Learning to OC... is it wise?

hello folks.

i started reading about overclocking and i am pretty interested in it. though i still have my doubts about it, and so i ask your opinion on a few of my questions. i hope you lads (i don't expect any girls in here, though correct me if i'm wrong :p) can help me.

so, first of all, my system specs:

motherboard: Asus P6T-SE
PSU: 550W (no idea what brand... it says "Advance" on the side.)
CPU: Core i7 920 Bloomfield (Stock Cooling)
Videocard: ATI Radeon HD 5850 (stock cooling)
no-name case

if i forgot anything you'd like to know, please say so :)

well, currently i just want to overclock my videocard, because i've read that it is the most reliable component and least likely to break when overclocking. maybe later on in my overclocking 'career' i'd like to OC my processor too, when i've collected some knowledge about the subject. when i am going to do that, i'll get a Corsair H70 cooler, because stock cooling sucks (85*C in 2 minutes of Prime95! luckily i don't reach that kind of temperatures when playing games, not even in Crysis :)).

anyway, the videocard might be the least likely component to break, but that doesn't tell me anything. I know i will lose my warranty when OCing the card, but can AMD detect an overclocked card when they get one in that's broken for some reason?
and how likely is it to break? i have had this PC now for about half a year, so is it safe for me to try and overclock my card? because i don't have any experience with overclocking or anything related.

that said, this is what i've already discovered about my card:

i've downloaded the programs it listed there and screwed around with the fan control (i couldn't resist, lol). from that i concluded that i can keep the card at 68*C with the fan speed at 70%.

the guide seems pretty clear to me, and i'd like to try it, up until the point of "over-volting". though i really don't have any idea of the risks involved. is it safe enough for me to try?

thanks in advance, your gaming-enthusiast Peter.

edit: i don't have a stock 5850, i have a "PCS+ Powercolor" 5850
39 answers Last reply
More about learning wise
  1. I am certainly not an authority on overclocking.
    With that in mind, check out this page:

    Also, as far as increasing voltage, you want to do that slowly and patiently, raising it in very small increments.
    After each small increase (or any change for that matter), run 3dmark stress tests and other stress tests suitable to
    your situation. If you don't get crashes or strange obvious stability problems, you should be alright.
    I'm willing to bet that you can find more specific info for overclocking that card in particular if you search around diligently enough.
    (Maybe I sound dumb because it might turn out that you can't o.c. that card at all- I just don't have any experience with it)

    Keep in mind that what I've written are simply general guidelines, and unfortunately I couldn't be of more help.
    Good luck to you.
  2. Depends which 550 watt PSU.

    GA-EP45-UD3P | Q9550 OC'd to 3.6 GHz (425 MHz X 8.5) C3 stepping :( | Corsair 750TX
    GA-EP45-UD3L | Q6600 OC'd to 3.6 GHz (400 MHz X 9) | Antec 550 TP3
    GA-EP35-DS3P | E7500 OC'd to 4.1 GHz (373 MHz X 11) | Corsair 550VX
    GA-G41M-ES2L | E6500 OC'd to 3.66 GHz (333 MHz X 11) | Corsair 400CX

    A good 550 watt PSU will power just about any CPU with any single GPU.

    I do not OC video cards. I figure that GPU's run hot enough already.
  3. OCing a video card is easy, and most 5850s can get pretty good improvements. IMO, you can try MSI Afterburner (even without an MSI card). There should be a fair bit of info on it. To play it safe you can OC and test it without changing the voltage (it's possible your card doesn't even support voltage control). My 5850s run 870mhz core, 1200mhz memory without any voltage change. Temps go up a bit, but you can also set custom fan profiles to keep it cool.

    And no, if you RMA a broken video card you had overclocked, the only way they can tell is if you BIOS flashed it with the new speeds like I did. ;)

    CPU OCing is a whole other ball game. Defintely worth doing a lot of reading before getting into it.

