Overclocking Megathread 3.11 for Workgroups

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > Overclocking Megathread 3.11 for Workgroups

«‹›5 »

future ghost: Dec 5, 2005; Gun Saliva

Badass OP, badass thread title.

Not really sure what else to say here, other than nice work. The OP is just so comprehensive :allears:

e: Forgot to add from the last thread - The overclockers.com Nehalem guide is the original revision (and there's a few minor errors in some of the steps).
Miahallen hosted the finalized guide on techreaction: http://www.techreaction.net/2011/01/04/3-step-overclocking-guide-%E2%80%93-sandy-bridge-v0-1beta/

Didn't read the last section fully so I missed it the first time around.

future ghost fucked around with this message at 05:44 on Feb 7, 2012

# ¿ Feb 7, 2012 05:41

Adbot: ADBOT LOVES YOU

# ¿ Apr 28, 2024 04:34

future ghost: Dec 5, 2005; Gun Saliva

Scalding Coffee posted:

It was thanks to the old thread that I had no trouble with overclocking. Now for some sort of futuristic dusting system, as this monster attracts more dust than a swiffer.

Pantyhose stretched over the inside of a case grill/outside a fan will do it on the cheap. I am in no way joking.

# ¿ Feb 7, 2012 05:57

future ghost: Dec 5, 2005; Gun Saliva

Factory Factory posted:

The one you linked there is a Sandy Bridge guide and I'm having a of a time trying to Google a more current version of the Nehalem one. Should I just nix that entirely?

Welp, I'm 0 for 2 tonight (oops).

The correct link is this one: http://www.techreaction.net/2010/09/07/3-step-overclocking-guide-bloomfield-and-gulftown/

# ¿ Feb 7, 2012 06:05

future ghost: Dec 5, 2005; Gun Saliva

Fil5000 posted:

An actual thread relevant question - should I be following the three step overclocking guide that grumperfish linked for overclocking my 760? I realise I won't be going for exactly the settings in the link as that's not specifically for the 760 but the principles seem clear(ish).

FF, the OP is superb - I especially liked the explanation for exactly WHY you need to raise voltages to avoid crashes really helpful. The time spent is greatly appreciated.

This version of the guide is specifically-tailored to Lynnfield chips. Forgot they did something wacky that generation:
http://www.techreaction.net/2010/09/07/3-step-overclocking-guide-lynnfield/

e: Zalman apparently just released a 6950/6970 VRM heatsink kit finally. I spent all that time custom-modifying copper mosfet sinks for my card since nothing was available at the time :colbert:

. Don't know if I'm going to bother switching at this point, but if anyone else has a reference-model 6950/6970 and they're looking to replace the stock blower, this is probably the easiest solution available to cover the VRM chips without using thermal glue or custom heatsinks.

future ghost fucked around with this message at 03:19 on Feb 8, 2012

# ¿ Feb 8, 2012 01:20

future ghost: Dec 5, 2005; Gun Saliva

Gorilla Salsa posted:

I made this dumb post about overclocking my i7 920 back in the old thread, and it was suggested that maybe the problems I was having with stability were a result of poor cooling. I installed a Coolermaster 212+ Evo and my temps at load dropped from 83 to 68, give or take a couple of degrees. I used the same settings and ran Prime 95, but the stress test only lasted 30 minutes before it bluescreened out. At least, that's what I'm thinking is going on based on the HWiNFO64 log.

Any suggestions? I'd really like to make this 3.6Ghz overclock work.

Your BIOS settings look alot better now, and temperatures are in check at least.

What BSOD code are you getting? Check the event log in Windows for critical errors/restarts.
E: Have you run memtest to verify that your RAM is OK at stock settings?

future ghost fucked around with this message at 17:53 on Feb 10, 2012

# ¿ Feb 10, 2012 17:46

future ghost: Dec 5, 2005; Gun Saliva

Alereon posted:

As long as temperatures are okay (not going above 100C)

ATI/AMD cards should stay at or below 80C at worst-case. 90C-100C temperatures would rapidly kill one of these cards.

PS. Love the cabin posted:

Can we take breaks during the 24 hour burn in?
Web browsing isn't going to be an issue but if I wanted to play something I'm not sure it'd like prime running in the background.

I don't really believe in the necessity of a 24-hour burn-in period myself, so at the most I'll run prime for like 4-6 hours maximum as an initial test after ~2 passes of memtest & 10 runs of max IBT. If it doesn't fail in that period, it's fine as far as I'm concerned. Beyond initial testing, day to day use will generally reveal any possible lingering problems.

If you're going to do a 24-hour burn in for whatever reason, you should do it in 1 stretch or at least 2 12-hour stretches ideally, although I still think it's excessive. If you want to play a game, by all means just turn off prime. Gaming will still be stress-testing the system.

future ghost fucked around with this message at 17:14 on Feb 11, 2012

# ¿ Feb 11, 2012 17:10

future ghost: Dec 5, 2005; Gun Saliva

Phuzun posted:

Looks like Nan's gaming has the 2011 retention kit now for the Thermalright heatsinks. Also, the True Spirit that they sell is another really good heatsink, from the reviews I've seen. I'll most likely be throwing on of those on a q9550 that I have laying around and use it as a server. The 120mm version is only $30, which is a great price.

