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

akadajet posted:

Uh... Yes. My point was that Speed Fan isn't giving me accurate readings, whereas RealTemp appears to be.

Oh wow, I'm sorry - I didn't even read speedfan in the initial post.

Running a bit of a fever at the moment, so please forgive my previous snideness :v:

# ¿ May 21, 2012 04:57

Adbot: ADBOT LOVES YOU

# ¿ May 12, 2024 12:31

future ghost: Dec 5, 2005; Gun Saliva

craig588 posted:

I decided to attach it as is, but It's a non issue now, the motherboard was DOA. I ordered a new one from Newegg this time instead of a weird Amazon 3rd party seller, hopefully I'll at least get a refund.

I had a 3.6ghz q6800 before and I didn't want the move to the 2500 to feel like a step down. The CPU is probably still good actually, the chipset fan failed and caused the motherboard to get so hot it burned the PCB. Maybe I'll see if I have any other 775 boards laying about...

I went from a G0 Q6600 @ 3.5ghz to a 2600K* and it's been all-around-amazing. Once you get good parts in and can overclock the 2500K, I can guarantee you won't be disappointed. It might be a good idea to use the Q6800 for awhile if you can to feel the full impact of the upgrade. You should consider an SSD as well if you want the most visible change.

*I actually used a 920 for awhile in-between the two and the 2600K eats it for lunch in single-threaded tasks.

# ¿ May 25, 2012 05:09

future ghost: Dec 5, 2005; Gun Saliva

Thom P. Tiers posted:

Just wondering if anyone can tell me if this is a safe overclock. I hit 72C towards the end of the 5 run burn (on high) which is what I am worried a bit about. My vcore seems to be fine at 1.30 though. Any input from someone who knows more about overclocking than me?

i5-2500k with hyper 212 with push pull fans overclocked to 4.2GHz.

The 72C figure is the CPU package. Look at the core temperatures instead. You're fine so far, and cores can get up to ~80C in max IBT before you really need to be concerned as you won't see temperatures that high normally.

# ¿ May 26, 2012 23:38

future ghost: Dec 5, 2005; Gun Saliva

ntrik posted:

Any thoughts on hyper threading?

On my i7 2600k, I find that temps lower by 5-6C at full load in stress tests with HT disabled and gives me room for a higher overclock.

HT on, 4.3 GHz
HT off, 4.5 GHz

My understanding is that HT only provides performance boost to tasks such as encoding which I do not do much of. Am I better off with HT on or off?

And if anyone is wondering, I got a great deal on the 2600k at the time which was priced equivalently to a 2500k.

If you know you'll never or very-rarely use it, turn HT off if you want the extra 200mhz.

# ¿ Jun 4, 2012 00:57

future ghost: Dec 5, 2005; Gun Saliva

Josh Lyman posted:

I don't know if this has been mentioned already, but there's a mistake in the OP. It advocated for having positive case pressure, which is correct, but then says you achieve this by having more exhaust fans than intake fans.

In any case, I came here for this question. Cooler Master Hyper 212 Plus is $20 AR at Newegg. My current CPU temps are 35C idle with 30C ambient and max out at 86C under Prime95, but realistically the most CPU-intensive stuff I do is play Diablo 3 and Starcraft 2 and Civ 5 (assuming H.264 video is being decoded by my 560 Ti). Is it worth the $20? Before you suggest I spend money on case fans, all 6 spots are occupied, albeit by lower dB/CFM 120mm fans.

What CPU and clocks? 80C+ is actually really high for Prime95 for an Intel chip, and way too high for an AMD CPU.

# ¿ Jun 16, 2012 07:44

future ghost: Dec 5, 2005; Gun Saliva

Alereon posted:

If you're not planning to overclock or otherwise run in a thermally stressful environment there is no reason to upgrade beyond the stock cooler. If you're using the stock TIM, replacing it with thermal paste (no need to buy Arctic Silver or other stuff) will improve temperatures noticeably and for free. You should also probably remove some of your case fans, six fans is pretty ridiculous and you're generating a lot of extra noise for not much benefit.

I would argue that a cheap aftermarket cooler (IE: 212+) is also a good purchase if you want lower noise levels at load, even if you do not plan to overclock. A user may want to overclock later anyways, and it's much easier just installing a $20 cooler at the beginning when the board's not installed rather than having to remove it for an install later on.

