|
Does Intel Turboboost need to be enabled for overclocking settings to take effect? I swear I had tested/verified and had overclocking working, but then I just increased the multiplier and now while running Intel Burn Test I never get above stock clock multiplier (35x in my case). i5-4690K on an ASRock Z97M Pro4 mobo.
|
#
?
Feb 25, 2015 19:19
|
|
|
|
| # ? Jun 14, 2024 12:15 |
|
|
So on December 1st I fixed my first overclock on a G3258. I only fiddled with 2 settings but that seemed to be a good solid OC. I had my clock set to 40, for a 4.0GHZ OC total, with my core volt at 1.17 and my Ring volt at 1.16. This sat stable, besides one blue screen, up until today when I had 3 BSOD in a row. It looks like the CPU isn't getting enough power, so I knocked the voltage on the core up to 1.19, and that seems to be stable now, with minimal heat increases (76 peaks in Prime95) is there a reason an overclock will become unstable all of a sudden like this?
|
|
#
?
Mar 3, 2015 23:59
|
|
|
Nemesis Of Moles posted:So on December 1st I fixed my first overclock on a G3258. I only fiddled with 2 settings but that seemed to be a good solid OC. I had my clock set to 40, for a 4.0GHZ OC total, with my core volt at 1.17 and my Ring volt at 1.16. This sat stable, besides one blue screen, up until today when I had 3 BSOD in a row. Short answer: Electromigration. Long answer: Go wiki electromigration. Abridged answer: Every processor will eventually erode to uselessness because integrated circuits are small enough that the continual energization and de-energization of their electrically useful components can knock atoms out of their proper places, and when circuits are relatively few atoms wide (read: dozens to hundreds), it doesn't take much of this misalignment to damage the pathway. Keep in mind that a processor depends on more or less every single pathway being to spec, and that there's several billion of them in a modern processor. The denser the processor, the smaller the manufacturing process, the tighter the tolerance, the lower the electromigration endurance, and the shorter its useful life. This is why it's so hard to push processors to smaller processes these days, because we're running into fundamental physical limits imposed by the size of the atoms in the processor's material components. A Super Nintendo may last until the stars go out. Your G3258 may only last a decade or two even if you never overclocked it, and a 4790K would be closer to the one decade under the same circumstances.
|
|
#
?
Mar 4, 2015 00:37
|
|
|
How long did you test the previous OC? It's possible that it was on the edge of instability and just never quite reached it during your tests. Degradation over time causing OCs that were once stable to no longer be is also a thing, but that seems unlikely to be the case here considering the time involved. Also, 1.19V is quite low for an OCing voltage. Any reason why you haven't just cranked it up? Just about anything less than 1.3V is basically risk-free. GokieKS fucked around with this message at 00:41 on Mar 4, 2015 |
|
#
?
Mar 4, 2015 00:38
|
|
|
I agree, 1.19 is low (virtually stock if not something stock can reach on its own). I wouldn't worry in the least until you're way over 1.30
|
|
#
?
Mar 4, 2015 01:10
|
|
|
I know about Electromigration but I figured seeing as the CPU is brand new and the volts so low, the chances of it being that are miniscule. I could be wrong though. I guess the volts being low was the issue, I was kinda worried about Electromigration a lot as well as heat (this CPU seems to run a bit hotter than most), so I have it set as low as I could. I'll see if this is more stable and if not, I'll knock it up to 1.2. Thanks guys!
|
|
#
?
Mar 4, 2015 01:53
|
|
|
Prime95 is not indicative of load temperatures unless you use applications which use AVX instructions (which will cause the CPU to automatically increase voltage regardless of what it's normally setting it to), if that's what you mean by "runs hot". Unless your ambient temps or case ventilation are completely awful, even the stock HSF should be capable of keeping an OCed CPU at acceptable temperatures. Mine that ran 4.6 @ 1.275V was fine with the stock HSF when my CM Glacer 240L started developing pump noises.
|
|
#
?
