|
Eletriarnation posted:Yeah, I believe that's right. Idle lowers the multiplier to a very small value and all you're doing by overclocking a K-series is increasing the maximum allowed multiplier. I don't think the load power consumption will go up that much either if you're not touching voltage, considering Ohm's Law. Alright, thanks. Finally getting around to this, and I have another noob question: how does overclocking interact with Intel's turbo boost feature? My 6600K has a 3.5 GHz stock clock with 3.9GHz turbo. But if I raise the multiplier only the stock clock changes, so I get e.g. 4.2 GHz stock / 3.9 turbo - while ideally I'd like the reverse (only hit the higher multiplier when necessary). I can't find any way in the BIOS to alter the turbo multiplier, and it seems silly that I'd have to turn off the feature entirely in order to OC.
|
#
?
Feb 14, 2016 20:24
|
|
|
|
| # ? Jun 12, 2024 12:10 |
|
|
Anyone have any experience at playing with the 6600k's igpu clock speeds?
|
|
#
?
Feb 15, 2016 15:08
|
|
|
Captain Hair posted:I didn't realise the q9550 was such a good overclocker! Just got one as a hand me down to keep my gaming computer limp along another year or so. Been running a core2duo overclocked to 3.6 for the past like ten years it feels like and moving over to a quad core has breathed new life in to my 560ti its incredible. Wreckfest went from 5-15 fps to 30or 60fps solid depending on how many cars on screen. Those old C2Q chips were pretty badass for the time, a PC I built a few years back in '10 or so was a Q9400 I got up to 3.76Ghz on air with a DDR2 setup. That thing lasted quite a while, I sold it to a friend who built a DDR3 based LGA 775 system and I think he was able to push it further near 4GHz by pushing the FSB harder (I think my old RAM held me back on mine, highest stable settings I got were 471 x 8 @ 1.31v). NihilCredo posted:Alright, thanks. IIRC you want to look for something called "uncore multiplier" or similar that will let you set the lower frequency, then lock the actual multiplier at what you want the OC speed to be (e.g. uncore set to 39 multiplier, OC set to 42 multiplier). At least I think that's how it's been on Haswell and previous anyhow. Anyone here can feel free to correct me though, I could be wrong but that's how I understood it. E: Broccoli Cat posted:I was messing with EVGA's Precision X, which seems like some bullshit graphic thing that displays some stats and pretends to change things which it really doesn't. Have you checked your clock settings by making changes in Precision & then using GPU-Z or a similar tool? Sometimes there's an option in the app settings that you have to tick to unlock certain features, I'd hunt around first to see if anything is "locked" at all. Otherwise you can try other tools like MSI Afterburner, I've used it for a while and never had trouble regardless of card vendor or branding. BOOTY-ADE fucked around with this message at 17:48 on Feb 15, 2016 |
|
#
?
Feb 15, 2016 17:04
|
|
|
Ozz81 posted:
turned out I was just being a drunken paranoid idiot, and the thing does actually work. Got my old 750 ti cranked up to, get this, 1337mhz...I laughed my rear end off. last night I did my first CPU overclock Got my 8320 running like a champ at 4.0, at around 55C....nothing special, but still getting over on the man.
|
|
#
?
Feb 18, 2016 04:43
|
|
|
So I have had a 2500k in the wifey's computer for the past 4 years or so and have never overclocked before. I went ahead and bought a Pure Rock Be Quiet! cooler to replace the stock Intel and decided to see if I could give this puppy a bit of an OC. It's running on a gigabyte P67A-D3-B3 Bios F7. I went ahead and just set the multiplier to 43 for the time being and didn't mess with any of the voltage settings, they are all left on auto. Running Prime95 Blend the temps are looking pretty good, staying in the mid 50s to low 60s but the VID stays sitting around 1.36 to 1.376 and I am wondering if that voltage is safe for this chip? Vcore is sitting around 1.26/1.27. I honestly have no idea. I am using HWINFO64 to monitor temps and voltages Secondly I did OC the RAM from 1333 to 1600. I ran memtest86 overnight for about 12 hours and that showed no errors so I assume that my RAM overclock should be fine then. B-Mac fucked around with this message at 13:26 on Feb 22, 2016 |
|
#
?
