C, so unless you get to 99+ its still within spec. Just keep in mind you may live longer at temperatures below 90C, but its still within the operational limits of the CPU. Indiana_Krom fucked around with this message at 03:33 on Feb 20, 2019 |
|
#
?
Feb 20, 2019 03:29
|
|
|
|
| # ? May 15, 2024 04:16 |
|
|
Indiana_Krom posted:XTU is probably the hottest stress test there is on an Intel CPU, its highly unlikely any real world workload will even reach 80% of that heat output. If it can handle running that, it should be fine running any other real application. Hotter than Prime 95? I thought Prime 95 was the "hardest" out there. What about max temperatures on more "real world" settings? 3DMark, Cinebench, playing an actual game, etc.? Should I dial it down if that's over 80? 85? From what I've seen googling around people talk about temperatures all the time but there isn't very good information about why people say, for example, anything over 80 is going to fry your chip/is bad. edit: ran the prime95 small FFT test for 5 minutes and these were my temps which were much, much higher than anything I've seen. No crash so that's good right? Assuming it passes an overnight x264 test I can declare the system stable @ 4.9 Ghz and 1.28 mV? 
axeil fucked around with this message at 04:23 on Feb 20, 2019 |
|
#
?
Feb 20, 2019 04:02
|
|