Mar 4, 2015 01:57
|
|
|
Yeah electromigration is a thing but it takes years and high voltages, if you upgrade your cpu like every 2 years don't even worry about it. lovely motherboards do degrade after OC tho, not sure what's the explanation behind it but I've seen it a few times. In other news my computer is randomly booting with either 12/16/20 gb of ram, this rampage 4 extreme is really lovely about ram.
|
|
#
?
Mar 4, 2015 03:11
|
|
|
GokieKS posted:Prime95 is not indicative of load temperatures unless you use applications which use AVX instructions (which will cause the CPU to automatically increase voltage regardless of what it's normally setting it to), if that's what you mean by "runs hot". Unless your ambient temps or case ventilation are completely awful, even the stock HSF should be capable of keeping an OCed CPU at acceptable temperatures. Mine that ran 4.6 @ 1.275V was fine with the stock HSF when my CM Glacer 240L started developing pump noises. I get 75c peaks at anything below 1.2v but as soon as I knock it over that I get issues with heat. Does ring volt have a lot of effect on heat? I've not quite got what that does yet. Also, no blue screen since I upped the volts, so I guess that was it. Weird, because I stressed the poo poo out of it in December and no issues and just now was blue screening over netflix and skype.
|
|
#
?
Mar 4, 2015 04:50
|
|
|
Nemesis Of Moles posted:I get 75c peaks at anything below 1.2v but as soon as I knock it over that I get issues with heat. I pretty much promise you were just barely on the edge
|
|
#
?
Mar 4, 2015 06:16
|
|
|
Fair enough. Hopefully that little nudge puts me in the clear. I might bump it up some more and shoot for 4.5 later this week.
|
|
#
?
Mar 4, 2015 06:30
|
|
|
I keep my voltage in the bare minimum because I believe (but I don't actually KNOW this subject, it's mostly faith) that it will reduce the decay of the processor and bump my voltage up 0.03v every 6 months to a year or so when it blue screen if I play a more demanding game. E: But I also remove the automatic clock boosts so my processor is most of the time at 40x 4.0 GHz and this probably causes more damage overall, I'm stupid. program666 fucked around with this message at 13:56 on Mar 4, 2015 |
|
#
?
Mar 4, 2015 13:53
|
|
|
I'm currently trying to overclock my 3570k on a p8z77-i motherboard and it isn't going so well. I set the Turbo ratio to 42 and I'm still crashing at 1.275v. Am I doing something wrong? Should I keep adding more voltage? edit: I tried the auto tuning program by ASUS and it set my computer to a 41 ratio and 103.00 blck. Upon restart the computer crashed once into windows. Apparently I'm have some sort of problem overclocking. No idea why. John Lightning fucked around with this message at 05:17 on Mar 8, 2015 |
|
#
?
Mar 8, 2015 00:04
|
|
|
My processor already runs perfectly at 4.0GHz at 1.163V but I was under the impression that I could make it reduce voltage and/or speed when idle, and Linus from Linus tech tips say that you should set your machine to "adaptive" voltage after the overclock is done. Problem is if I just set it to Adaptive and leave everything else on auto it will increase voltages too much and produce more heat than needed, same thing if I try to set "Additional Turbo Mode CPU Core Voltage" to 1.163V and offset to 0.001V instead of leaving them on "auto". The voltage will go above 1.19 regardless. So I came to the conclusion that the best way would be to force everything the way it was. I kind of wanted a mobo that would vary clock speeds between 3.0 and 4.0 with a max voltage of 1.163V but this just seem not possible. Am I doing something wrong? program666 fucked around with this message at 23:56 on Mar 9, 2015 |
|
#
?
Mar 9, 2015 23:50
|
|
|
Use a negative offset.
|
|
#
?
