To those of you that have overclocked do you mind answering one or all questions for me?
1. What device did you OC?
2. By how much did you OC?
3. How has it affected power draw? Both at idle and at full load?
4. Same question but with temperatures
5. Do you find it worth it?
1. 4690k
2. Now running at 4.5Ghz
3. Desktop and I don't pay the power bill so I don't care
4. Below 80 degrees at load
5. Yes, even just for peace of mind that when waiting around for some CPU intensive process to finish there's nothing I could do to make it faster.
>What device did you OC?
Every CPU I've owned for quite a long time, and most of the graphics cards
>By how much did you OC?
Varies drastically. From 10% to 50%.
>How has it affected power draw? Both at idle and at full load?
It always goes up. Generally the difference is negligible at idle and only noticeable at load.
>Same question but with temperatures
Same answer.
>Do you find it worth it?
Yes. If you don't like doing it as a hobby, though, then you should quit doing it since the gains are no longer large enough to be noticeable without benchmarks. It also doesn't save you money like it once did. If you do like doing it as a hobby though, there's no reason why you shouldn't.
My mobo has the ability to auto-OC my CPU but from my understanding it's typically more conservative. Should I just let my mobo do it for stability or is it worth doing trial and error?
any overclock that's too small to need manual stability testing is too small to bother doing, even if it's just "push the button"
>To those of you that have overclocked do you mind answering one or all questions for me?
>1. What device did you OC?
Cpu and gpu. I7-6709k/989ti classified
>2. By how much did you OC?
Stock boost on i7 was 4.2ghz. I got up to 4.7ghz boost on all cores at stock voltage. 980ti is up +750mhz on memory and hits 1550mhz core under boost. Up from 1420mhz stock.
>3. How has it affected power draw? Both at idle and at full
The cpu still dials down to under 700mjz during idle, and even overclocked, I'm still at stock voltage. So same draw.
The card uses maybe an additional 15w under full load? Only using +50mv
>load?
>4. Same question but with temperatures
Less than 1C in the card. Unchanged on cpu.
>5. Do you find it worth it?
Yes.
Alright so does that entail I crank up the clock rate some amount and test using the computer for crashing. If it doesn't crash crank it up if it does crank it down?
>i7 870
>2.93 to 4.17ghz
>not sure about idle, but power draw has gone up from 95w to over 200w
>temps have gone up a lot, needed a beefy cooler to keep it below 80 degrees.
>not quite sure how to answer this one, it wasn't bottlenecking at the time I did it. I only did it out of curiosity, to see how hard I could push this chip.
If I hadn't yet though, it would probably be worth it to overclock it now though, 7 years is a nice long life for any computer part.
>fx 6300
>50 dollar mobo
>4.3ghz
>had to turn up voltage a good bit in biosnto be stable, will not handle 4.4ghz
>runs hotter but with a shitty air color it never gets hot
It made a big difference, had to do it take make gta5 playable.
I have overclocked my gtx 970 a lot but its binned like shit and crashes in cs go so i stopped
For a CPU you want to run Prime95 (as many threads as you have cores, hyperthreading included) for 24 hours at a given speed and voltage. If it reports no errors then it's stable. If it reports any errors at all it is not.
Run about 12 hours of memtest86 if you're running your memory any faster or at any tighter timings than what's on the sticker.
Running prime95 for an excessively long time is such a bad meme.
I've had it fail after seventeen hours, which makes me glad I didn't call my machine stable at that speed after only one or three.
>not considering 17 hours on prime stable
overclocked a fx 8320 from stock to 4.4 ghz on all cores .
cvore @ 1.430 ~
it died after a few years.
well if it was stable prime wouldn't have barfed at all.
I knew a guy who used to do three 24-hour runs, one for each kind of test.
He sounds more ecen autistic than you
Might have been the motherboard voltage boosting the living piss out if the chip. Did you set proper LLC?
A single 24 hour run is excessive. 3 24hour runs is ridiculous. I usually run 2 20 minutes runs. Each different test. 20 minutes on blend to make sure it's decently stable and that the ram isn't spitting out error. Then 20 minutes on small fft run for max cpu stress and you'll know almost immediately if the cpu is unstable.
20 minutes is fine for finding out "is this the CPU's ceiling, or can I push it farther?" since you don't want to take two weeks of bumping up speed and voltage in tiny increments, and if it primes out within 20 minutes it's certainly unstable and you need to bump the voltage or back off a bit.
for when you've settled on where you're gonna plant your flag as your everyday this-is-what-I-run-it-at-all-the-time overclock though, two 20-minute tests is ridiculously inadequate for ensuring that it is actually stable. It's not hard to find settings that prime out only after about an hour or so.
Not to mention stability testing is supposed to be a worst-case test. "If it handles this with no errors, it'll handle anything". You're saying you never play a game or do anything CPU intensive for longer than 40 minutes? You aren't wondering if there's some marginal bit in the CPU that's mostly okay but can't quite take the heat or something, and that might or might not crash you at some point in the future just because it picked that moment to flake out?
Twelve hours is a minimum, 24 is not at all excessive. My friend's 3x24 test probably was though.
I thought you were talking 24 hour run(s) per test. Not your final "this is my forever clock speed" run. I thought you were talking when you're still bumping up incrementally to find the ceiling.
Yea I usually do a full 12 hour run when testing for the final clock. your friend's test is overkill unless he needs to have his PC locked at 100% full tilt for days on end as a standard occurrence.
no, no, I always meant for the final "this is it, y/n" run. fair enough then.
that was a standard occurrence for him since he was into distributed-computing projects, btw. I was too at the time but usually only did 24 hours anyway
1.) Xperia Play
2.) Original Clock: 1 GHz, New Clock: 2 GHz
3.) I set it so at idle it drops back down to 1 GHz Max and using governors it would then drop to 233 MHz at idle i guess, but during use, idk, I used an extended 3000mah battery so it lasted a great amount of time.
4.) It didn't get noticeably hotter than normal.
5.) It kind of was, emulating PSX games was kind of laggy but OC'ing literally stopped all that lag and made shit smooth, although I'm sure running at 1.8 GHz is probably fine too and that 2GHz is risky and overkill.
sabertooth 990fx
i did oc always auto it's between 20 -30% like this user says, i always get some good cpu cooling first and set to auto OC, auto oc werks for me since intel p4 and core 2.
>7970
>from 950/1300(?) to 1175/1765(ish)
>idle not at all, 100W more at full load
>ran at 90°C 24/7 for several months
>totally. Needed that high memory clock to mine Litecoins when it was still worth it
Can't remember the exact figures for the OC and mining speed anymore, but it ran at around 850KH/s of I remember correctly.