Nice find. Looks like it handled a 920 C0 at 3.9ghz with reasonable temperatures, so it has alot of cooling capacity for the price (and it's presumably easier to mount than a 212+ due to being a Thermalright product) :

http://www.overclockers.com/cogage-true-spirit-lga1366-heatsink-review/

The one on Nan's Gaming Gear appears to ship with TR's Venomous-X style mount instead of the TRUE-style crossbar mount. The Ven-X mount is great because you can re-mount the heatsink without needing to remove the board from the case.

future ghost fucked around with this message at 01:44 on Feb 12, 2012

# ¿ Feb 12, 2012 01:41

future ghost: Dec 5, 2005; Gun Saliva

Gorilla Salsa posted:

So here is what Memtest 86+ v4.20 came up with:

I'm guessint his is good, which leads me to guess that I'm hosed as far as a potential overclock is concerned. Right?

Your RAM is fine (it's a bit below stock speeds, but don't worry about that at the moment), so there's only a couple other things to check really that I can see..

In BIOS, disable turbo to keep the multiplier at a specific setting. I think from your list it's: "CPU Ratio Setting: Auto"

Check the Windows Event Log for the BSOD codes, so we can rule out harddrive failures, etc.

Did you run through the Nehalem overclocking guide in the OP? It might be a good plan to revert to stock settings and go through it step by step. You've already got BCLK stable at 180 which is a safe, conservative point for X58, so build around that. It'll probably be easier ramping up BCLK and the CPU multiplier without worrying too much about the RAM overclocking just yet. Since X58 ties the RAM multipliers to QPI and BCLK (unlike Sandybridge where DRAM can be set independently without having to worry about other multipliers), you'll want to isolate those values at a low RAM clock first, and then attempt to raise the memory clocks once you have a stable CPU/BCLK overclock set.

Since you have 6 RAM modules running, try reverting to just 3 modules in parallel when you're going through the guide. It's possible that your CPU's IMC is having trouble with 6 modules & overclocking.

If your board supports saving separate OC profiles, save one before clearing CMOS/reverting to stock. Worst-case, you recorded the values on the last page to work with.

What motherboard are you using by the way? I think it was listed at one point but no idea where that post is.

future ghost fucked around with this message at 06:57 on Feb 12, 2012

# ¿ Feb 12, 2012 06:54

future ghost: Dec 5, 2005; Gun Saliva

Druuge Fuel posted:

Great OP! I initially thought I was buying an awesome motherboard for OC'ing when it first came out -- as it turns out it's one of the worst. I've purchased a hyper 212+, additional ceiling fans for my case, but it still throttles down after prolonged use under ~full load. It boggled my mind for a long time until learning all about 4+1 power phases and un-sinked mosfets!! That knowledge will certainly factor into my next mobo buying decision.

This is me: http://www.asus.com/Motherboards/AMD_AM3/M4A87TDUSB3/

Relevant:
http://www.overclockers.com/forums/showthread.php?t=679903
http://tinyurl.com/7dggoqm

Ouch, yeah you're completely naked there.

If you want to attempt a fix on your own, grab some of these mosfet heatsinks:
http://www.sidewindercomputers.com/enmofocomohe.html
12 come in each pack, so one pack should be enough. They're tiny, and one goes on each mosfet. They don't look like they'd be that effective, but I use them for GPU VRM chips all the time (sidenote: gently caress you, Zotac).

And either (one of these will suffice, not both):
Thermal adhesive (you can get the non-"Premium" version if it's cheaper) :
http://www.newegg.com/Product/Product.aspx?Item=N82E16835100005

or Sekisui thermal tape (only get Sekisui. 3M thermal tape is easily-available but it is garbage) :
http://www.ebay.com/itm/Sekisui-5760-Dbl-Sided-Thermal-Tape-8inX-1in-4-strips-/180521089422?pt=LH_DefaultDomain_0&hash=item2a07e5398e

The adhesive might be easier to work with, although it would have a longer set-in time. If you're using the adhesive, DO NOT spill any on the board as it's conductive. Just use a tiny bit applied to the base of the heatsink. The tape isn't conductive, so just stick the heatsink on one side, cut it from the strip, and then apply to the board with a little [not too much] pressure. You can use a blowdryer to heat-cycle the heatsinks for a couple seconds to ensure they stay put, but it's not expressly necessary. I generally prefer tape as it's harder to gently caress up, and you can remount it if needed.

You'd want to remove the board from the case, and then either using the thermal tape or the adhesive you'd attach the heatsinks to the VRM/mosfet chips on the board (near the I/O plate).

This would at least ensure that the board's VRM's won't overheat, and hopefully stop it from throttling. No guarantees though, so blame the board-maker if it doesn't work. YMMV. You should strongly consider sinking the VRM's on your board if you're using a high-TDP CPU, as eventually the board will get unstable or stop booting if it's over-taxed for too long.

E: Just noticed that the overclockers.com page you linked mentions MOS-C1 heatsinks. Those are the ones I linked above, just with more detail on heatsink application.

future ghost fucked around with this message at 07:23 on Feb 12, 2012

# ¿ Feb 12, 2012 07:07

future ghost: Dec 5, 2005; Gun Saliva

Gorilla Salsa posted:

What should I set it to? There was a guide that you had linked me that suggested 15, but when I tried that, the system didn't boot.