Mar 10, 2015 00:38
|
|
|
John Lightning posted:I'm currently trying to overclock my 3570k on a p8z77-i motherboard and it isn't going so well. I set the Turbo ratio to 42 and I'm still crashing at 1.275v. Am I doing something wrong? Should I keep adding more voltage? I've got the same CPU and a similar MB from Gigabyte, follow this blokes guide, it did me well: https://www.youtube.com/watch?v=mBEeXajbG2o I'm pulling 4.3 GHZ on 1.115v, I could push it higher but I'm liking have my voltage so low. Also dont change your blck, it will make it alot easier to crash your system for almost no gain over just changing the multiplayers
|
|
#
?
Mar 10, 2015 10:59
|
|
|
craig588 posted:Use a negative offset. but even if the mobo obeys the limits I impose to it, I imagine it will keep the voltage at 1.163 at all times anyway since it thinks it's too little voltage for any clock speeds right?
|
|
#
?
Mar 10, 2015 12:58
|
|
|
It's overriding what the CPU is requesting. If the CPU thinks it needs 1.163v and you have a -.005v offset the motherboard will give it 1.158v. Those numbers probably will never line up so perfectly because of the intended use of vdroop to control thermal limits depending on different load requirements, if it's requesting 1.163, but you're also doing something that needs 150 watts it'll drop the voltage automatically regardless of what the CPU is requesting, and normally the clock speeds would go along with it, but pretty much every overclocking capable motherboard will ignore power limits for clock speeds. Basically, whatever the CPU requests is going to be offset by -.005 and everything else will move around with that new -.005v adjusted voltage as a target. If you want less voltage increase the offset, I think I'm using -.02v right now.
|
|
#
?
Mar 10, 2015 20:31
|
|
|
I think I finally understood how this works thanks to your post and some fiddling. The voltages without offset are: -1.192 when idle -1.284 when under load I could use an offset of -0.121, so it would hit 1.163 under load at 40x, the value I understand is the ideal one for the 40x multiplier, the problem is my machine won't even boot with the idle voltage at this offset. at -0.114 it won't go past the splash screen, at -0.100 it will bluescreen before windows' login. Another thing that I was looking for and I kind of understand there is no way to set is to make the clock vary between something like 3.4 and 4.0 GHz, I can only afix it at 4.0 GHz or leave it at stock settings so it will vary between 3.4 and 3.8. But in the end, why people suggest that you use adaptive? Isn't it better to just leave it at the bare minimum amount of voltage? I'm guessing that I could let it fluctuate between let's say 1.120 and 1.212, so I could benefit from the power saving at 1.120V when idle, but it's kind of funny that I can't set the maximum to 1.163V. E: no, I can't let it vary between 1.120V and 1.212, it will bluescreen almost immediately under prime95 lol, I don't get adaptive. program666 fucked around with this message at 00:44 on Mar 11, 2015 |
|
#
?
Mar 11, 2015 00:36
|
|
|
Is there a watercooling thread ? Got the EK waterblocks kit for the Rampage 4 and I need an intel 1150 waterblock, while I'm there I need a new pump and rad and I'm looking for advice and used parts cause my waterpump exploded and I'm pretty sure it sent something in my rad because it doesn't flow well anymore. Pricing a decent setup on EK website is like 400$ for a 360 rad with D5 pump and accessories, those fittings are loving expensive and I'm not sure I want to blow that much on cooling, I'd rather get a GPU for that price but I also want to overlock a bit and i can't trust my jury rigged adapter I've been using after the pump fiasco, pic for giggles : This is an evo 212 without mounting bracket, used stuff like pci plate, washers, amd bracket and rubber bands  
|
|
#
?
Mar 24, 2015 20:04
|
|
|
Out of curiosity, are you doing that more for fun or are you actually trying to get more speed out of it? I hate to be a buzzkill but delidding and a plain jane AIO brought down my temps so far that they were no longer the bottleneck.
|
|
#
?
Mar 24, 2015 20:20
|
|
|
Yeah I know, couldn't care less about watercooling but having all this hardware unused is just taunting me and I really enjoyed the silence, also bought this board for overclocking but had so many problems that I didn't try it yet, my CPU is a 3930K I bought used and it's running pretty hot already at stock speed, might be defective or I've been too used to my AMD's running ice cold
|
|
#
?