I run 6 case fans, although that's with 7 HDD's, high-end fans, and a weird full-tower case (although many mid-towers are approaching its size at this point). For most usage cases, 6 basic case fans are far too many.

# ¿ Jun 17, 2012 16:55

future ghost: Dec 5, 2005; Gun Saliva

Agreed posted:

You sure you're thinking of the right heat sink? The XP-120 was famous for _{giving me the biggest bone back in 2004, gently caress copper block arrangements, I wanted one of those so bad} being the first commercially available 120mm cooler ever. And it "only" has 5 heat pipes, too. Is that definitely the one you meant?

From his post he meant the TR Ultra-120 Extreme (TRUE). Some of the early rev.-A TRUE's had slightly concave/convex bases at the time, although they've since corrected their manufacturing. I have one with a slightly-warped base on the fileserver, but the base is hardly as big a deal as people made out, and standard TIM fills in any gaps. The earlier TRUE models benefit greatly from dual-fan configurations (the thickest & fastest fans you can strap on the better). The Rev. C TRUE was introduced to allow for lower-speed fans at the same performance level with a revised fin placement. So the earlier TRUE's were evolutions of the XP-120 design, and the later ones were meant to be a hybrid between the HR-0x and TRUE lines.

# ¿ Jun 19, 2012 16:24

future ghost: Dec 5, 2005; Gun Saliva

VorpalFish posted:

So it's summertime, at it looks as though the thermal performance of my overclock, which I set up last summer, has gotten quite a bit worse - Last year I was under 72C during IBT in ambient temperatures up to 95F, now anything 85f for ambient temperatures is pushing me up to 78C on my hottest cores. My case is relatively dust free.

I'm thinking maybe I should try reseating my heatsink and re-applying paste. I guess these things can dry out over a year? Other than that I'm out of ideas. In the meantime I've dropped my clockspeeds back to stock and undervolted the poo poo out of my 2500K.

TIM usually won't dry out that quickly. Did you increase the voltage at some point in the last year? You may want to reseat just to verify though.

# ¿ Jul 1, 2012 18:06

future ghost: Dec 5, 2005; Gun Saliva

si posted:

Did I tl;dr myself, or just no real suggestions? Ordered a couple of new case fans as my rear 120mm isn't blowing very well, so maybe putting in a front and a new rear will help my temps also. If that doesn't, I guess I'll pull the cpu cooler and re-apply the thermal compound and see if it does anything.

If my temps are actually normal, please stop my insanity.

"77-83-81-80" is alright during maximum IBT, and your chip doesn't seem to want to take higher clocks/voltage with your cooler, so it looks like you've found your limit. You can try the additional case fans and see if it gives you more room, but you're probably up against a wall there.

# ¿ Jul 9, 2012 15:48

future ghost: Dec 5, 2005; Gun Saliva

nrr posted:

So what sort of temps are acceptable running IBT on a 3570k? From what I'd read earlier in this thread, it looks like 72C is as high as you want to take a 3570k, but now I'm hearing it's cool if IBT pushes things a bit above that? What sort of temps are acceptable while running IBT then?

I'm not Intel, so I can't be blamed if your poo poo blows up, but I always cut temps some slack when running maximum IBT, as it's an artificial load that you'll never really see in standard usage scenarios. I let my (C2/SB) chips get closer to 80C during max IBT runs as long as prime95 smallFFT runs stay at 72C or below. Below 72C is better for Prime95 as it's a good idea to account for occasional higher ambient temperatures to avoid baking chips due to a freak heatwave or broken A/C unit.

# ¿ Jul 9, 2012 23:23

future ghost: Dec 5, 2005; Gun Saliva

nrr posted:

Yeah, that sounds fair enough. Ok, well I may as well sort this out once and for all, I'm getting 32-38C on my hottest core while idling with an ambient temp of 25C. 3570k, ASUS p8z77-m, Hyper Cooler 212+. Did I gently caress up the thermal paste application again, or are those sort of temps somewhat normal?

Last time around I used the pea method from the vid in the OP, although it's possible it may have shifted a little as I was trying to screw the brackets in. Those things never want to sit still.