Feb 22, 2016 12:49
|
|
|
B-Mac posted:So I have had a 2500k in the wifey's computer for the past 4 years or so and have never overclocked before. I went ahead and bought a Pure Rock Be Quiet! cooler to replace the stock Intel and decided to see if I could give this puppy a bit of an OC. It's running on a gigabyte P67A-D3-B3 Bios F7. I went ahead and just set the multiplier to 43 for the time being and didn't mess with any of the voltage settings, they are all left on auto. Running Prime95 Blend the temps are looking pretty good, staying in the mid 50s to low 60s but the VID stays sitting around 1.36 to 1.376 and I am wondering if that voltage is safe for this chip? Vcore is sitting around 1.26/1.27. I honestly have no idea. I am using HWINFO64 to monitor temps and voltages If it's still set to auto then it should be running at stock voltage. Those numbers don't sound like anything unusual to me so I think you're just fine. You can quite likely go higher, I have the same chip at stock voltage too and it's been set to 44x for years with a Hyper 212. Tcase for the 2500k is 72C so that's the temperature you should try not to exceed. If the RAM got through memtest and Prime95 just fine then I would consider it stable until seeing evidence that it isn't, and it's reasonable to expect that a lot of 1333MHz kits could do 1600 no problem.
|
|
#
?
Feb 22, 2016 19:50
|
|
|
Eletriarnation posted:If it's still set to auto then it should be running at stock voltage. Those numbers don't sound like anything unusual to me so I think you're just fine. You can quite likely go higher, I have the same chip at stock voltage too and it's been set to 44x for years with a Hyper 212. Tcase for the 2500k is 72C so that's the temperature you should try not to exceed. Awesome, thanks for the reply! Ill probably try 44 or 45 here, pretty amazing how much extra life you can get our of a five year old chip!
|
|
#
?
Feb 23, 2016 00:21
|
|
|
I need to buy a new power supply and I was wondering how much I needed to support mild overclocking. I have a i5-4690k, ASRock z97m-itx, and plan on getting that MSI GTX 970 4G model graphics card that people are recommending. PPP estimates a wattage of about 300W, so by the standard 120% estimate that's 360W so I should get about a 450W supply? Again if I wanted to do some light overclocking to both the CPU and GPU, is 450W enough or should I jump up to 550W? By light overclocking I'm really just talking about getting whatever "free" performance out my computer I can get without going nuts. I was also going to get a Cooler Master Hyper 212 cooler for the CPU as well.
|
|
#
?
Feb 28, 2016 15:04
|
|
|
Boris Galerkin posted:I need to buy a new power supply and I was wondering how much I needed to support mild overclocking. I have a i5-4690k, ASRock z97m-itx, and plan on getting that MSI GTX 970 4G model graphics card that people are recommending. PPP estimates a wattage of about 300W, so by the standard 120% estimate that's 360W so I should get about a 450W supply? Again if I wanted to do some light overclocking to both the CPU and GPU, is 450W enough or should I jump up to 550W? By light overclocking I'm really just talking about getting whatever "free" performance out my computer I can get without going nuts. I was also going to get a Cooler Master Hyper 212 cooler for the CPU as well. The 550w EVGA G2 Supernova is $75 on Amazon Prime and has a 7 year warranty and exceptional components, I'd spring for that. Cryorig coolers tend to have easier installation than the 212, as well as better noise/cooling performance for the money. http://pcpartpicker.com/search/?cc=us&q=cryorig
|
|
#
?