Mar 24, 2015 20:28
|
|
|
I have an i7-3820 with a Corsair Hydro H60 on it. It performs lovely at a 38x multiplier (stock is 36x), but I've been thinking of going for more. Should I just bump the multiplier or should I follow a guide for BCLK multiplier overclocking? I've seen one for pushing it up to a stable 4.625 GHz and it seems like it could be fun...
|
|
#
?
Mar 24, 2015 22:44
|
|
|
Think I found my temperature problem but not sure, Ai suite put my vcore at 1.325V with stock speeds and I believe it is not normal but any time I try to change the vcore in the software it just freeze the computer.
|
|
#
?
Mar 25, 2015 03:22
|
|
|
Don't use the AI suite or adaptive in the BIOS, instead use Manual.
|
|
#
?
Mar 25, 2015 04:14
|
|
|
I have no idea what's going on, this is with the Vcore set as 1.2V manual, HWinfo stills shows me a VID of over 1.3V, what is happening ?
|
|
#
?
Mar 25, 2015 12:35
|
|
|
Sh4 posted:I have no idea what's going on, this is with the Vcore set as 1.2V manual, HWinfo stills shows me a VID of over 1.3V, what is happening ? PNG now comes in 24/32 bits. Also Vcore is what you care about. VID is Intel's specification for nominal voltage-in. That is: this is the voltage before droop at which Intel considers the processor Intel-stable for the given combination of core clock rates and other circumstances.
|
|
#
?
Mar 25, 2015 12:51
|
|
|
Sir Unimaginative posted:PNG now comes in 24/32 bits. Is it before Voffset too ? edit: nevermind, figured it's AI Suite loving everything up and pushing all the settings to "extreme" at boot overriding what I set in BIOS, that is some bullshit, especially since I just want to use that software for the neat custom fan curves. Sh4 fucked around with this message at 17:51 on Mar 25, 2015 |
|
#
?
Mar 25, 2015 17:25
|
|
|
Sh4 posted:Is it before Voffset too ? In-environment overclocking tools are almost uniformly a bad idea. Intel has some official ones (which don't necessarily work on your motherboard if its manufacturer would rather you use their tools and/or are just dicks) and even those should probably be reserved for zeroing in on an overclock before affixing it in firmware.
|
|
#
?
Mar 25, 2015 19:08
|
|
|
Removed Ai suite (which is a pain in the rear end since it installs services and scheduled tasks everywhere), tried a few things for OC, I was aiming for 4.2Ghz and had unstability even at high voltage offsets, tried the asus presets and they were garbage, basically cranking up voltage everywhere and it was unstable as hell since it changed the BCLK (I believe it's the equivalent of FSB ?) to 150+ Mhz. The only way to have stability was to use fixed Vcore, I would like to keep dynamic voltage tho, is there anything I'm missing ? edit: pic 
Sh4 fucked around with this message at 22:26 on Mar 25, 2015 |
|
#
?
Mar 25, 2015 22:24
|
|
|
I think I'm hitting a wall on my G3258, but its much lower than I'd expect and there's some funky issues I would like some advice on. For reference, I'm still on stock cooling, and my motherboard is the MSI z97 PC Mate. My current 'totally fine completely stable and everything is a nice temp' settings are a multiplier of 41 with a CPU Core voltage of 1.195. Those are the only things in the BIOs I've messed with. That gets me sub 70~ temps even on high load and with Prime95 running on Blend. I wanted to bump that up a little today, after reading that even poo poo-tier chips can reach 4.3ghz. So I set the multiplier to 42 and started bumping up the voltage. It took till 1.25v to get anything near stable (boot to windows, short stress test) but the temps were spiking at 85~ and generally hanging around the high 70s. Trying to bump up to 43 (just as an experiment at that point) took 1.27, and shot my temps up to the mid 80s at peaks, with 60~ at idle. Anything above that and my PC refuses to even boot. Is this just lovely cooling? Should I upgrade that and then shoot for a better OC? I'm a bit disappointed after hearing tales of completely stock chips hitting 4.5ghz and 70~ temps. Maybe I just got a dud. Are there any other settings I should be tweaking to get better results?
|
|
#
?