Idle temperatures aren't really useful for determining proper heatsink application, although 32-38C is about what you'd expect if things were installed correctly. Load temps are a better way to judge however.

# ¿ Jul 12, 2012 16:33

future ghost: Dec 5, 2005; Gun Saliva

Liu posted:

Is an i5 760 that hovers around 80c under load something to worry about? I'm thinking it may be the cause of some hardlocks I've been getting.

80C is relatively high for a 760, although it's not at critical shutdown-level as far as I'm aware. Is that with the stock cooler, and which app are you using to check temperatures? Hardlocks could be due to various factors, including a bad or insufficient power supply.

# ¿ Aug 4, 2012 17:55

future ghost: Dec 5, 2005; Gun Saliva

It's leak-testing on a custom watercooling loop. The idea being that you run the pump with the PC unplugged/off for 12-24 hours to check for any slow leaks.

# ¿ Oct 19, 2012 21:54

future ghost: Dec 5, 2005; Gun Saliva

Imodium AD posted:

Why wouldn't you do this with the case oriented to let gravity bring the droplets down away from the boards? Maybe the resivoir needs to be oriented up?

I've seen pictures of it being done with the case tipped to the side before, so I assume it can be done that way. I haven't messed with PC watercooling in years though, so I don't know if reservoir orientation would affect anything or not.

Glen Goobersmooches posted:

This isn't a particularly bad idle range. The ambient temp of your computer room is always a more significant factor than the case (unless it's literally a tomb of dust).

I was talking with someone on mumble yesterday with a 45C outside case temperature. :stare:

This is why you don't skimp out on the bargain-basement Aspire/Apevia cases.

# ¿ Oct 19, 2012 22:56

future ghost: Dec 5, 2005; Gun Saliva

Beefheart posted:

I've been casually looking into overclocking my i5-2500k via Turbo Boost with my Gigabyte Z68XP-UD3P and I have a few questions:
By default are you limited to 3.4GHz via Turbo Boost with an i5-2500k with four cores running?
Is it normal that I can't change the Turbo Boost setting for 4-cores from [Auto] in the bios for these boards? Would updating to the latest UEFI possibly help?

You don't need to change that setting. Set your amp and wattage maxes to 300 and adjust the multiplier to overclock. You'll need to use offset voltages to keep idle down-clocking/down-voltage on the CPU. I use +.80 on my 2600K, but yours will probably be lower.

Make sure your RAM timings and voltage are set manually, and give the VTT voltage a tiny boost over stock.

future ghost fucked around with this message at 09:31 on Oct 28, 2012

# ¿ Oct 28, 2012 09:24

future ghost: Dec 5, 2005; Gun Saliva

It should drop the vcore and CPU clocks dynamically depending on what you're doing with the PC (check voltage/clocks with HWiNFO). If it's working on auto, verify using IntelBurnTest set to 5 runs on maximum to make sure the overclock is stable. If it's not changing clocks & voltages dynamically you'd need to test setting offset voltages to get a stable load vcore for the overclock.

# ¿ Oct 29, 2012 06:12

future ghost: Dec 5, 2005; Gun Saliva

Dogen posted:

Probably tough for a small shop to compete with all these pre-done closed loop systems that Corsair and the like put out

Performance-pcs and Sidewinder Computers still seem to be doing fine.

Dangerden stayed focused solely on custom water-cooling well after all-in-one coolers hit the mainstream and even the rare 'enthusiasts' moved to tower coolers. It didn't help that they got completely outmaneuvered in their own product space by both XSPC and Swiftech.

# ¿ Nov 2, 2012 21:37

future ghost: Dec 5, 2005; Gun Saliva

UndyingShadow posted:

What the gently caress? I know push/pull configs can get a degree or two better performance, but does all that bullshit have any realistic chance of affecting temps at all (other than being so god drat noisy that every human being and warm animal avoids the room entirely?)

The ones in the center are probably empty spacers to mimic the air-pressure from a 38mm fan. It generally makes things quieter, assuming they're tied to a rheobus, although a single decent 38mm fan would probably be a better use of space.

That's not to say that stunningly-insane 72mm-thick dual-fan parts don't exist however:
https://www.youtube.com/watch?v=IJOkZx_FjUY

# ¿ Nov 7, 2012 03:54

future ghost: Dec 5, 2005; Gun Saliva

Factory Factory posted:

That is brilliant. Next up, combine a Raspberry Pi and a number of those to create a computer whose cooling system also hovers it.