    Also 85C... at stock speeds? Sounds like you need to buy some fans for your case. Stock cooling isn't that good but it should keep it at a reasonable temp at least. Tho, ambient temps matter a lot. If your room temp is like 28 then maybe 85C isn't so bad lol, but all the more reason to buy aftermarket cooling anyway.
  4. ati drivers have an overclocking utility built in there drivers under performance -- amd overdrive it even has an auto overclocking called auto tune just don't raise anything more then 5 mhz at a time then check to see if there is any corruption memory overhaeting or reached its limit back rown 5 mhz same with gpu need a benchmark utility program to test it with be carefull if anything bad does happen shut off computer ?
  5. wolfram23 said:
    (removed this bit to shorten my post a bit :p)

    CPU OCing is a whole other ball game. Defintely worth doing a lot of reading before getting into it.

    Also 85C... at stock speeds? Sounds like you need to buy some fans for your case. Stock cooling isn't that good but it should keep it at a reasonable temp at least. Tho, ambient temps matter a lot. If your room temp is like 28 then maybe 85C isn't so bad lol, but all the more reason to buy aftermarket cooling anyway.

    useful info! thanks! ^^
    and i know about that CPU OC'ing. luckily when playing any games my CPU doesn't get over 70*C, so that's within limits i guess. and i'm not going to OC my CPU just yet. maybe after i get some experience :p. though that pc already heats my entire room, when i want it warmer in here i just play crysis for 30 mins, lol. my room temp is usually 20*C - 22*C. and my awesome case-temp-LCD-display doesn't usually go hotter than 25*C.

    anyway, i am worried about my GPU temps. because at 70% fan speed my GPU already gets up to 68-70*C. what is the maximum i should want? how hot does your GPU get? could you please give me some more info? :)
  6. To answer your last question, your GPU temps are fine. As uriahk referenced, 85 degrees is acceptable for normal use, 75 degrees for maximum overclock. Here is another reference, if you want to see more.

    Have you downloaded the AMD CCC tool yet? This is what I use, and it seems almost too easy. Plus there are a lot of other options offered here for GPU management.

    As for your power supply, 550 watts is minimum I would use. If you look at the first link above, it also shows power consumption under load for the graphics card at 305 watts. Add the 125 watts for the CPU and you are at a comfortable operating range for the PSU. Power supplies will burn out and cheap ones take other components with them. Did you find the make and model of your PSU?

    Finally, 85 degrees on your CPU when running prime 95 indicates a problem

    Here is an overclocking guide for the 920. Your CPU is one of the best around to overclock. I have one, and had a blast playing with all the settings. Reached 4.3 Ghz and did not go above 85 degrees, though.

    Your case probably needs more airflow, or you may have a failing thermal pad between the heatsink and cpu or even a poor cpu itself (unlikely). Check your existing fans first and make sure they are all working properly. Add fans to increase in flow at the bottom and front and increase outflow at the top and the back.

    Before you purchase a new heatsink, you need to make sure that it will fit inside your "no name" case. The Corsair h70 is big, and so are all of the air cooling options for your huge CPU that gets HOT when it goes fast.

    Good luck!
  7. ok, thanks eloric. though i don't think i have a "Toxic" 5850. i think i just have a stock one. is there a lot of difference between two versions of the same card in terms of power usage? because the only thing that's different is the cooling solution right? and yes, CCC comes as standard with the driver, so i have that up and running :p

    if i get a new PSU, how much wattage should i get? is 700W enough if i get an H70 and OC my CPU too?

    anyway, i'll be back at that pc by monday. i'll keep this thread updated, and see what happens :)
  8. wow, for some reason i'm not allowed to edit my posts o.o
    i'll put it here then:
    what errors should i expect when overclocking my videocard? i've read that once you go too high with the GPU you get artifacts on your screen, but are there any other things i should look out for? (besides the obvious overheating part :p)
  9. The difference between boards that have the same GPU is relativley small, so the fact that you don't have a Toxic makes little difference.

    700W is an ample power supply, but remember that there are good ones and bad. Antec, Corsair, Seasonic, Sparkle, Xigamatec and XFX make good products: others are a toss up.

    JonnyGuru is the toughest reviewer around. If it is recommended by Jonny, then you're safe. Aw heck, I'll make a confession here: Jonny is so tough, even if he gives it a 7.5 rating, I would still consider purchasing a PSU for the right price. Don't tell anybody else, though.