Mar 29, 2015 20:58
|
|
|
Currently I'm running:
I'm running Prime95 balanced test. Total package power is about 85-88W, and I hit 83-85C after a couple minutes (at which point I've been killing it). It then returns to 45C within a couple seconds and idles down to 35C. I've never tested using Prime95 before - is it normal for it to heat processors up that hard? Or should I use another stress tester? Am I already overclocked a bit just from the XMP memory profile? I realize a single 140mm cooler isn't exactly a powerhouse but I was hoping I could overclock at least a little bit, and I don't feel comfortable with those temps. I've got a couple things I'm wondering about. First is whether I hosed up putting the paste on. Worth taking it off and trying again? Also, my fan configuration. I've got 2 140mm slots in the front, a 140mm on the back, and a 140mm on top. Currently the CPU cooler is on the front blowing in, with a 120mm fan below that blowing in, and the rear/top fans pushing out. I wanted the front coolers to blow in since there's a dust screen there, and it's too thick to fit on the top mount with both fans. Should I try rearranging that? Finally, I'm thinking the radiator fans might not be the best in the world. Would I be better off getting some aftermarket high-static-pressure fans?
|
|
#
?
Apr 3, 2015 02:39
|
|
|
I've got a similar question, I've got a new 4970k running at stock clocks, temps are good for absolutely everything except Prime95 small FFTs, where it goes to ~95C, it is stable at this temperature and does not seem to throttle itself down. Short test:  End of longer test:  I've got a Noctua U12S on it and am wondering if it's fitted tight enough/needs to be re-seated... I'm mostly concerned because of how rapidly the temperature ramps up and down, the load temperature is a delta of ~55C over idle which doesn't seem too unthinkable but it gets there instantly. So, worth openning things up and reseating cooler?
|
|
#
?
Apr 5, 2015 00:06
|
|
|
Small fft will always do that
|
|
#
?
Apr 6, 2015 02:14
|
|
|
Looked at my S340 further - I think the configuration I had was optimal (top front radiator sandwiched with intake, lower front fan intake, top fan exhausting, rear fan exhausting). The S340 can't do a radiator on top due to clearance issues with the RAM - the only way it'd fit is with the radiator against the case with a fan outside, which I don't think will have enough oomph in either exhaust or especially intake. The rear slot is 120mm which won't fit my 140mm radiator. So - the only real choice is whether to intake or exhaust on the front, and I still think intake is a better idea. GPU temps aren't a problem right now. I benched using AIDA64 instead and I'm getting 100% load temps around 60C so I think it's just Prime95 being a hellish stress test. Paul MaudDib fucked around with this message at 03:47 on Apr 6, 2015 |
|
#
?
Apr 6, 2015 03:38
|
|
|
Paul MaudDib posted:Looked at my S340 further - I think the configuration I had was optimal (top front radiator sandwiched with intake, lower front fan intake, top fan exhausting, rear fan exhausting). The S340 can't do a radiator on top due to clearance issues with the RAM - the only way it'd fit is with the radiator against the case with a fan outside, which I don't think will have enough oomph in either exhaust or especially intake. The rear slot is 120mm which won't fit my 140mm radiator. So - the only real choice is whether to intake or exhaust on the front, and I still think intake is a better idea. GPU temps aren't a problem right now. Prime95 is not real world at all, especially small FFT.
|
|
#
?