I think I finally found a use for the 250CFM Deltas someone gave me..
https://www.youtube.com/watch?v=qZcCuXDBuoY

# ¿ Nov 7, 2012 23:07

future ghost: Dec 5, 2005; Gun Saliva

Tres Burritos posted:

Also, I keep reading that you shouldn't do this because condensation is an issue. Is he combating this by just slapping the cooler straight on the CPU so that it doesn't get cool enough to form condensation? I mean, he gets like 10 degrees Celsius so I imagine that would be pretty cool anyways, right?

Either way, really interesting read.

You'd have to add insulation to prevent condensation for sub-ambient cooling, and it's generally not a 24/7 thing. Most of the people running things like LN2/LHE, staged phase-change units, or water chillers will be doing them for benching runs only.

TECs like the one the guy posted are inefficient as hell, and they dump a ton of heat, requiring a (second) separate cooling solution.

With modern chips none of that extreme-end cooling poo poo is needed to overclock given high-efficiency aircooling, and watercooling is relegated to vanity or space considerations.

Dogen posted:

Right, but I meant in this specific case where you're just screwing two 120mm fans together and none of that is taken into account.

The 'fan' closest to the heatsink is probably an empty spacer. I'd be surprised if there's actually anything in it.

At any rate the noise from that 295 would drown out the fans on the heatsink.

# ¿ Nov 10, 2012 21:19

future ghost: Dec 5, 2005; Gun Saliva

Destro posted:

I have a Radeon HD 6950, if I use MSI Afterburner to OC it as well as set the voltages do I still need to turn the power control up to +20%, or does setting the voltages in Afterburner take care of that? On that note do I even need to have the graphics overdrive option ticked in Catalyst Control Center if I am using Afterburner?

You might need the overdrive option enabled in CCC to overclock (I don't know if this is accurate though as I use the non-CCC drivers). You have to turn power control up to +20%, especially with overvoltage, as the card will throttle once it hits TDP regardless of cooling.

If you want to overclock beyond overdrive limits you'll probably need to follow the instructions on this page.

# ¿ Dec 2, 2012 00:55

future ghost: Dec 5, 2005; Gun Saliva

Destro posted:

What do you mean by throttle? I checked the OP and it is mentioned, but not defined.

The 6950 has a TDP limit (of ~200W) so once it hits TDP through overclocking and/or overvoltage it'll throttle speeds back down, leading to erratic performance. Bumping up the power limits to +20% gives you more headroom for overclocking.

# ¿ Dec 2, 2012 20:52

future ghost: Dec 5, 2005; Gun Saliva

I just dropped the Intel binary files into the folder and replaced whatever it asked. :v:

# ¿ Dec 15, 2012 09:31

future ghost: Dec 5, 2005; Gun Saliva

icantfindaname posted:

For Ivy Bridge on Asus uefi bios, where is the setting to turn off turbo boost overvolting? I'm completely lost. I managed to find the TDP limit settings. All I want the chip to do is to run at the overclocked speed 100% of the time and not do the reverse turbo boost underclocking thing, and also to have static voltages. Am I on the right track?

Why? Other than testing, there's no reason whatsoever to do this. The underclocking/undervolting is completely dynamic and has no impact on stability, so you might as well save on power and heat production (unless you're trying to use it as a room heater or something).

# ¿ Dec 16, 2012 09:25

future ghost: Dec 5, 2005; Gun Saliva

Naan Bread posted:

It's kind of amazing how manufacturer-provided tools inevitably fail to read settings from their own boards.

# ¿ Dec 21, 2012 19:49

future ghost: Dec 5, 2005; Gun Saliva

El Grillo posted:

Holy crap, in the UK I'm looking at �100 for refurbished Panaflos. Is that normal???

e: so it seems all retail suppliers of Panaflo fans in the UK are out of business or don't stock them anymore.
I take it there're no other fan maker's who've done anything similar to Panaflo's 'Hydrowave bearing' design?

Scythe's Gentle Typhoon (Nidec Servo) line are fairly close to Panoflos.