That's odd - it should have just dropped the CPU clock to test BCLK limits. Did you revert to stock settings before you tried it?

quote:

The closest thing I could find was this:
code:
The computer has rebooted from a bugcheck.  The bugcheck was: 0x00000101

Usually this would indicate inadequate vcore, but 1.35V should be more than enough for a 920 at 3.6ghz, so there's not enough to go on right now.

quote:

What would happen if I tried to re-insert the other three after getting everything situated?

Assuming your overclock was stable by that point, you'd have more available RAM. The point of removing the sticks for now is to stabilize the OC & minimize stress on the CPU's memory controller. If it turns out that it works fine with 3 sticks but it isn't stable with 6, you might be better off getting 4x4gb RAM and selling your current kit. You'd lose triple-channel that way, but there wouldn't really be an appreciable performance difference with a dual-channel memory configuration and it may make overclocking easier.

# ¿ Feb 12, 2012 09:23

future ghost: Dec 5, 2005; Gun Saliva

Gorilla Salsa posted:

I ended up scaling back to 3.36 Ghz instead of 3.6. Load temps are around 65 with prime95's max heat setting. I'll leave Prime95 on overnight to make sure that it's stable enough. The vcore is at 1.25 I believe. I really hope this poo poo works out, because I'm kind of tired of dealing with this bullshit.

It figures I'd get the (apparently) solitary 920 that can't do 4Ghz on air.

3.4ghz is pretty low for 1.35V. Even with a C0 chip, you shouldn't really hit any limits until up to ~3.8ghz.

What cooler are you using? It really sounds like you've either got insufficient cooling or something's not mounted correctly.

# ¿ Feb 16, 2012 17:42

future ghost: Dec 5, 2005; Gun Saliva

Emanuel Yam posted:

It gets here tomorrow and I'm a total scrub when it comes to computers but ive been learning a bit in the last week reading this forum/system building stuff.

I want to know whether the OP was being facetious with the 'Set the CPU multiplier to 42, save and exit' advice? Because i dont really want to crank it too much before i really know what im doing but it seems crazy not to go for a near 1 ghz increase for so little effort.. do i need to adjust anything else in the BIOS? is it not that simple?

Make sure the PC works fine and that it is completely stable at stock before attempting it, but yes you can generally just leave all other BIOS settings on auto and change the multiplier for 4.2ghz.

# ¿ Feb 21, 2012 17:53

future ghost: Dec 5, 2005; Gun Saliva

SRQ posted:

I'm already running an "unsupported" processor (For one thing it was made 2 years after the bios date, and for another it's got a higher FSB) so I'm sure any P3 other then the server variants will work. I would love to get a Coppermine-T 1.2 in there.
I also managed to shove 768 megs of SD ram into it, which frankly shocked the hell out of me.
Also doesn't that mean that you could easily bump that multiplier up to 10, with no fuss?
E: Heh, that review says it supports 128 meg sticks max, I've got 3 256s in it. I didn't even know SDRAM came in 256 until I found these.

The reason I use this rig instead of just using something new, is that it's the same machine I first got in high school, so it has a bit of a nostalgia thing going on, plus it runs almost silent and I put an AWE64 in there for some old school PC Gaming bliss.

You wouldn't be able to bump the multiplier up on an OEM board AFAIK, however a 100mhz FSB chip should work. I have an old compaq P2 SFF box, and I shoved in a 100mhz FSB P3 slotted CPU for awhile. The one I had also listed 128mb SDRAM support but took 768mb fine. I'm fairly sure that a 440BX chipset should accommodate any slotted P3's really, you'd just need to find an adequate 100mhz chip.

# ¿ Feb 24, 2012 08:03

future ghost: Dec 5, 2005; Gun Saliva

22 Eargesplitten posted:

I've got an EVGA GTX460 SSC+, which comes at 850mhz core, 3900mhz effective ram speed. The EVGA GTX460 FTW runs at the exact same core speed, but has 4000mhz effective ram speed. Is it worth trying to overclock the ram a little bit, or would it be about the same level of uselessness as buying faster ram for my motherboard?

Not really. VRAM overclocking doesn't really make much of a difference in most cases. As long as the core speed is the same for both they should run fast enough for most tasks.

# ¿ Mar 12, 2012 23:54

future ghost: Dec 5, 2005; Gun Saliva

HalloKitty posted:

680s? Nah, to make use of that power you want to downgrade to 480 SLI.

Yeah, I have a case absolutely packed with hard drives, an unlocked Radeon 6950 and an overclocked 2500K, and it doesn't burn through all that much power. I think the Seasonic X-660 running it is basically the perfect size for it. I'm pretty sure my (don't laugh) Radeon 2900XT used more power.

Outside of the CPU, I have a similar system to yours, and it's about the same deal. I think the highest it drew with the kill-a-watt was around 380W last time I checked.

Still too much for the UPS I have though... I'm pretty sure the battery's going out on it, as The Witcher 2 triggered the overdraw alarm yesterday and it shut down. I need to pull the unit this weekend and see if ~~APC~~ Amazon has replacements available.