Feb 28, 2016 18:39
|
|
|
Boris Galerkin posted:I need to buy a new power supply and I was wondering how much I needed to support mild overclocking. I have a i5-4690k, ASRock z97m-itx, and plan on getting that MSI GTX 970 4G model graphics card that people are recommending. PPP estimates a wattage of about 300W, so by the standard 120% estimate that's 360W so I should get about a 450W supply? Again if I wanted to do some light overclocking to both the CPU and GPU, is 450W enough or should I jump up to 550W? By light overclocking I'm really just talking about getting whatever "free" performance out my computer I can get without going nuts. I was also going to get a Cooler Master Hyper 212 cooler for the CPU as well. You should get the gigabyte 970, it's better. For PSU 450W is enough but if you're upgrading often I'd get something like 600w, for me the best brand for PSU's is beQuiet but Corsair is pretty good too. While the 212 is a pretty good cooler, if you're serious about overclocking it will not be enough and you will hit the heat wall pretty fast. The best air cooler imo is the bequiet dark rock pro 3, it has amazing cooling but is also nearly silent compared to something like the cryorigs and phanteks that are really loud.
|
|
#
?
Feb 28, 2016 20:29
|
|
|
Sh4 posted:You should get the gigabyte 970, it's better. The gigabyte 970 is substantially louder than the MSI, and is basically identical in all other aspects. BeQuiet and Corsair are both mediocre brands for PSUs (they have both good and bad units, and their good ones are often overpriced)
|
|
#
?
Feb 29, 2016 10:22
|
|
|
BurritoJustice posted:The gigabyte 970 is substantially louder than the MSI, and is basically identical in all other aspects. BeQuiet and Corsair are both mediocre brands for PSUs (they have both good and bad units, and their good ones are often overpriced) Though the MSI could still be quieter when they're at 40% or whatever.
|
|
#
?
Feb 29, 2016 10:40
|
|
|
BurritoJustice posted:The gigabyte 970 is substantially louder than the MSI, and is basically identical in all other aspects. BeQuiet and Corsair are both mediocre brands for PSUs (they have both good and bad units, and their good ones are often overpriced) You can flash fan curves on the gigabyte and turn off fans at idle as well so they should be on par but the MSI's don't have a backplates and they are not as good with OC. The only problem with G1's is that they exist in 2 revisions and 1 is better than the other and there's no way to know which one you're getting.
|
|
#
?
Feb 29, 2016 15:03
|
|
|
I am interested in putting together a very cheap junkyard/Home Depot water cooling setup with my nephew. I see there are VERY cheap (<$5) water blocks on ebay, along with little brackets to hold them down securely. I imagine we could use something like an automotive heater core or maybe an oil cooler for a cheap radiator. It seems like the reservoir could be anything from a pickle jar to one of those plastic vodka bottles. The pump is where I get somewhat worried. I want something built to last. Is there any reason that an automotive electric water pump wouldn't work? I am specifically thinking of the unit used in the Toyota Prius to cool the inverter module. These units are available used from junkyards for $24-40 or brand new from Toyota for $80. I am confident that these pumps are built to a very high level of quality, but I can't seem to find specs (flow rate/pressure) on the pumps. The guy in this video: https://www.youtube.com/watch?v=1ojNID2jDz4 demonstrates that the pump is pretty powerful. He also demonstrates a power draw of 1.8A at 12V. This seems reasonable from an automotive electronics perspective; will a ~500W PC power supply be able to handle this? Physics class tells me that a 1.8A at 12V is 22W, which ~seems~ reasonable compared the the 75W that a 6-pin PCIe connector can provide. What are the chances I could run this pump at 5V for reduced flow? I know the real improvements in water cooling come on the GPU side. I know I can figure out a way to fasten a cheap water block onto the GPU. What is the cheap way to cool the memory and VRMs? can I just attach little cut up heatsink pieces on them with thermal paste? Is this even necessary? I have a background in aftermarket automotive electronics, including custom audio installation, so mild fabrication work is not out of the question. Any thoughts?
|
|
#
?