If you can locate any San Ace/Sanyo Denki/Nidec Beta (basically anything made by Nidec)/or NMB-MAT fans in the UK they would also work well. Most of these run really nicely on fan controllers. Avoid Deltas.

GreenBuckanneer posted:

Why it is that some people can get higher stable frequencies out of their Ivy Bridge than others? What are the variables?

Depends mainly on the CPU. What motherboard are you using? What is your current CPU voltage setting?

# ¿ Feb 6, 2013 19:51

future ghost: Dec 5, 2005; Gun Saliva

GreenBuckanneer posted:

CPU-z reports anywhere from around 1-1.2v volts or so. I have a Z77A-GD55 MSI board.

As long as you stay below "safe" Ivybridge limits for voltage and temperatures look ok, feel free to push up the overclock. It may be that your board just isn't capable of higher overclocks, although it has a fairly-robust vrm section so it should be fine for higher clocks. You might need to play with your BIOS settings, or you might've just gotten unlucky with the chip.

Your only other recourse is to violate your warranty and replace the internal TIM with better stuff, although 4.2ghz+ is still plenty fast for anything current, other than video encoding or maybe Planetside2.

MSI Afterburner chat:

So I was trying to add the new Catalyst 13.2 beta drivers (for Crysis3 performance improvements), and I ran into an issue with unlocking voltage/clock limits with MSI Afterburner. The old method to bypass overclocking limits no longer works, however if you want to go beyond the "approved" settings, the new way is much easier, and it works for both AMD and Nvidia:

1. Install MSI Afterburner.
2. Run afterburner at least once (and reboot if needed).
3. Modify the Afterburner shortcut so it has /xcl on the command line.
4. Run that shortcut, and it will tell you it's increased your overclock range and ask to reboot.
5. Reboot.
6. Remove /xcl from the Afterburner shortcut.
7. Now you can enable upper limits for clockspeeds in Afterburner with the latest Catalyst drivers, without having to deal with flickering windows (if you're still seeing these on an AMD card, you may need to disable ULPS via regedit - Nvidia users won't need to do this). Powerplay remains active with this method, so you won't lose power-saving.

Most people probably already figured this out, but I figured I'd try to save someone else the trouble of doing this the old/hard way.

The /xcl tweak convinces Afterburner to broaden the range of "approved" clocks rather than bypassing the BIOS clock ranges, and it apparently tells the CCC to allow the higher limits as well (I don't use the CCC so I can't test it). Please note that you can fry your card with extreme clock increases, so don't go overboard, and only use either the CCC or Afterburner to overclock, not both.

future ghost fucked around with this message at 02:20 on Feb 8, 2013

# ¿ Feb 8, 2013 02:16

future ghost: Dec 5, 2005; Gun Saliva

SocketSeven posted:

Then it is way to much to re-seat that copper monstrosity. I guess it works though, Intel burn test and prime95 show it stable at 4.4ghz, with a maximum temp of 75c.

There are a lot of apps that can disable Turbo mode on the i7, but I can't find one that looks at CPU temperature and disables it when it gets above a certain temperature for a period of time. You'd think thermal control would be the obvious application of a feature that forces clock speeds down to stock.

If it's at stock clocks it'll likely stay below 70C for the most part, so I wouldn't worry about it.

# ¿ Feb 12, 2013 03:07

future ghost: Dec 5, 2005; Gun Saliva

SocketSeven posted:

It would be nice to have the control is all. Thermal protection is all automated.

As stated, you probably have user-configurable thermal limits that you can set in BIOS.

On the other hand, as long as voltage is in check and the CPU's not hitting 80C regularly, it's not really worth worrying about. A couple extra C won't make any difference in the CPU's longevity over the long term.

# ¿ Feb 12, 2013 21:21

future ghost: Dec 5, 2005; Gun Saliva

Trilin posted:

Edit: I dropped it down to 3.6 @ 1.4 vcore and I feel a lot better about it. I could probably downvolt it further but I just got done with an hour long Linpack to make sure it's in a good spot and I think I've had enough tests and crashes for one night. I can't justify the massive spike in power demand just to squeeze that much more out of my CPU.

200mhz isn't worth 1.48V, so it's better to leave it.

Regarding overclocking an i5/i7, it won't make a difference currently other than specific applications (PS2, MWO, or video encoding), but it'll probably extend the useful life of your CPU considerably. I ran a Q6600 at 3.6ghz all the way through to Sandybridge, and it only started to feel slow by the very end. Might as well get a couple more years out of it if possible.