# ¿ Mar 28, 2012 17:25

future ghost: Dec 5, 2005; Gun Saliva

protobyss posted:

[edit] Resolved the issue, left the CPU Ratio in CPU Power Management to Auto and not 45.

I was able to run the Standard Intel Burn Test and the Very High Intel Burn Test and the highest CPU Temp was the third core reaching 71-73*C and the other cores fluctuating between 68-70*C.

Do these temperatures seem reasonable or should I dial it back because I don't necessarily want to be riding "the wall."

The chip will never get that hot in "normal" use, so those temperatures are fine for IBT. With maximum (or close to it) RAM settings on IBT, anything south of 80C is acceptable for a SB chip, although obviously you'll want to stay closer to the low-70's.

# ¿ Apr 5, 2012 17:41

future ghost: Dec 5, 2005; Gun Saliva

Blackclaw posted:

Should upgrading two generations to IVB and installing a 212 be enough for viable overclocking when I'm almost edging the limit as it is already?

If it's meant to primarily be a gaming platform, just get a better cooler for your CPU (either a 212+/EVO or a Thermalright Macho if you want higher-end) and overclock the 760. There's not much sense in doing a full platform upgrade unless you just want to switch things out, as your current CPU will be more than enough for quite awhile with overclocking and better cooling (at least for gaming).

E: Noticed your post in the parts megathread..
Put the 212+ on the 760. Overclock the 760. Upgrade the board & CPU some point after IVB as you won't need to before then.

future ghost fucked around with this message at 17:04 on Apr 6, 2012

# ¿ Apr 6, 2012 17:01

future ghost: Dec 5, 2005; Gun Saliva

Blackclaw posted:

Was my fan behaving properly under that load? Do I need to worry at all about that?

The fan was maxing-out around 2500rpm/70C so that's normal for the stock cooler. The 212+ will work wonders and allow you to overclock. Given that it'll be more than enough performance for gaming at 3.8ghz or so, I'd go with that option rather than a full rebuild. Other than that you should be good with basically any GPU depending on how much you want to spend for performance, as that CPU overclocked won't hold you back.

e: Have a question of my own. Anyone successfully using Afterburner with the CCC? Trying to decide if I should install the newest catalyst drivers to get powertune options, but I'd also like to overclock beyond what the CCC allows. Right now I'm stuck at 880/1350 on my 69*0 as pushing it any farther hits the TDP limit and the card starts throttling. I'm currently using the non-CCC catalyst drivers with Afterburner and it's working fine, but I'd like to push the card a bit more if the CCC won't conflict with AB. Cooling isn't an issue as I'm using an aftermarket kit.

future ghost fucked around with this message at 00:00 on Apr 7, 2012

# ¿ Apr 6, 2012 23:52

future ghost: Dec 5, 2005; Gun Saliva

I used to go for full painstaking CPU coverage with my hand in a plastic bag, but I just do the 'ricegrain in the center of the chip' method nowadays and temperatures are slightly better (with 70lbs compression on an HR-02, they better be). Also it's easier if you're lazy, and you don't have to worry about TIM overspill. As Factory Factory said, with the 212+ it has its own TIM pre-applied so just use that and don't mess with it.

Factory Factory posted:

Definitely install new drivers/CCC. Afterburner works fine with them and will override the CCC clock settings as long as it's running.

Thanks, that's what I was wondering about. +20% powertune isn't much, but I'll take it. :getin:

e: FF, if you're getting tired of the MLP avatar (I know I am) let me know and I'll buy you something else. Probably best to send me a pic/BB code though because I'd have no idea what to change it to beyond like a CPU die on fire or something.

future ghost fucked around with this message at 01:06 on Apr 7, 2012

# ¿ Apr 7, 2012 00:58

future ghost: Dec 5, 2005; Gun Saliva

grumperfish posted:

e: Have a question of my own. Anyone successfully using Afterburner with the CCC? Trying to decide if I should install the newest catalyst drivers to get powertune options, but I'd also like to overclock beyond what the CCC allows. Right now I'm stuck at 880/1350 on my 69*0 as pushing it any farther hits the TDP limit and the card starts throttling. I'm currently using the non-CCC catalyst drivers with Afterburner and it's working fine, but I'd like to push the card a bit more if the CCC won't conflict with AB. Cooling isn't an issue as I'm using an aftermarket kit.

Just an update to this - it was unnecessary to use the CCC with the latest Afterburner beta as the beta includes powertune controls now. Which is a good thing as I failed at getting the CCC and AB to play well together. :colbert:

future ghost fucked around with this message at 05:31 on Apr 8, 2012

# ¿ Apr 8, 2012 05:08

future ghost: Dec 5, 2005; Gun Saliva

Blackclaw posted:

I'm having trouble finding the definitive limits for Lynnfield (i5 760). I've got a Vcore limit of 1.4 (I believe), not too sure what the temperature limit is. Multipliers are unlocked but I don't know if I'm expected to play with the BCLK at all...

e: Apparently BCLK can affect a bunch of other parts of the computer thanks to QPI so I'll start with tweaking the multiplier. I'm not aware of the 'hard limit' on temperature though.

e: Turns out I can only change the multiplier to one above the default (22 x 133MHz = 2.93GHz). Looks like I'll be tweaking clocks!