Mar 15, 2016 16:58
|
|
|
A reservoir is not required for a waterloop. A t fitting and a bit of hose right before the pump is fine.
|
|
#
?
Mar 15, 2016 17:16
|
|
|
I have to say this is something I've thought about. In concept most auto parts will run just fine, and AIO radiators are simply just... tiny radiators with a healthy markup. A heater core would probably provide incredible cooling performance. ... but its all going to come down to the details. Going to be a lot of little problems along the way hooking this poo poo together (and mounting, leakproofing, etc) with plenty of cursing. But it would definitely be a journey is the reward sort of thing, at least for me. Plus it is nice when it does work it'll be a kick rear end cooler.
|
|
#
?
Mar 15, 2016 17:29
|
|
|
Arn't most heater cores going to be aluminum though? PC watercooling gear is generally designed with metals that do react poorly with aluminum.
|
|
#
?
Mar 15, 2016 17:38
|
|
|
Don Lapre posted:Arn't most heater cores going to be aluminum though? PC watercooling gear is generally designed with metals that do react poorly with aluminum. I'm pretty sure all radiators are going to be aluminum. The only part that matters (I thought) was the interface between CPU and water block, and I cant remember why even (besides better thermals but)
|
|
#
?
Mar 15, 2016 17:46
|
|
|
Radiators on pc's are 99% of the time copper/brass. The only exception are thermaltake and other lovely chinese companies. Fittings/blocks are also 99% of the time copper/nickle. Mixing that with aluminum is a big no no in your loop.
|
|
#
?
Mar 15, 2016 17:54
|
|
|
I think the watercooling community freaks out about metals because of reactivity of different metals, because they tend to use straight water. Automotive coolant has anticorrosive additives to take care of this problem.
|
|
#
?
Mar 15, 2016 17:55
|
|
|
It does but distilled water with a biocide is better for watercooling than automotive coolant. Unless you are running your pc at sub zero temps of course.
|
|
#
?
Mar 15, 2016 17:56
|
|
|
Don Lapre posted:Radiators on pc's are 99% of the time copper/brass. The only exception are thermaltake and other lovely chinese companies. I'm pretty sure the majority of PC radiators (didnt really mean open loop, which makes sense for them to splurge on copper) are aluminum but I did not consider mixing fittings, which sounds very important when making your own stuff out of car parts lol. Although a lot of heater cores are copper too Hardly a source I know but asetek does provide most radiators to AIO's if I recall http://www.asetek.com/desktop/oem-cpu-coolers/545lc/ edit: http://www.cgj.com/2013/07/02/aluminum-vs-copper-brass-radiator-corrosion-susceptibility/ was an interesting read for me, but yeah mixing copper and aluminum is a bad idea. So for whatever core you get, gotta make sure you keep the fitting material consistent. Fortunately that would be easy to do in this case penus penus penus fucked around with this message at 18:11 on Mar 15, 2016 |
|
#
?
Mar 15, 2016 18:03
|
|
|
Yea, the all in ones are almost all aluminum unless they allow you to open and expand the loop. Thats because AIO's are poo poo and using the lowest quality parts. They also tend to have very toxic and corrosive chemicals in them. Heaven forbid they drip onto your equipment.
|
|
#
?
Mar 15, 2016 18:33
|
|
|
You mean its antifreeze? I'm not sure I'd consider AIO's poo poo because they use aluminum, considering their performance. I thought the appeal to open loop was being able to cool multiple things rather than, well, quality of parts. I am also quite sure the chance of leaks in open loop are greater than a sealed AIO just based on the nature of its assembly. But... anyway... I support the guy using junkyard parts lol because the potential there is very great but to your point be careful on materials. penus penus penus fucked around with this message at 18:40 on Mar 15, 2016 |
|
#
?