Kilazar, what's your current FSB speed and motherboard voltages? What's your RAM model and current clocks/timings/voltages? What CPU cooler are you using, as it's possible that it's throttling due to heat (although unlikely), and you can check temperatures with HWiNFO.

future ghost fucked around with this message at 06:30 on Feb 14, 2013

# ¿ Feb 14, 2013 06:27

future ghost: Dec 5, 2005; Gun Saliva

Imapanda posted:

Is there any applications I could download similar to ATI Tray Tools, but for Nvidia cards?

Or more to the point, Is there any way I can increase my Maximum Pre-rendered Frames beyond the 4 that the NVIDIA Control Panel gives?

There's not really a comprehensive tweaking application for videocards now that resembles ATT. The closest combination would be MSI Afterburner combined with the Nvidia control panel.

If the feature you're talking about is quad-buffering, there's no way to increase that as far as I know. Your only option is the Nvidia control panel.

Kilazar posted:

Information as follows

Proc - Q8300 2.5Ghz, clocked at 3Ghz 1.2 volts Stock cooling.
Mobo - Asus P5Q-VM
Mem - http://www.amazon.com/gp/product/B0027P9C9G/ref=oh_details_o00_s00_i00 BIOS is set to auto for the mem timings/management. If I have the voltage modified (I cannot remember at the moment cause I have made so many changes) the voltage would be at 1.8

FSB is set to 400.

AI clock twister turned to moderate.

Set your RAM timings manually to 5-5-5-18, and set RAM voltage to just over 1.8V (but below 1.9V).

The reason I asked about motherboard voltages is because a 400mhz FSB might need a voltage boost on the northbridge. There's not really anything you can do about the FSB since the CPU multiplier is so low, and pushing beyond 400mhz FSB is not advisable for stability.

You're limited in CPU vcore due to cooling, so you can't really go any higher there, although I'd suggest pushing vcore beyond 1.2V if possible given temperatures.

I don't know what AI clock twister is.

future ghost fucked around with this message at 17:05 on Feb 14, 2013

# ¿ Feb 14, 2013 16:57

future ghost: Dec 5, 2005; Gun Saliva

Dogen posted:

http://www.anandtech.com/show/6716/closing-the-loop-ii-new-liquid-coolers-from-corsair-and-swiftech/6

The H220 is a new dual fan design, as opposed to the H20-H220.

The pump is louder, but more effective, and also you can turn it down via PWM apparently...

edit: That anandtech review has it showing 30dba under both idle and load, also

Other than pump noise, you could always strap on a push-pull fan setup with something like this. Most of the "silent" watercooling setups I've seen have usually been managed with 2+ 38mm San Ace fans in a push-pull setup, since they maintain air pressure for radiators (while being quiet) at 5-7v without needing to run full-tilt all day.

The key advantage of the hybrid accelero card cooler that I can see (~~that almost makes it worth the cost~~*) is the active VRM cooling. Liquid VRM cooling is pretty great if you want to bypass power throttling on the card and overclock it to hell and back compared to what you can get out of passive heatsinks. The 120mm radiator is a bit restricting though, and the 350W heat rating on that isn't really much higher than the air-cooler versions, ~~at least not high enough to justify the added expense.~~*

Although given the pump and risk of leaks, I don't see why you wouldn't just buy whichever accelero air-cooler fits your card and strap that on instead, assuming you have the room (crossfire probably wouldn't work). I have a triple-fan model and I can't hear mine at all .5m away even with it set to 100%. At least with the universal versions you can probably re-use it later if you get another reference-PCB card, and if not they generally hold their value for quite awhile for resale on eBay/forums.

*edit: Nevermind the cost argument above. Didn't realize it was available at newegg. For 100bux the hybrid model is actually pretty compelling:
http://www.newegg.com/Product/Product.aspx?Item=N82E16835186067

future ghost fucked around with this message at 21:58 on Feb 17, 2013

# ¿ Feb 17, 2013 21:49

future ghost: Dec 5, 2005; Gun Saliva

SocketSeven posted:

Already running the latest Bios.