Apr 6, 2015 04:59
|
|
|
My friend gifted me his liquid cooler as he recently upgraded, and I decided 'hey, why not see if I can squeeze a little more life out of my old rear end x2 255?' (3.1ghz stock) . I haven't OC'ed since the 90's when I got my old pentium 75 to nearly 100mhz and I'm happy jumpers are no longer needed. To make a long story short my multiplier is locked so I got the FSB up to somewhere around 233-240, can't remember off hand, but with no other changes it's running stable at 3.7. I booted fine with 3.8 but it crashes on the prime95 blend test within seconds. Small FFT ran more stable so I thought maybe it was a ram issue, but as soon as I bumped the ram voltage up by 0.0675 (whichever the smallest increment is) it spiralled into an unrecoverable reboot cycle and I had to clear the cmos. The system started throwing out bios corruption warnings and poo poo before I cleared it- it was super messy but thankfully no harm done. Real panicky moment though. I'll be honest, the ram fiasco wasn't a surprise to me as I'm running a matched pair of 2x2gb + a single 4gb stick. They're all 1.5v and the same timings, but the single stick is 1600 and the matched pair are 1333 and I had a feeling that funny things would happen if I messed around with them. Now that the system is running 'stable' (1 hour of prime95 blend, probably about 6-8 hours of Cities Skylines and general usage between Sunday and today with 0 problems) I'm wondering what to do next. Leaving everything other than FSB at stock also leaves my motherboards vcore setting in Auto mode, which according to hwinfo has been stepped up to 1.45 (1.4 stock). Should I try to manually enter a lower vcore and see if it's still stable? At this point heat is a non-issue, 30 mins of the small fft test got me to a whopping 40c, though the ambient room temp is usually below 20c which is a factor.
|
|
#
?
Apr 7, 2015 16:27
|
|
|
You didn't specify your ram settings, changing FSB also plays with the DDR speed so you should pay attention to that. I too have unmatched 1333/1600 sticks and I find it better to lock them all around 1333. With some tinkering I'm pretty sure you could reach 4Ghz, I switched to intel after hearing everyone bashing AMD all day but I must say I've had better experience with them so far and they were easier to OC.
|
|
#
?
Apr 7, 2015 16:40
|
|
|
Sh4 posted:You didn't specify your ram settings, changing FSB also plays with the DDR speed so you should pay attention to that. I too have unmatched 1333/1600 sticks and I find it better to lock them all around 1333. With some tinkering I'm pretty sure you could reach 4Ghz, I switched to intel after hearing everyone bashing AMD all day but I must say I've had better experience with them so far and they were easier to OC. Sorry, you are right. I'm not at my home pc right this moment but if memory serves me correctly I left the ram settings on auto (other than my failed attempt to increase the voltage) however I did try one configuration with it manually dialed down to 1333 after cranking up the fsb so I can't with 100% certainty say that it's not still on that setting . I'll verify tonight after work and maybe get some screenshots for more relevant info.
|
|
#
?
Apr 7, 2015 16:57
|
|
|
|
| # ? Jun 14, 2024 12:15 |
|
|
I'm fiddling with my new i5, and I wanted to run some overclock specs past you guys for your thoughts. For reference, I'm on a i5 4690k and using a Hyper 212 heatsink/fan. I started off with similar settings that I had on my G3258, that is, about 1.95vCore and a 40~ multiplier, everything booted up fine, and after some fiddling I knocked the multiplier up to 42 and left the voltage the same. It boots fine, seems to be running no problem, idle temps at 30-28. When I run Prime95 Blend, it hovers between high 60s, mid 70s, which I would kinda expect, except that it very occasionally spikes up to mid 80s for about 60 seconds, before going back down to the 60s-70s range, in an hours worth of benchmarking, it hit 88 as a peak once, but never went over 85 after that. Otherwise, it works fine. Should I be at all concern about the peaks? I'm going to run something like Cities Skyline or Dragon Age and see what a real world load temp looks like in a moment, but I'm thinking that those extreme peaks are just Prime95 being insane, especially considering those jumps to 80 never lasted more than 60 seconds.
|
|
#
?
Apr 7, 2015 17:31
|
|