    The catlyst tool will tell you if the setting is not acceptable most of the time. After that, the most common error when overclocking is that you will see skips or artifacts in your screen. That is when you know to back off to the previous setting.
  10. umm guys, i went to my other home (divorced parents) already, because i couldn't wait. i started ramping up the memory, and waiting 5-10 mins every time before adding another 5-10 MHZ. i was just going to leave and let it test for the night, but when i upped it from 1205 MHZ to 1215 MHZ i get a bluescreen and the PC restarts.
    as i expected this, i didn't worry about it, but here comes the weird part: i let it boot and i started up furmark and MSI afterburner, loaded my fan curve and backed down to 1205 MHZ. all was fine until i (10 secs later) started furmark. i get a blank screen and no recovery whatsoever. i had to hard-reboot the system.

    then i lowered the mem to 1195 MHZ and tried again, same problem! though at stock speeds it (thank god) still boots like normal.

    is this normal? have i screwed up?
  11. still can't edit posts...

    well, waited a night, then tried again. i put it down to 1195 MHZ and then went to school to let it test. when i came home there was again a blank screen, and i had to hard-reboot again. when i fired up the system, it said "overclocking failed" in BIOS POST, and asked me to enter setup or load default values. but i didn't make MSI Afterburner run the overclock at startup, so what's wrong?

    also, my CPU fan went roaring to 100% and wouldn't go down, i downloaded speedfan, put it down to 35%, and set it to auto again. after i closed it everything was fine (luckily).

    and, windows doesn't detect one of my mem sticks for some reason, though CPU-Z does! what have i done wrong? o.o

    help please!

    edit: after a reboot the CPU fan again jumps up to 100%, and after POST, when it normally goes to normal it just lowers for 0.1 seconds and then keeps on making noise at 100%...

    and quick-edit does work o.o

    edit 2: great, after setting the CPU fan down, it doesn't automatically go up again when the CPU gets hot... any way to fix this?
    edit 3: checked BIOS, now it says "no hard drives found" and then still boots.
    i know my storage config. should be AHCI, but in BIOS it said IDE. that was after i discovered it said "no hard drives found". i thought that was the problem, but it wasn't.
    also, "Intel Speedstep" is still on in BIOS, and there it says the CPU fan should be controlled by the OS...
    my system seems to get worse after every reboot :(
  12. You don't need more than 10-15 mins of Furmark... trying to let it go for hours and hours is asking for trouble, it stresses the GPU far more than any game possibly can.

    Anyway sounds like you need to reset your BIOS to default and start over. Your CPU fan speed should be auto regulated. In the BIOS there should be something to set the fan speed limits and the temp it will hit max speed. (like 100% at 70C, for example)

    Also have you reseated the heat sink yet? Like we have mentioned, your temps are way too high on the CPU. Stock speeds and stock cooling shouldn't go over 70C at the most... I'm thinking that it's probably maxing the CPU fan right now because it's completely off the CPU and so your core temps are sky rocketing.

    As for the GPU, 1200 mhz memory is pretty high. A few lucky people can get up to ~1250. Mine are fine at 1200 but even 1210 I get issues. If you can only get 1190 or 1180 that's still good. Also, when doing this, what is your core speed set to? You need to determine if the issue is the GPU core or memory. I mean if you're at stock voltage and trying to get 900mhz core then to me that's obviously where the issue is... but you didn't really say what setting your at.

    Also, use GPUZ to check temps. Afterburner only reads the core temp, but in GPUZ you can also see memory and VRM temp (VRM is the voltage regulator). If either of those goes over ~120C it'll crash.
  13. yay! infos! tnx wolfram!

    well, to answer your questions, i was following this guide, which stated that once i found my final memory speed i should let it run for at least 6 hours to test stability. is this guide just plain wrong?

    as for resetting BIOS, how do i do that? and is it possible to make a backup of my BIOS? just in case something goes wrong?

    and if i reseat my CPU, there's probably thermal paste underneath that, do i need to reapply that? because i didn't put it on myself so i don't have any paste...
  14. Overall that looks like a good guide. I just don't think 6 hours of FurMark will show you anything different than 20 minutes so long as your temps even out for at least 5 mins of the test. If all is good just start gaming. Should you have any issues then, back the clocks off 5mhz and try again. GPU core instability is pretty easy to find. If the temps evened out, it'll generally artifact pretty quick.