I'm resigned to let it run full tilt all the time at this point. Even with EIST and C1E disabled, it still drops the CPU multi to 16 at idle, it just jumps immediately to full speed the moment any process seems to do anything.

Another side effect, the PCIe Link states no longer adjust themselves according to load and neither do the video cards. I was not expecting that, but I only skimmed the docs on what C1E and EIST can do.

Turning the options off doesn't even raise core temps any significant amount, and looking over at my kilowatt, My rig isn't even pulling that much more power at idle.

I wonder how much power all this adjusting core multi's and link states really saves on a desktop.

It sounds like you have your OS power settings set to low-power or balanced. I leave power-saving options on in BIOS without issues, however I have manually-set power options configured in Windows, based on the high-performance profile. Try changing the OS options to high-performance, leaving EIST/C1E engaged. Make sure you raise any power limits in BIOS (I set CPU Amps/Watts to 300 each in my BIOS as an override).

Early X58 chipset revisions had some weird power bugs, but anything P67 or later should handle BIOS power savings without problems.

# ¿ Feb 20, 2013 17:18

future ghost: Dec 5, 2005; Gun Saliva

Femur posted:

As a matter of fact, any turbo charge setting above .200 will bring the CPU-ID numbers to 1.5ish Volts; so I don't even know why there are so many steps if it doesn't even matter?

Sorry to dogpile you on the CPU voltage thing, but there's a reason why the specific overclocking guide was picked in the OP, since it used conservative choices for CPU vcore (also I know the guy who wrote it). The only time you'd ever go upwards of 1.4V intentionally on an SB chip/IB (if you're crazy) is if you're a professional reviewer or forums benchmarker and you're intentionally testing the upper limits of a chip's clocks without caring about longevity. 1.5V is a terrible idea and that guide you linked makes alot of bad assumptions and shortcuts, particularly when it comes to voltage and temperature min/max levels. He completely bypasses any requirements to keep VTT voltage with .5v from RAM voltage, and about the only thing the guide does right is recommending not to touch SA voltage levels.

tarbrush posted:

Welp, it didn't like that. Bumped it 50mhz and had a lockup on MWO almost immediately. Moving it down to a 25Mhz boost and trying again.

Which specific 4870 is it? Some of them had weaker VRM sections and couldn't really tolerate overclocking well. 4870's and 4890's generally ran very warm unless they had custom manufacturer or aftermarket coolers.

Also MWO depends heavily on your CPU, more than most games. What cooler do you have on your 920? Overclocking it should give you a nice performance boost in the game, but it'll depend on what heatsink you're running. Also, which power supply do you have?

future ghost fucked around with this message at 20:03 on Feb 23, 2013

# ¿ Feb 23, 2013 19:56

future ghost: Dec 5, 2005; Gun Saliva

tarbrush posted:

Sapphire HD 4870 1024 is the best I can say. I threw the box out ages ago. I have a Coolermaster Realpower Pro 850W (don't laugh, it's an old computer and I was young and foolish and paranoid about underspending on the PSU when I got it) and a Thermalright Ultra 120 Extreme cooler, so my hardware should present no OCing issues. It's just a matter of having the patience to figure out how one overclocks and how far to push it.

Powersupply isn't great honestly, but it should do for your sytem assuming it isn't too old (since they came out in 2008). When did you buy it?

If the PSU checks out, you can use the Nehalem guide located here to overclock the 920. I'd shoot for ~3.6 to 3.8ghz for a good boost in MWO, with voltage limits depending on whether you have a 'C' or 'D' version chip (check CPU-z). Other than overclocking the 4870 you're not going to get much more out of the videocard, which without seeing the VRM section or the model # (check the physical card inside your case) we can't really know whether that'd work well or not. With MWO, for whatever reasons setting low settings won't really net you better performance than setting a mix of medium or high settings, but CPU overclocking will make a significant difference in your game performance.

# ¿ Feb 23, 2013 21:22

future ghost: Dec 5, 2005; Gun Saliva

tarbrush posted:

Bought it early 2009, so it's pretty old. I guess I'd like to be able to squeeze another 18 months or so out of it before I have to start from scratch. I have no trouble running MWO, GW2 and things of that ilk on medium to high settings, I'd just like a little extra prettiness.

Is there any reason not to use the overclock utility that the guide you linked recommends? 15% boost gets me to 3Gz.