Mar 15, 2016 18:35
|
|
|
I just got my new case fans in yesterday so I started playing around with overclocking. I have an i5 4690k and every guide I've read says to start off by bumping vcore up to 1.2v or whatever. But if I understand right, you only need to bump up the vcore setting if your CPU is giving errors which always(?) results in a BSOD so it's easy to tell? So I decided not to listen to the guides and manually set my vcore override to the default state of 1.009v (according to the UEFI HW Monitor). I also set the uncore/cache ratio multiplier to 35 (default) like I read somewhere to keep it from auto-adjusting. Then I increased the core multiplier from 35 and got it up to 40 in steps of 1/100MHz before I ran out of time and had to go to bed. After each step I ran prime95 with "in-place large FFTs" since it's labeled as "maximum power consumption" and I'm assuming that I'm more worried about whether or not my CPU is getting enough power at this point. Each run was roughly 10-20 minutes while I was off doing other stuff and when I stopped it it gave me no warnings or errors. I also ran the small fft mode once to see what my max temperature was at this vcore setting which was only about 60 C so not too bad. I ran prime95 in blended mode over night and this morning I stopped it and it gave zero warnings and errors at 4000 MHz and 1.009v, with max temperatures of 60 C and idle temperatures of 20-25 C. I also dialed the CPU fan way the gently caress down in my BIOS to keep it at the lowest speed setting when under 30 C and ramp it up linearly to maximum rpm at 60 C. It seems like I can't tell my CPU fan (Cryorig H7) to spin at less than 30% which is a bummer because my idle temperatures are still pretty drat low at 30% rpm. And I set my case fans to spin at "silent mode". The max 60 C temperature I got was with all of these fan settings. Is it seriously this easy to overclock? Like wow I vaguely remember when I built my first computer ever with an Athlon ... Thoroughbred(?) that you had to take a pencil and connect a trace together on the chip and changing fsb settings or whatever the gently caress. Anyway I guess my next steps when I have time are to keep increasing the core multiplier and running prime95 with large ffts until my computer BSODs from not getting enough power, and then either increasing the vcore setting and repeat. I'm trying to shoot for idle temperatures of 30 C max and loaded temperatures of 70 C max with prime95 which I read is "safe" territory. It's also my understanding that the small fft test in prime95 is going to be my absolute worst case maximum temperature reading since games I'm playing aren't generally going to put so much loading on my cpu, is that right? e: During all of this testing last night I realized I set my processor minimum power state to 100% at some point, not sure why. Is there a reason not to bump this back down to the default 5%? I'm assuming that Windows would automatically adjust the clock speed back up to 4000 MHz when needed so there's no point in keeping it running at 4000 MHz when I'm just sleeping. Boris Galerkin fucked around with this message at 10:03 on Mar 16, 2016 |
|
#
?
Mar 16, 2016 09:22
|
|
|
Boris Galerkin posted:I just got my new case fans in yesterday so I started playing around with overclocking. I have an i5 4690k and every guide I've read says to start off by bumping vcore up to 1.2v or whatever. But if I understand right, you only need to bump up the vcore setting if your CPU is giving errors which always(?) results in a BSOD so it's easy to tell? It seriously is that easy, but your CPU is probably auto volting. I doubt 1.009v can handle stock full load clock much less any overclock. Are you running CPUz (or HW monitor or some variant) while stress testing? That will show you real time voltage. In any case, 4.0 ghz is an easy overclock that is very likely to be handled well by auto voltages (if it had to alter them at all over stock values). Auto voltages are not generally recommended due to instability, heat, and unnecessarily high vcore but at 4.0 ghz I would doubt its even worth setting anything in the bios. Also it will depend on your motherboard and it has gotten a lot better overall. But you have a lot on the table. You have an easy 20 degrees of thermal overhead to go under the same stress test situations which is a lot. I'd expect a fairly simple 4.5 ghz overclock out of that chip, but you will have to manually test and set voltages.
|
|
#
?