I'd suggest ignoring the turbo (+1x) multiplier setting, as it will make overclocking your chip more difficult initially.

Temperature limits are similar to SB chips, in that you'll want to stay around 72C at highest (Prime95) whereas you can allow higher temperatures in max IntelBurnTest, up to ~80C at the maximum limit. It'll never get that hot during normal use, however it's fine to go up to that temperature as long as voltage isn't too crazy for testing-only. Technically you could take it up to 90C+, however you won't be pushing enough voltage through it even up to 1.4V for it to get that warm in IBT.

With your CPU, you'll be overclocking primarily via the BCLK, and RAM/QPI/etc speeds will be adjusted with multipliers. RAM should stay at or below the modules' rated speeds. You'll want to keep an eye on QPI/VTT voltage as it should be no more than .5V below RAM voltage (so if your RAM is 1.65V sticks, VTT voltage would need to be at least 1.15V - if you're using older 1.8V RAM sticks, consider replacing them, but you'd need 1.3V VTT voltage minimum). For nehalem/lynnfield chips, 1.2V to 1.3V is usually safe, although your chip may need higher voltage. I'd aim for 1.3V or lower as any increase in VTT voltage will increase CPU temperatures as well.

QPI speeds don't really need to be super-high as it won't increase performance, so going beyond ~6.4ghz is pointless (so a 32x multiplier for QPI at 200mhz BCLK = 6.4ghz which is more than enough bandwidth).

If your board supports uncore frequency changes, keep the uncore ratio at 2X RAM speed (so if you have 1600mhz RAM you'd need an uncore frequency of at least 3200mhz), but try to keep uncore at or below 3700mhz as it'll get unstable the higher you go beyond 3600mhz depending on the board. Uncore frequencies will likely limit you before your CPU unless your board allows for a high BCLK. If you're using 2000mhz RAM you're going to have a hard time running it at full speed, although there's few benefits from faster than 1600mhz, so don't worry about it and run it slower.

If you're really lucky with CPU voltage, you might be able to get away with 1.3V at your maximum stable overclock. If your board has LLC options, consider using them as it'll make your life a bit easier.

Check out this guide here for your chip:
http://www.techreaction.net/2010/09/07/3-step-overclocking-guide-lynnfield/

Read the first few pages to understand the steps, and when doing the BCLK isolation stage, try to get your BCLK as high as it'll go without getting unstable, then back it down a few mhz from that speed.

future ghost fucked around with this message at 03:54 on Apr 9, 2012

# ¿ Apr 9, 2012 03:47

future ghost: Dec 5, 2005; Gun Saliva

Blackclaw posted:

Thanks for the link, last successful stress test was on BCLK=160 (using IBT 'High' 10 times), now I'm getting BSODs and warnings from IBT at 170. I've got the CPU offset voltage at .025V. From the last OC megathread here apparently BSODs are indicative of a memory instability (which may be true as BCLK alters that too I believe) but I wouldn't know where to start making memory more stable. For the meantime I'm staying with BCLK=160 but is there anything I can or should do to push it further?

What was the BSOD code, and what voltage does that offset give you in BIOS? Make sure you've set your RAM speed & voltage manually, and that you've got uncore high enough for the RAM speed. Manually set your VTT/QPI voltage rather than leaving it on auto.

The BSOD code is important as windows7/vista will give specific ones like xx124 that indicate that CPU voltage is too low (particularly if you're getting IBT warnings). You'll want to have CPU voltage (with the offset) at 1.3V while you're testing BCLK so you can see where that runs you (you can dial it back down later but setting it artificially-high will let you test BCLK without having to constantly adjust voltage).

# ¿ Apr 9, 2012 03:56

future ghost: Dec 5, 2005; Gun Saliva

Posted about this awhile back, but this crazy dude on overclockers has been modding Voodoo cards and he just posted another one he's working on with custom watercooling:
http://www.overclockers.com/forums/showthread.php?t=702722

# ¿ Apr 11, 2012 23:29

future ghost: Dec 5, 2005; Gun Saliva

Endymion FRS MK1 posted:

I'm kinda new to overclocking here, and I'm wondering the consequences of overclocking my HD6950. I already have the additional shaders unlocked, and I would like to bump the clokcs and voltage up to 6970 levels. Will this damage my card in the long term? Its a Sapphire, so it is at least a good card.

For consequences: more heat and more noise with the reference 6950 cooler as long as you keep your overclocking reasonable. Overclocking to 6970 speeds is highly-unlikely to damage the card in the long or short term (maybe shortened lifespan in the very-long term, but by then you'd have needed to upgrade anyways).

As for the overclocking, it's fairly easy, especially if you're already unlocked. 880/1325 or 880/1350 on my card (XFX) only needed a small voltage bump to 1.1V which didn't add much heat with the reference cooler (using 950/1350 @ 1.2V now with better cooling though).

Download the MSI Afterburner beta and modify the Afterburner .ini to enable overclocking beyond the CCC per:
http://forums.overclockersclub.com/index.php?showtopic=182403

You can disable ULPS as they've mentioned but it's not strictly necessary with a single card. It may do nothing, or it might increase stability, so that step is skippable.

