|
OK, seeing some odd behavior. I have a 8700k and a Noctua nh-D15 on an ASRock Extreme4 board. Temps have been very good, usually they don't go above 65C even under normal stress tests, but I have had clock speed issues. I've been trying to run a 4.7-5.0GHZ OC. I have had the voltage kinda high (1.4 or so) but I haven't been very stable lower than that and the temps haven't been much of a problem so I wasn't too worried about the higher voltage. The problem is that I can keep that speed with the CPUID stress test, but if I do Prime95 it drops to 3.3GHZ across all 6 cores, even though the temps are fine. I ran userbenchmark.com and scored below average for my CPU and it looks like the CPU just didn't ramp up: http://www.userbenchmark.com/UserRun/5626353 Am I hitting some thermal throttling? It does this while the max temp reads 64C, which should be plenty of headroom. The MB won't update the Bios, so is it a problem there or with the CPU? Or am I missing some other setting? It's otherwise been pretty stable.
|
#
¿
Nov 2, 2017 14:40
|
|
|
|
| # ¿ Apr 25, 2024 12:52 |
|
|
I was using HWMonitor Pro, which has been reliable for me in the past. I'll try realtemp tonight. I agree, it sounds like thermal throttling except for the readouts. I am tempted to turn off the throttling in the Bios and run it old-school for these tests, but I wanted to try some less riskier options first.
|
|
#
¿
Nov 2, 2017 17:14
|
|
|
Ok, figured it out. The extreme 4 Z730 has a MASSIVE default AVX offset, to the tune of 800-1.4GHZ MHZ clock drop. And it doesn't cleanly release it after the AVX test is done. Apparently setting it to 0 sets it to "Auto"(aka, 800MHZ+), and the lowest you can set it to is 100MHZ. That seems really frickin dumb and I just spent several hours chasing down potential power and thermal throttling.
|
|
#
¿
Nov 3, 2017 02:13
|
|
|
84C shouldn't be causing errors in your CPU, that is well within the operating temp. If you're seeing errors immediately when running smallfft's in prime95 I think you have an issue what might be worth digging into. Userbenchmark does a CPU test using AVX that is a lot less intensive, it might be worth trying that: http://www.userbenchmark.com Otherwise I think your voltage is a little low. I'd start setting it at 1.3volts and maybe bring your RAM down to 2.133 and run the Prime95 small FFT test. Then start bringing your ram up first, then your voltage down.
|
|
#
¿
Nov 4, 2017 15:02
|
|
|
eames posted:At stock clocks, yeah, because Intel bins for relatively large stability margins. (those are the margins we use for overclocking. You are conflating a few ideas. Heat generation is going to be a function of voltage and load. More heat doesn't mean more power, more power means more heat, because that energy has to go somewhere. Heat can cause instability, as can higher clock rates, which is why LN2 is how you get stable at insane clock rates. Voltage can help stabilize a CPU, but can create more heat, which is why you hit a ceiling that can only be fixed by better cooling. (This is oversimplified, but for what you're looking at here its a pretty decent model) The fact that you are having stability problems at relatively low temps (which 90C is relatively low for an 8000 series intel CPU) is telling me you have something else going on. I think you are probably keeping your voltage a little low, or else you are outrunning your memory. OR, you have a faulty CPU/MB/PSU which is why I'm suggesting testing now while stuff is under easy return windows. Prime95 is a power hug and not a "realistic" test, but more and more stuff will likely use AVX so if it is throwing up problems for you now it might be a browser in the future using parallelization or a game that causes instability in the future for you, or whatever problem that might exist might start showing up in non-AVX workloads.
|
|
#
¿
Nov 4, 2017 20:05
|
|
)
|
eames posted:I don't have time to respond to the whole post right now but please read this, tia. None of that data remains so it's hard to understand the methodology here. There will be a difference in power consumption, but I am shocked it would be that high. Was he controlling for load across all the tests? Because what your CPU is doing matters quite a bit. GRINDCORE MEGGIDO posted:EFB Of course, lower voltage means less current and wattage over the same activity. There is current leak at higher temps, but for the 8700K you should only see a difference of 15-25% from 60C-85C. My point to Eames is if you are seeing stability problems at or below 90C, it's not because you are butting up against thermal limits or even power draw limits (unless your PSU isn't up to snuff). AVX is going to draw more power, but that's not because of the heat, that is because of the workload too. I can show you an example of a 8700k running stable at 90C pretty easily.