Mar 16, 2016 15:41
|
|
|
THE DOG HOUSE posted:It seriously is that easy, but your CPU is probably auto volting. I doubt 1.009v can handle stock full load clock much less any overclock. I thought that setting my vcore override to 1.009v disabled auto volting? I have HWInfo64 running still from last night and just got home so these are the values I'm seeing:  Current/Min/Max And in the motherboard category  Current/Min/Max I don't see a vcore measurement from the motherboard somehow. e: I just started up prime95 and cpuid is giving me these values:  and cpu-z  Nothing makes sense anymore. edit2: here's my uefi setting  According to the screen it says override mode "the voltage is fixed" to my input of 1.009v. I guess technically I have the additional offset left as auto, but i thought it only used this if I chose "adaptive mode." I'll set it to 0 so it doesn't auto anything and set my core multiplier back down to 35 and try again. Maybe that 1.033v value in cpuid is giving me the adapted vcore? Boris Galerkin fucked around with this message at 18:00 on Mar 16, 2016 |
|
#
?
Mar 16, 2016 17:32
|
|
|
For fun I just changed my core multiplier back to stock, and changed the 'vcore override voltage' to 1.02v just to see what hwmonitor, cpu-z, and hwinfo64 measures. HWInfo64 reports 1.019v under "cpu core # vid", everything in the motherboard section is unchanged. CPU-z reports 1.019v as well. HWMonitor is interesting though. The numbers under the motherboard aren't really different but under the cpu tab I get this: It's showing "LLC/Ring Offset" as 0v now at a stock clock speed, but in the other post when I had it at 4 GHz it was showing "LLC/Ring" (no "offset") at 1.229v. I can't really find out any information about what "LLC/Ring Offset is" or why it would be different. e2: after playing around with these settings and reading the reply in this http://forums.tweaktown.com/asrock/57879-want-overclock-i5-4670k-asrock-z87-extreme-6-motherboard.html post it seems like vcore readings in Haswell systems are a crapshoot because Intel hasn't publicized a method of extracting this data from the CPU. I'm going to go back to my assumption that override mode fixes the vcore to whatever I typed in. From what I read the field labeled "LLC/Ring" voltage deals with the uncore/CPU cache (what ASRock calls it) voltage but since I'm fixing it to 3500 MHz it shouldn't really do anything. Either way I tried setting the "CPU cache voltage mode" to override and setting it to a value of 1.009v and booting. CPUID reported 1.033v, a difference of 0.024v from what I set. In the UEFI I now set "CPU cache voltage offset" to -0.024v and CPUID now measures 3+ volts. So clearly this field is nonsense, and hwinfo64 doesn't even report a voltage set to 1.009 or 1.033 other than the "CPU vid" fields. For fun I also set "system agent voltage offset" to a value of 0.204v because this is what hwmonitor reports at stock settings. On boot it reports 0.209v instead. So yeah I'm gonna assume override mode fixes the vcore to what I set and this is really the best option I have given the limitations of measuring vcore in Haswell chips. Boris Galerkin fucked around with this message at 19:51 on Mar 16, 2016 |
|
#
?
Mar 16, 2016 18:33
|
|
|
Well the values seem close enough to 1.009 to assume you really are running close to that. It really isn't, but it shouldn't sway too far from it. That's a fairly low voltage and you have a ton of room to go. I've seen factory Haswell processors run at 1.20 vcore. You're doing it right so, keep going. I know this is rear end backwards but I've messed with OC so many drat times all I do is set the voltage I want to be at and see how fast the processor can go then back the voltage down. 1.30 vcore is generally considered moderate and safe for Haswell. 1.375 is the "safe limit", but I kind of doubt you'll hit those voltages anyway without running too hot.
|
|
#
?
Mar 16, 2016 20:43
|
|
|
Just to chime in, don't worry about the power settings thing if you're running a desktop. As far as I understand, it doesn't affect speedstepping or adaptive at all. For example, mine is set to 100% all the time, and it idles here:  And goes to here  Incidentally, I feel I should probably be pushing this chip a bit, but I'm really pleased with this sweet spot and haven't screwed with it. The volts go lower, but it gets screwy after suspend on adaptive, and I understand from the super overclocking professional smartmen that adaptive is bad and kills puppies. I don't know if this is true or not.
|
|
#
?