Install the AB beta and reboot when it asks you to do it. After that you'll want to play around with overclocking in AB. You should enable voltage modification in settings as you'll probably need to lock it at 1.1V to get the card up to 880mhz core clocks (keep voltage monitoring disabled if you're using HWiNFO though as it'll already give you a readout). Setting a manual fan profile in Afterburner is a good plan when overclocking (check the settings menu) as the stock fan profile will let the card get too hot and then ramp up the fan at 70C. You'll want the fan to go up in steps, so at 50C you'd want it running at ~medium speed then around 70C you'd want it running pretty high as you'd be nearing the card's temperature limits. The "K" button on the left side of Afterburner's menu will run Kombuster which, like furmark, will stress the card allowing you to test your clocks/fan profile. After you've got your fan profile & clocks set how you want them, save it as a numbered profile. You can also toggle the button at the bottom to apply the overclocking settings at boot.

Without increasing powertune levels you can reasonably get up to ~900mhz (although 840-880 is safer) at 1.1V-1.17V, but beyond that you'll be hitting the card's TDP limits. Increasing the powertune levels will allow you to go higher, although you'll also see alot more heat, which is problematic with the stock cooler as it'll mean the fan will be running at a higher speed more often. The VRAM will (safely) top out around 1400mhz on a 6950, but in my experience there's really no point to running it beyond 1350mhz as you'll see more benefits from core overclocking.

e: You may also want to grab HWiNFO64 to monitor temperatures (including VRM temperatures which AB won't give you) while you're setting up a fan profile in Afterburner. You can set HWiNFO to run in sensors-only mode in the "configure" menu and it'll just get you configurable temperature & voltage readouts.

future ghost fucked around with this message at 00:43 on Apr 12, 2012

# ¿ Apr 12, 2012 00:29

future ghost: Dec 5, 2005; Gun Saliva

Endymion FRS MK1 posted:

Oh ok, thanks. I already had the clocks and voltages saved as a profile in Sapphire Trixx though. Would you recommend using Afterburner over Trixx?

Also, I just want to be clear on this, as long as I keep everything in check, temps and voltage, this won't hurt my card permanently?

If you already have your clocks & volts saved in Trixx and you're familiar with it, just stick with Trixx. Either application will do essentially the same thing as far as overclocking the card is concerned. If it's stable in games and not artifacting or overheating, you should be good to go.

# ¿ Apr 12, 2012 03:36

future ghost: Dec 5, 2005; Gun Saliva

Endymion FRS MK1 posted:

I played with Trixx, and I have my 6950 running stable at 880/1325/1.14v. It runs Battlefield fine, temps stay under 60 and no hiccups. When I run Uningine Heaven though, sometimes (uncommonly) I get screen flashes during the benchmark. Settings for that are everything at highest except no AA. Is this bad?

Screen flashes are a little unusual but it doesn't sound like GPU artifacts. It's probably not anything to worry about if the card's working fine in games.

# ¿ Apr 12, 2012 18:46

future ghost: Dec 5, 2005; Gun Saliva

Haeleus posted:

I have an i7 2600k and an ASUS Sabertooth P67 board, with the turbo multiplier set to 43. All I want to know is if there is a point to adjust the TDP settings to have turbo running indefinitely as mentioned in the OP; is there some downside to having the CPU downclock itself to 1600MHz when in low use?

The chip will downclock itself to 1600mhz when it doesn't need full power, and it will increase speeds per-core dynamically as needed. Dynamic overclocking will not affect stability or maximum overclocking or anything really, so there's no downsides to it. Unless you somehow count using less power to be a downside :psyduck:

Increasing the TDP levels will prevent the CPU from throttling at higher load, so it's not a bad idea to raise it, but you don't need it running at max overclock 24/7.

Grab HWiNFO64 and check the Summary window, and you'll be able to see it actively switching per-core speeds & voltage.

# ¿ Apr 13, 2012 16:52

future ghost: Dec 5, 2005; Gun Saliva

Endymion FRS MK1 posted:

Ok, question. What keeps resetting my voltages? every so often I'll look at Trixx, and voltage has reset to 1.1v, but the clocks stay 880/1325. I just reload the profile and it stays that way. Could Catalyst be affecting this?

I noticed something similar when I was using Afterburner with the CCC loaded (before I reloaded the non-CCC drivers), where the GPU voltage was dropping back to 1.1V randomly.

Will your card not run 880mhz at 1.1V, or does it need the additional bump to 1.14V for stability?

# ¿ Apr 13, 2012 20:55

future ghost: Dec 5, 2005; Gun Saliva

Carecat posted:

My 2500K is sitting at 1.361V with a 47x. It gets to an eyebrow raising 1.396v and 78c peak in IBT on normal which makes me hesitate to try it on maximum.

Gigabyte boards will raise voltage slightly at load as some LLC is applied when using offset voltages above ~4.5ghz, although you should have the option to lower the LLC level if this is the case (not sure what board you're using so just throwing that out there). Are you setting vcore voltage manually or are you using offset voltage? What level of LLC have you set?