|
|
#
¿
Nov 5, 2017 02:50
|
|
|
craig588 posted:Dead photobucket fix for Chrome http://i272.photobucket.com/albums/jj163/idontcare_photo_bucket/Intel%20Core%20i7-2600K/TempvsPowerfor2GHzat1290V.png Oh, yeah for the same clockspeed you are talking about 10% power consumption from very low-to-very-high. Note that that is not TDP. That much power consumption is probably within the regular normal variance for desktop OC'ers. That is a pretty cool article, but I still think something is wrong in your 8700k setup if you are having stability problems at 85C.
|
|
#
¿
Nov 5, 2017 03:45
|
|
|
AVX is for parallelization, I believe. That's different than multithread and not something that will be used universally. But it's used in some video and photo editing effects today, and we'll may see it in web browsers (which likely wouldn't tax a system) at some point in the future. Games probably won't use that instruction path in the short term but who knows?
|
|
#
¿
Nov 5, 2017 14:55
|
|
|
So I contacted ASRock about the inability to set the AVX offset to anything less than 100 MHZ. They replied back with a custom bios for me enabling that. Pretty cool response to what was essentially a request for benchmarking. I'm going to keep AVX offset on during my day-to-day OC, but it's nice that I have a bit more fine control over it.
|
|
#
¿
Nov 8, 2017 00:32
|
|
|
Or do a larger AVX offset. AVX instruction set stress tests are not going to lineup with what you want for everyday temp. Your numbers look correct for your situation.
|
|
#
¿
Nov 29, 2017 21:21
|
|
|
Zigmidge posted:I've been bouncing between AVX offset of 0 and 3 with no discernible effect during AVX activities/tests. Right now at 4.6ghz I've got it set to 0 but had it at 3 when pushing for higher clocks. What board are you using? ASRock treats 0 as "auto" which is usually 3. Zigmidge posted:El oh el that was a typo, sorry. 1.37. Oh, that's not as bad them. Better cooling is really your next step on this, I think you'll be able to run at 4.9 right around 1.3. I wouldn't stretch to 5 as your day-to-day OC if it means going that high, but I also think you can probably get stable at 5GHZ. How is your airflow in your case?
|
|
#
¿
Dec 4, 2017 21:52
|
|
|
Maybe its different now, but the last time I experienced the Corsair water coolers I was really turned off by the sound. Even if it was quieter (which I am not sure it was), the "whhrrr" of the pumps sounded different than the "whhooosh" of a large air cooler and was less appealing. That may be a preference though.
|
|
#
¿
Dec 5, 2017 23:24
|
|
|
Sync all cores at 4.7 and try again. 4.8 is starting to push a bit but most CPUs should be able to handle that. You might have a poor OCer. Alternatively, start at 1.35v and work down. You really can't hurt your CPU, thermal throttling will keep it well within parameters. Trying the "inch up" approach just means you'll have to deal with more crashes.
|
|
#
¿
Feb 26, 2018 16:21
|
|
|
1gnoirents posted:I didn't get into this earlier when it was mentioned to me but since it seems to be common knowledge now, whats wrong with P95? Especially relative to IBT? Its funny because in the past IBT was looked at with suspicion rather than P95. Afaik small FFT P95 and linpack are virtually identical workloads on memory and CPU, though they are solving different kinds of math, with IBT being clearly more stressful. Whats changed? I don't know about relative to IBT, but P95's AVX stresses ago well beyond real-word situations (today). Since AVX is a specific outlier that runs up heat I think its warranted that a special caveat is given when running tests on it. It's much more likely with the newer architecture that you can have a system that runs cool and stable on literally everything but running AVX full bore for 20 minutes straight. If you are trying to test for workstation-level stability, that isn't super realistic. Basically, you want a stability test that is outside normal working conditions, but the AVX testing is now so far outside the regular performance curve it makes sense to work around it.
|
|
#
¿
Feb 26, 2018 17:00
|
|
|
eames posted:Does anybody else have problems with the AVX offset kicking in non-AVX programs such as games? That is really weird. Why would it kick off the offset, but then not even adhere to the offset itself? AVX offset isn't a throttle (that I've seen), its usually just pretty simple logic that kicks off a different frequency. What board do you have?
|
|
#
¿
Feb 27, 2018 09:47
|
|
|
AVX is really aimed at parallelism, and it's not very widespread. Note that in typical use, AVX instructions aren't going to sit there processing for minutes on end. They will only occasionally be used. It's not like "This game that uses AVX will stress your CPU the same way Prime95 will". Web Browsing will sometimes use AVX instructions, but very minimally today. This may increase in the future. OpenSSL and some games do, but there honestly isn't much need for it and I don't expect it will be heavily utilized for a while. If you are running Blender you can see significant AVX instruction use, but again it'll only be a small percentage of your CPU cycles there. If your concerned about your heat or a CPU dip when benchmarking AVX, don't be.