Mar 17, 2016 07:20
|
|
|
Processors have various features aside from adaptive voltage that let them clock down at idle and deeper. C3 and Intel's EIST drop voltage and frequency respectively at idle, in addition to the various things that Intel Turbo boost does. I'm not familiar with which features override what, however. (my previous rig ran most of its life on Adaptive, but this time I'm just so  that I'm just going to rely on these features, hope that's ok)As for Boris Galerkin, LLC/Ring and System Agent/FCLK voltages have been a crapshoot for me, too.  I think only Skylake has allowed end-users to change the FCLK on their processors, so that voltage is definitely not useful. From what I've read, LLC/Ring voltage should only be messed around with when pushing up your VIN/VRIN/whatever your motherboard calls your CPU INPUT voltage is no longer an option. Anyways, I have stuff to post too! It's about a month late and half a year behind the curve but Broadwell trip report, day 1: Intel Gave Their Special Butterfly An Artificial TDP Wall I was pretty proud of myself, having put together my new rig with few to no issues. Enjoyed myself doing exciting stuff like  but then I realized that my frequency was throttling really hard (down to 2.1GHz in some cases) on various different stress tests, even though THREE HUNDRED AND SIXTY MILLIMETERS OF SLOVENIAN STEEL--i mean copper rads easily kept temperatures below 65 Celsius. I took forever trying to enable/disable a bunch of relevant motherboard settings, to no avail. All I had to do was look one entry above Clocks in HWMonitor to realize that my fun was being killed by 65.00W 64.99W 65W 64.90W 65W 65W 65W etc. This power envelope significantly slowed down my sustained processing speeds (worse than stock!) at full load from 1.225V and onwards. thanks intel Raising my Intel Turbo power draw limits in BIOS didn't work, so I had to dig around on various forums for the answer to this roadblock. Strangely, Intel lets you past their own limitation.  This is Intel's Extreme Tuning Utility, and these sliders are available on Broadwell but not on Haswell, at least from what I've read. Will play with them tomorrow and try not to Chernobyl my 5775C.
|
|
#
?
Mar 17, 2016 08:05
|
|
|
Yeah I understand what you guys are saying about just pushing the voltage up to 1.2v because it's been known to be stable and seeing how high the frequency then goes. If I do this again in the future with future builds then I'll probably just do that but for now it's pretty fun just learning by trial. I ran another six hours of prime95 blend mode last night at 4100 MHz at 1.009v still and hit a max temperature of 63-65 again with no errors so at this point I'm just really curious how high I can push it. e: re: vcore measurements on Haswell this is what that guy in that forums I linked to had to say quote:Now you've encountered a reality about Haswell processors and ASRock Z87 mother boards, a true VCore reading is not available. The data from the sensor chip on the board does not have a VCore voltage reading. The best we get are the VIDs. That is the case in the UEFI/BIOS, A-Tuning, and any other monitoring program. Boris Galerkin fucked around with this message at 08:49 on Mar 17, 2016 |
|
#
?
Mar 17, 2016 08:47
|
|
|
Broadwell-C trip report, day 2: That was easy. Running 4.2GHz like most people, at 1.35 volts. To prevent having to boost that VCore even higher, I goosed it a little with a couple of extra tenths of a volt to CPU INPUT/VRIN as well as a few hundredths to uncore/cache. It looks like I got a fairly middling result from the silicon lottery, and I certainly  to handle that kind of voltage for constant use, so I'm satisfied. I'm going to check out what kind of clocks I can squeeze out of the uncore, L4 cache, and memory, and then it'll be on to the GPU. Sidesaddle Cavalry fucked around with this message at 06:18 on Mar 19, 2016 |
|
#
?