Assuming the voltage isn't too high it's not terrible, but you'll get higher overclocks with less vcore if you do it manually in BIOS. If you have a D0 chip 3.9ghz to 4.1ghz should be doable with your heatsink, although 3.6ghz to 3.8ghz is going to be an easier target.

2009 purchase date isn't awful for the PSU, although I'd suggest replacing it within 8-10 months at the latest if you can.

Most of the MWO guys I've talked to with socket-1156/1366 boards saw significant improvements in the game with 3.6ghz+ CPU overclocks on the same mid-range videocards. If you're running 1080 you'll probably want to look into a videocard replacement at some point (1680 or lower isn't worth doing yet), but I'd start with the CPU overclock.

future ghost fucked around with this message at 23:28 on Feb 23, 2013

# ¿ Feb 23, 2013 23:25

future ghost: Dec 5, 2005; Gun Saliva

tarbrush posted:

Yeah. It's just a matter of figuring out how to do it and setting aside a day to gently caress around with it. Next question I suppose is how do I update my BIOS without a floppy drive?

You can usually use a USB for that. Check google for the HP USB tool, and it'll set you up with a DOS boot disk. From there you would do your board's normal upgrade using a copy of the BIOS. For gigabyte boards you would install it through the BIOS instead of using a disk, but I'm not sure what options your board will have. Check the manufacturer site for specific BIOS instructions.

# ¿ Feb 24, 2013 01:28

future ghost: Dec 5, 2005; Gun Saliva

spouse posted:

I know this is probably a dumb question, but I'm pretty new to this whole thing. I just ordered all my parts for the first computer I've built (for myself) in 12 years. I got an i5-3570k, a gigabyte z77 of some sort (don't remember model number), and a Sapphire HD 7850 2GB OC edition. I know the i5 is made for overclocking, and I assume I can do something mild with the 7850. Is it going to be a pain in the rear end to switch from the stock cooler on the CPU to a 212 evo or something in the future (not planning to OC just yet, seeing as how my budget is maxed out just using the stock cooler)? I haven't touched thermal paste since I was 13...

edit: I'm using a 520w power supply. Is that going to be enough to OC at all?

As said, it's a pain to switch later on as you'll have to remove the motherboard from the case, so it's better to grab a cooler now.

Depends on the power supply model, since 520W could be fine, or it could be awful depending on who made it.

# ¿ Feb 24, 2013 20:35

future ghost: Dec 5, 2005; Gun Saliva

Naan Bread posted:

Holy poo poo 6.6GHz? How is that possible to have 1.9V and not have the chip explode?

Liquid nitrogen. After a run like that and results verification, the chip's just about useless though.

# ¿ Mar 1, 2013 20:44

Adbot: ADBOT LOVES YOU

# ¿ May 12, 2024 12:31

future ghost: Dec 5, 2005; Gun Saliva

Factory Factory posted:

poo poo yo, it's a fuckin' air cooling throwdown at AnandTech. The CM Hyper 212 EVO and Noctua NH-D14 get a test in earnest, alongside the NH-L12 (low-profile), NH-L9i (mini), SilverStone Heligon HE01, and be quiet! Dark Rock Pro 2. Does the common wisdom about air vs. closed-loop liquid cooling hold up?!?!!!

NO

The CM Hyper 212 EVO is still a price/performance beast, even edging out the much larger Heligon HE01, and the air coolers in general can offer some really buff performance, but the current gen of closed-loop liquid coolers are pretty darn competitive on noise and cooling performance.

Price/performance is a little wackier, but really that's because the 212 EVO is just on a completely different level than everything else.

That's a neat article. The H80-tier of liquid coolers still doesn't look that impressive, although those NZXT coolers are interesting. Shame that they're still outperformed when it comes to noise - not sure if it's the restrictive tubing or the radiators doing that though. Corsair's higher-end closed-loop coolers look like they'd be really nice for an ITX build however.

I finally got around to using a 212+ the other day that I had in storage. It was actually really easy to install, although running Thermalright heatsinks for years probably contributed to that. It's a shame they didn't have any Thermalright heatsinks to test against though, since most of them compare with the higher-end Noctua models (plus I like graphs).

# ¿ Mar 15, 2013 00:29

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

«‹›5 »