I'd be more concerned with the temperatures than the voltage (assuming it's vcore) as long as it doesn't pass 1.38V during normal load or prime95. 1.396V during short IBT tests is unlikely to do any real long-term damage, but it could be an issue during real-use load if it's sustained - you should of course aim for lower even in IBT though. 78C at normal IBT is pretty toasty - which heatsink are you using?

# ¿ May 2, 2012 04:38

future ghost: Dec 5, 2005; Gun Saliva

Carecat posted:

Motherboard is a Gigabyte Z77-D3H. Bought an overclocked bundle from Scan so I didn't set it up. I was looking at VID which was 1.36-1.38. The vcore is 1.428V

Kind of not surprised now that it rebooted itself before getting to bios around six times before it started today. Changed to the unclocked profile. Looked like everything was on auto except the multiplier and voltage.

CPU was manually set to 1.425V, CPU VTT at 1.1. BLCK is 100. Everything else is on auto. Not sure what I paid for other than a 24 hour stress test and a kamikaze vcore setting.

Play around with a mild overclock (~43x) and offset voltage (DVID) instead of a static manual vcore voltage. You'll need to experiment to find out what +.x DVID setting you need to add for the load vcore voltage you want. Try to keep vcore at or around 1.38V maximum, and you can give VTT a slight bump if necessary.

You want to use offset voltage (DVID) as it'll keep the C-states undervolting in play at idle, but still bump up vcore at load. Once you go above 4.4ghz+ you may need to reduce the LLC level.

# ¿ May 2, 2012 15:55

future ghost: Dec 5, 2005; Gun Saliva

Endymion FRS MK1 posted:

Ok, thanks guys. Now I just need to hunt around the settings and limit the voltage, unless you think 1.275 would be safe for a 24/7 overclock? You said under 1.3 should be good, and I have power saving on so it downclocks to 1.6 when I'm not doing anything intensive.

1.275v should be alright if temperatures are in check and it's max-IBT/Prime95-stable. You can probably fine-tune it for lower voltages and temperatures if you get the BIOS settings down.

# ¿ May 6, 2012 06:20

future ghost: Dec 5, 2005; Gun Saliva

dunkman posted:

I have an extra 80mm fan laying around, are you saying I should screw it to the "back" of the H60 radiator to get better temps? It's already screwed on to the one that comes with it on the back of my case.

Edit: Did it anyways, now OCCT tops out at 60 degrees. Not bad, but a bit noisier.

38-mm thick fans generally work better on restrictive radiators and heatsinks, so you could mod-in 2x 120x38mm fans for better performance. The H50/H60's were more restrictive than the later closed-loop designs by Corsair, so you'd see more benefits with a thicker fan than a faster fan, and the most benefit with dual 38m fans in a push-pull configuration.

This, but with 38mm-thick fans instead:

Sanyo Denki/Panoflo (NMB) fans are generally designed in that form-factor with the intention to hook them to a fan controller at any rate. You'd see lower noise and temperatures (than the stock fan and an additional 80mm) with two of them in push-pull, with the caveats that it would take more space due to the added thickness, and you'd want to get medium/high-specced fans and undervolt them on a controller for best results. SD fans have a gentle-typhoon-style noise profile, so they're barely audible at 5-7V. You could also just get 2x 120x25mm fans, but the tiny radiator on the H60 means you'd have to run them faster (more noise) to overcome the resistance on the radiator.

You'd have to get a little creative with the mounting solution. Zip-ties are easy but messy, although it wouldn't be that difficult to grab some wider bolts at your local hardware store as the H60 mounting bolts are a standard size.

future ghost fucked around with this message at 06:40 on May 13, 2012

# ¿ May 13, 2012 06:37

future ghost: Dec 5, 2005; Gun Saliva

For anyone with an ATI/AMD card running catalyst 12.3 or higher:

If you want to use MSI Afterburner (the lastest version is the non-beta -> Beta15 just expired today) with catalyst version 12.3 or higher, and you want to unlock the overclocking limits, you'll need to do some extra steps.

AMD removed a couple .dll files from the newer catalyst drivers, so the files that Afterburner looks for are no longer there. You'll need to add them in.

You need to save the .dll files from here and then extract both of them into Afterburner's root directory. I copied atipdl64.dll to \Windows\system32\ & atipdlxx.dll to \Windows\SysWOW64\ per the instructions, but the AB root folder should suffice.

After this you'll want to run AB once (allow it to reboot when asked) to get the correct configuration profile for your card. This first-run creates the /profiles/ folder in AB's root directory.

Once it's back in windows, you'll need to make sure that BOTH MSIAfterburner.cfg files have the following modifications (make sure AB is closed when you do this) :

[ATIADLHAL]
UnofficialOverclockingMode=1
UnofficialOverclockingEULA=I confirm that I am aware of unofficial overclocking limitations and fully understand that MSI will not provide me any support on it

The cfg files are located at:
1)
/MSI Afterburner/Profiles/MSIAfterburner.cfg
(you'll need to add the above section to the bottom of this file)
2)
/MSI Afterburner/MSIAfterburner.cfg
(add the above section to the lines in this file as they're already present)

After you make the above modifications, run afterburner and you can overclock beyond CCC limits and enable voltage control or setup fan profiles or whatever. Enter your clocks manually as the toggles are pretty sensitive.