Mar 19, 2016 06:12
|
|
|
So just out of curiosity what would happen if I accidentally set my UEFI vcore to 1.91 v instead of 1.19 v? I mean everything I've read said I'll most likely kill my CPU but are we talking instantly booting off and not being able to boot on again (so kind of boring) or am I gonna see sparks/hear capacitors exploding in my case (which is objectively more fun)?
|
|
#
?
Mar 21, 2016 13:41
|
|
|
Not sure, but I wouldn't expect fireworks. Either the cpu\mobo would shut off the system quick I'd hope. If not I'd assume it would just die internally and quietly? Only time I've had an actual fireworks type situation was back on an a-bit nf7s. I turned on the pc and was greeted with plumes of grey smoke and a small motherboard fire near the ram. Fun fact, a-bit rma'd the board and said it was a "known issue"  and sent me a new one. Swapped it in and the system ran perfectly again. Only the eternal stench of burning pcb remained...
|
|
#
?
Mar 21, 2016 14:05
|
|
|
I'm going to upgrade my RAM today from 8 GB @ 1600 MHz to 16 GB @ 1866 MHz. I know I need to run memtest to test the RAM itself but do I need to redo my stability testing for my CPU overclock? Both the old and new RAM should be using 1.5 V according to the xmp profiles.
|
|
#
?
Apr 13, 2016 09:28
|
|
|
I'm fairly new to overclocking and have a question for the thread. I have an i5 6600K overclocked to 4.4GHz on a Gigabyte Z170x-Gaming 5 motherboard. My RAM is 2x8 DDR4-3000, and initially was clocked to 2133MHz when I started my system. However, as I overclocked my CPU, it brought my RAM down to 1066MHz. Should I bother about getting it back up to speed? If so, what's the best/easiest way to go about it?
|
|
#
?
Apr 29, 2016 16:14
|
|
|
Did you check the clock speed in the same place both times? DDR memory sends information twice per clock cycle (hence Double Data Rate) and a lot of utilities will report the actual clock speed, which is half of the effective, advertised clock speed. Your RAM is almost definitely still running at 2133. As for getting it up to its rated speed of 3000, you should enable Extreme Memory Profiles (XMP) in your motherboard's UEFI (or BIOS if you're familiar with the older term) and then set your memory profile to, most likely, XMP Profile 1. You should be able to find information about this in your motherboard's manual or online, if you can't find the option. HMS Boromir fucked around with this message at 16:34 on Apr 29, 2016 |
|
#
?
Apr 29, 2016 16:31
|
|
|
Alright, just turned XMP profile 1 on, and that got me up to 1600MHz. Checked in BIOS, which is where I initially saw it reading at 2133. Just to make sure, I ran Aida64's cache and memory benchmark and got this: Also found out that apparently my computer doesn't like me exiting BIOS without saving. Didn't bother to note the code it was throwing on account of a reset fixing it right up.
|
|
#
?
Apr 29, 2016 17:06
|
|
|
|
| # ? Jun 12, 2024 12:10 |
|
|
Huh. That does look like it's actually running at 1600 MHz. Was there another XMP profile available? If there's multiple profiles there's sometimes a more conservative one in case you can't actually run the module at tested speeds but 1600 is pretty much hitting bedrock as far as DDR4 clock speed. From your description it doesn't sound like it but if that error code came after attempting to boot with your RAM at 3000 it could've been it telling you the overclock failed. You could also just try setting it to 3000 MHz manually - the timings are mostly the same across the board with DDR4 from what I've seen so you'd just have to make sure it's getting 1.35V. EDIT: Found someone having a similar problem with a similar Gigabyte Z170 motherboard (theirs was a Gaming 7 model) that was solved by a BIOS update so you could look into that as well. HMS Boromir fucked around with this message at 17:37 on Apr 29, 2016 |
|
#
?
Apr 29, 2016 17:30
|
|