|
|
#
¿
Mar 9, 2018 19:54
|
|
|
Revol posted:I was running 4.5ghz last night, but I came home today to my system unresponsive, so back to 4.0. FYI, my old super-star 2500k was rock solid at 4.4GHZ and needed TLC to get 4.5 and above. I think you can probably get to 4.2-4.4 stable, 4.5 is one of those barrier points on that CPU.
|
|
#
¿
Mar 11, 2018 00:30
|
|
|
jonathan posted:Between die and IHS I used arctic silver. I was reading that the biggest difference is closing up the tolerance between the die and IHS by removing the glue between the IHS and PCB. So I popped the IHS, cleaned all the sealer/glue off, already a thin layer of arctic silver over the die, then reglued the IHS using the weakest JB weld. Then used the motherboard CPU retainer to keep the IHS clamped. I only glued down the 2 side tabs on the IHS. Not the entire perimeter. Let's back up a second. What load tests are you doing? AVX (prime95 small fft) with no offset? Or like, the CPUID stress test? What happens if you run a game or something for 20 minutes, what temp ranges does it stay in? What voltage are you running at? The speed of a spike is not an indication that something is wrong. What is concerning is that your hitting the thermal ceiling doing tests, but without knowing the tests and conditions its hard to say if its really wrong. I could make a perfectly water cooled system hit the thermal limit if I really wanted to.
|
|
#
¿
Mar 13, 2018 20:19
|
|
|
jonathan posted:Sorry, that was a 5am post after a lovely work night. I'm using cpu-z's stress test. I was able to repeat it with Asus's CPU burner test within the tweak GPU utility. Then I swapped motherboards and now the core temps are 80c using cpu-z's stress test. That is really high, what voltage are you running the OC at? Otherwise to me this sounds a lot like bad thermal contact or maybe an issue with the heatsink itself. I would be very surprised if damage would cause thermal behavior like that.
|
|
#
¿
Mar 14, 2018 01:04
|
|
|
jonathan posted:Yeah I guess I'll order in some liquid metal and redo the die to IHS. What's the best for IHS to cooler ? In my experience its usually not very significant what brand you use, though I have seen batches that were bad. Arctic Silver 5, or Noctua NT-H1 are both fine. I have heard this stuff is good though I've never used it. https://smile.amazon.com/Thermal-Gr...a-no-redirect=1
|
|
#
¿
Mar 14, 2018 13:51
|
|
|
Minorkos posted:I'm afraid I don't know the terminology yet so I don't really know what moving the multiplier up. Does that simply mean moving up a setting (CPU ratio?) without tinkering with other settings and seeing if my computer starts with it? Set the CPU multiplier to 42x and the VCore to 1.28V. That should be pretty stable and shouldn't generate too much heat. Do some testing with some CPU Burn-in applications to make sure you're stable. Play with DRAM OCing next, but if nothing is working your RAM just might not be very overclockable, mine wasn't. That should give you a nice boost pretty easily. I assume by "Rat Game" you are playing this, so every CPU cycle helps https://www.youtube.com/watch?v=YnO406cOVmM
|
|
#
¿
Mar 16, 2018 21:32
|
|
|
Minorkos posted:I did what you said and I booted up A-ok. Is the matter of testing for stability as easy as running CPU-intensive software and seeing if I blue screen or not? I ran CPU Burn-in for a few minutes and it only took up 25% of my total CPU in Windows, and then stopped with no errors. I also ran the stress test on CPU-Z and it didn't do anything especially scary either, besides slowing down everything and occasionally forcing Core Temp to reload its window. My temperatures seemed to cap out at 66C after running at 100% for a while. Congrats, you just got a cpu ~33% faster and its running reasonably cool. If you get any weird crashes you might want to consider stepping the voltage to 1.3, but I think your in a good spot. Trying to get much higher than where you are will require more fiddling and won't result in much performance improvement, so I think your probably best just to leave as is. I'm a Hard Over Clocking Professional Gamer Dood and I used the 2500k until about 5 months ago. This should buy you significantly more time if you were doing ok at 3.3GHZ.
|
|
#
¿
Mar 17, 2018 18:04
|
|
|
I suspect there was something wrong with the application or just a bad batch of Arctic Silver. Liquid metal will be better, but that spread is WAY more than you'd normally expect.