This is the guru3d thread talking about the .dll change with 12.3 and higher:
http://forums.guru3d.com/showthread.php?t=359671

I've been playing around with the new AB version, and while there's not many changes from the latest beta, they added powertune options to the main window which is a nice feature if you want to pump more voltage into your card for some maxx-epeen-overclocking (aftermarket cooling whatup).

# ¿ May 15, 2012 01:22

future ghost: Dec 5, 2005; Gun Saliva

KillHour posted:

Noctua:

Airflow 110.3 / 92.3 m�/h
Airflow with U.L.N.A. 83,7 / 63,4 m�/h
Acoustical Noise 19,6 / 19,8 dB(A)
Acoustical Noise with U.L.N.A. 13,2 / 12,6 dB(A)

Thermalright HR-02:

Airflow: 0 m�/h
Acoustical Noise 0 dB(A) :colbert:

Technically only passive at idle :ssh:

# ¿ May 15, 2012 21:25

future ghost: Dec 5, 2005; Gun Saliva

Factory Factory posted:

Okay, here's the deal: everyone pick your favorite heatsink, then we draw a circle on the ground, stick some Delta fans on them, and play heatsink sumo.

Sadly (?) mine doesn't really benefit from faster fans as it's more whatever flows over is enough. Kinda like a modern Tuniq Tower that was specced for 1366 hexacore chips and became kinda ridiculous when running a 1155 CPU.

I actually had a 250CFM Delta monster that would move across the table when you kicked it on... I got rid of that one when it tried to remove my finger.
Rather quickly got over the ludicrous-speed class of fans after that happened.

# ¿ May 16, 2012 05:06

future ghost: Dec 5, 2005; Gun Saliva

Lars Krimi posted:

I tried to pull the side off the case which causes CPU temperatures to be about 10 degrees lower under stress test. Perhaps an extra case fan is in order?

Yep - if temperatures decrease when you remove the case side, you need to work on your case airflow.

# ¿ May 16, 2012 15:57

future ghost: Dec 5, 2005; Gun Saliva

movax posted:

Yep. I have the same cooler, just order a 1155/1156 adapter kit and you'll be a-ok. I don't recall the model # off-hand.

e: From FrozenCPU it was Thermalright LGA 1156 Bolt-Thru-Kit Rev B. (Ultra Series / MUX-120 / IFX-14 / Venomous X / Ultima 90 / Cogage True Spirit)

Here's a direct link to the Venomous-X bolt-through kit from Sidewinder:
http://www.sidewindercomputers.com/thvebtkinso1.html

And here's the one on FrozenCPU:
http://www.frozencpu.com/products/1...ue_Spirit_.html

The Ven-X style mounting kit is the same one that shipped with my HR-02. It's nice because you can tighten it for more pressure and you can re-mount the cooler without removing the motherboard from the case (note that this is tough when you're talking about the larger models). You may need a long philip's head screwdriver for the installation - I can't remember how wide the TRUE's are vs. the mounting kit's bolts.

# ¿ May 18, 2012 05:27

future ghost: Dec 5, 2005; Gun Saliva

A 2600K's already going to be ridiculously-fast at 4.6ghz.. Just for single-threaded tasks I get like a 7.7s 1M superpi run with mine for instance. So personally, given the headache involved I wouldn't bother switching for nearly-identical performance. Especially given the extra work involved with your small case layout.

# ¿ May 19, 2012 05:26

future ghost: Dec 5, 2005; Gun Saliva

Star War Sex Parrot posted:

3770k temps are okay at 4.2 during IntelBurnTest but I'm really disappointed since my 2600k got to 4.4 under the same conditions. 4.6 was even good for gaming, though too hot for IntelBurnTest.

I'm less than impressed with Ivy Bridge.

I almost wrote that you'd risk like a 4.2-4.3ghz max with the IB chip, but I decided against it when I thought it'd be incredibly-unlikely (I mean, 4.3 or so should be easy at least, right?). Guess I could've left that in.

Weird that it'd be so warm since the heat/power difference between a 2500K & 2600K isn't really that much, but there seems to be alot of things about the IB chips that are strange. To be expected with several new technologies in play at any rate.

I'm not sure I'd attempt the IHS removal though. Getting it off would be easy-enough, but IIRC you'd need to modify the cooler for better contact without crushing the core, and that's generally harder to pull off successfully. Even with a Thermalright cooler you'd have to be mm-exact with any shim you use, and it wouldn't give you much room for error.

# ¿ May 20, 2012 01:08

Adbot: ADBOT LOVES YOU

# ¿ Apr 28, 2024 04:34

future ghost: Dec 5, 2005; Gun Saliva

akadajet posted:

RealTemp seems to do the job as well. Speed Fan is telling me that my cores are at a frosty 13c.

I hope this is a joke, because that's patently impossible. Your CPU cores are not running at 55.4F unless you are posting from Antarctica.

e: What's likely going on with the monitoring programs is that the internal temperature offsets haven't been properly adjusted for the new chips yet. Same thing that happens with every new architecture.

future ghost fucked around with this message at 03:57 on May 21, 2012

# ¿ May 21, 2012 03:52

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > Overclocking Megathread 3.11 for Workgroups

«‹›5 »