|
|
#
¿
Mar 27, 2018 03:09
|
|
|
Eletriarnation posted:It was several years ago but HardOCP did a test and determined that American cheese was only about 6C worse under load than their best thermal grease, so I'd be willing to wager that the vast majority of thermal greases are going to perform similarly. There is definitely such a thing as bad batches though. So while the brand doesn't matter, if you see poor results from one tube it might be improved by getting another one.
|
|
#
¿
Mar 28, 2018 15:51
|
|
|
jonathan posted:My 7600k @ 5ghz is score a lot higher than an 8700k on cpu-z benchmark for single core. Does this make sense ? It scored 608 while the 8700k is high 400's. If it's a stock 8700k this makes sense. If they are both at 5GHZ, I am guessing the 8700k is getting throttled somewhere. The best test would be to just set 1 core to 5GHZ to give an apples to apples. Also, even if everything is fine I'd expect the two probably within margin of error of each other in single core. If the 7600k is just barely edging it out I wouldn't be shocked. Edit: High 400s is stock 8700k, I think it throttled to stock. 8700k @ 5GHZ will be closer to 600 http://valid.x86.fr/bench/a5wyah https://valid.x86.fr/lmdr5k Lockback fucked around with this message at 16:12 on Mar 29, 2018 |
|
#
¿
Mar 29, 2018 16:10
|
|
|
jonathan posted:Yeah sorry I meant my 7600k @ 5ghz is benchmarking quicker than the reported stock 8700k. I thought the single core performance was a big bump from the 7600k but I guess not. They increased the single core stock speed, but not dramatically (about 100MHZ across similar levels). It is a big jump in stock speeds from the 7600k to the 8700k, but OCing covers that ground.
|
|
#
¿
Mar 30, 2018 02:08
|
|
|
jonathan posted:So in the few recent generations, the single core gains have been in clock speed stability rather than doing more at a given frequency. Gotcha. Some other features too that won't show up in all benchmarks, but yeah. The big performance jump is in number of cores.
|
|
#
¿
Mar 30, 2018 13:29
|
|
|
Zero VGS posted:Hmm thanks, I changed the AVX from Auto to 0 and now it is using the correct frequency on Prime95 for a minute or two (instead of dropping by 100mhz instantly), but after that it lowers the frequency down to things like 4700, 4701, 4559, 4699, etc. It's strange that it seems to be aiming for 100mhz increments still, and takes them down even further now, but only after it's been grinding Prime95 for a few minutes. It's thermal throttling. AVX benchmarks are not a realistic benchmark (even for stress testing) for most user activities. I would set your AVX offset to 3 and your other stability tests seem to indicate you are in good shape. The AVX offset won't hurt your performance in any meaningful way unless you are doing some heavy duty video effects editing. AVX is used in some web-browser activity but extremely lightly (so the offset won't be used for very long). Some games might use AVX in the future, but again I would expect the usage would be a very small % of the CPU instruction set.
|
|
#
¿
Jun 27, 2018 14:58
|
|
|
eames posted:Have you tried this? I found that even letting Firefox minimized in the background will reduce my 8700K CPU performance in non-AVX benchmarks by the amount that the offset is set to. I haven't seen this myself, but I don't use firefox which I think has the most AVX-compiled features. Alternatively, as you mentioned, you can just leave the offset to 0 (or the minimum of 1 which some ASRock boards don't let you go below for some reason) and make sure your thermal throttling is sane.
|
|
#
¿
Jun 27, 2018 21:33
|
|
|
Zero VGS posted:Ultimately I decided that even though 5000mhz is a nice round number, 4900 is a billion times easier to manage; I can drop voltage by 0.1 and keep AVX at -2 and it is stable on everything including 15 minutes of RealBench, with a lot less power and heat. This is basically exactly where I landed. I was able to get it stable at 5GHZ, but I found keeping an "Everyday" OC at all cores 4.8 was way cooler and used way less power.
|
|
#
¿
Jun 28, 2018 14:36
|
|
|
Dumb question but are you sure your CPU cooler is seated properly, with the right amount of thermal paste, all that fun stuff? If you are getting overheating issues at stock voltages (unless I am misunderstanding) there might be an easy physical fix.
|
|
#
¿
Aug 9, 2018 14:19
|
|
|
Lolcano Eruption posted:Talk about horrible luck. I would expect the mobo to be dead, but why the PSU? I assume the brownout killed the PSU unrelated to the BIOS update. I suspect you can recover the Motherboard though. I've had bad luck/incredibly-stupid-on-my-part stuff happen and you can almost always recover a corrupt flash.
|
|
#
¿
Aug 13, 2018 19:03
|
|
|
Endymion FRS MK1 posted:Ok, thanks guys. I think I'll still set a power limit of like 250W in the bios too I think that's an abundance of caution but also probably won't hurt you. Unless you have a really odd-ball use case of professional video rendering you won't be continually running AVX instructions for a prolonged period of time like it does in the benchmark. Plus your PSU already passed the test, there's no guarantee even a 750 PSU would pass it.
|
|
#
¿
Aug 27, 2018 15:25
|
|
|
Endymion FRS MK1 posted:Nope no rendering, it's purely a gaming PC. Do any modern games use AVX? I vaguely recall reading the PS3 emulator did Web Browsers can for certain things and probably at some point some games will use the instruction set more, but even then the amount of time the CPU spends in that instruction set is going to be very, very low for anyone not doing specialized rendering or video converting. Setting an AVX offset is the best solution here (since the AVX instruction set literally exists to perform the certain FP calculations faster). It's a PIA for benchmarking right now since it's realistically the most intensive thing your CPU can do, but running AVX instructions for hours at a time is not a realistic scenario by any stretch. If a game uses AVX for something, it's not like it's going to use those instructions for the entire time. I mean, if your concerned setting a more conservative power throttle is fine, but I don't think you are actually in any sort of real risk.
|
|
#
¿
Aug 27, 2018 16:58
|
|
|
Endymion FRS MK1 posted:I am extremely impressed by this cooler. I had the chip delidded by Silicon Lottery and during all of this stress testing temps never rose past 80C, even on the max heat Small FFT torture test... but the best part is the fact that I never heard the cooler unless I actively listened for it. I mean no its not like 0db or something stupid, just really smooth and quiet. Even with my case open and laying beside me (Weird circumstances caused this temporary setup) I had to put my hand behind the exhaust or the cooler itself to actually verify I was blowing hot air and not just on idle. That seems pretty good to me, especially @ 1.3v. I suspect due to the nature of confirmation bias you get told it sucks if it's not top 10%. What are they saying the 8086k usually hits?
|
|
#
¿
Aug 30, 2018 16:32
|
|
|
spasticColon posted:Well my video card is an EVGA GTX1070 SC so would EVGA precision be a better choice? Doesn't matter. Use whichever you like functionally they are identical.
|
|
#
¿
Sep 4, 2018 04:44
|
|
|
fknlo posted:I've decided to finally start playing around with overclocking since I have a new build on its way. I'm currently on a 4690k on a Z97-AR. I used the preset overclock in the BIOS and it's running at 4.49Ghz at 1.25V. Stability seems great and temps are good as well. Leave it there for now or play with it when I have the time to learn some things? That's pretty good for a 4690k. If you want to start messing with it I'd put money on hitting 4.7, but you'd probably be running warm at that point. I think 4.8 was basically the realistic ceiling there for most people but probably not worth getting to. If 4.5 was that easy on a relatively low voltage that would probably be where I'd leave it for "day-to-day".
|
|
#
¿
Nov 26, 2018 17:53
|
|
|
Yeah, my impression on the 9900k was there wasn't a ton left on the table and the stock turbos are already pretty good. I haven't had much experience with the 9900ks though.
|
|
#
¿
Nov 30, 2018 16:53
|
|
|
What is your AVX offset set to? You said it was happening both with and without AVX tests, but if that is the offset maybe your "non-AVX" test isn't really non-AVX.
|
|
#
¿
Dec 3, 2018 20:53
|
|
|
|
| # ¿ Apr 25, 2024 12:52 |
|
|
Red_Fred posted:Any good guides around for overclocking an i5-9600k? I just built a new system the other day and I’m keen to get this thing as close to 5 GHz as possible. It's hard to screw it up. Set voltage at 1.3v to start or use voltage offset (I wouldn't use auto). Lock frequency on all cores and start at like 4.7 or 4.8 and stress test non avx to watch temps and stability. Go up from there. If you crash increase voltage. 9600k should be totally safe for a casual overclock to 1.4 but for everyday use I'd probably try to keep it around 1.35 or so, it'll depend on your cooling and use and such. AVX tests are hard stress, and not a realistic load (you'll use avx instructions but not even close to at the levels a stress test will do). I'd say do an AVX test but don't overreact to heat levels. You won't break it.
|
|
#
¿
Dec 28, 2018 15:58
|
|