    One thing about their instructions (I just did a scan over not a full reading). I noticed it said to use a low resolution. IMO use the highest resolution you can while being able to also keep GPUZ/Afterburner open on the side to monitor temps and usage. Also for memory OC, use at least 2x or 4x MSAA to stress the memory a lot. If you don't use AA settings you won't really be stressing the memory. I ran some Furmark benchmarks comparing memory clocks and the FPS. Without AA the FPS didn't budge, only core speed changed the FPS. But with AA, only memory speed changed the FPS while core speed barely did anything (helped the min FPS a bit).
  15. allright, tnx.

    though i do have another question about that BIOS reset.

    i could also update it, because i think there is a newer version. but i have no idea where to find my own BIOS version... and i haven't set-up this BIOS, nor have i done anything in BIOS whatsoever, so i don't have any experience with BIOS. how do i do a BIOS reset, and can i backup my current BIOS in case something goes wrong?

    and do i need new thermal paste if i reseat the CPU cooler?

    and thanks Wolfram, you've already been a great help for an OC nub like me :) i also think this thread will help a lot of people that want to start OC'ing to get a clear picture of what to expect xD
  16. 1) You can usually find the BIOS version by simply going into the BIOS at start up. It should be somewhere.

    2) Updating is usually a good idea. You can check the notes by the manufacturer to see what sort of things they fixed.

    3) Most motherboards will save their stock BIOS permanently (the one it shipped with) so you can revert back to that. You'll have to read your manual or check out the manufacturer website for more info on that.

    4) To reset the BIOS either use the jumpers or unplug the battery. Basically near the CMOS batter (it's like a nickel) there should be a 3 pin jumper with a sleeve around 2 pins. Just move that sleeve to the other 2 pins (as in from 1, 2 to 2, 3) and count to 10 then move it back. During this time you should have the PSU either switched off or unplugged. Same goes for simply removing the battery.

    5) Yes you'll need new thermal paste. Clean all the old stuff off, usually an alcohol swab is best. Don't use water. There's plenty of guides for applying new paste. Make sure you get the cooler on right, it should click down, and the pins should stick out the back side a bit.
  17. 1) done that. i have 0805, and they just fixed support for some keyboards, support for bigger HDD's (which i don't have) and added support for new CPU's. so i don't need to update. right?

    3) found that. i have the CD and the manual which'll guide me through everything should that happen :)

    4) i cannot find the battery nor can i find the "jumpers". i've looked all through the Motherboard, but i can't see it. the only thing i can come up with is that it's either located underneath my graphics card or underneath the CPU cooler. both of which i find very unlikely... sorry if i'm being really noob...

    5) allright. i'll see to it :)

    and dinner's ready... i'll reply in 30-60 mins

    TNXX!!!! =D
  18. ok. found it. it's right below my GPU, lol. so i'll have to take off my GPU. is there anything i should worry about or take any precautions?

    this is what i'll do:

    1) power-off computer
    2) switch off PSU
    3) remove PSU cable
    4) grab something metal to get rid of static electricity
    5) screw off the Graphics card
    6) remove the graphics card & cables if necessary
    7) remove the battery
    8) wait 10-20 secs
    9) put battery back in
    10) GPU back in & power-on

    and will it then work? or do i need to run a BIOS setup or something?
  19. That should work, but if the jumpers are accessible it might save the trouble of having to remove the GPU.

    They look like this:

  20. nope. i see a lot of those, but not one of them says something like "cltrc". one says "chassis" and the other one isn't marked. and both are nowhere near the battery. so i'll see you in a few minutes :p
  21. that didn't work :(

    i got POST, then it says "no drives detected", then fully boots. but still with the CPU fan at 100%.
    btw, i know it's not my temps, since they're running at ~40*C (idle), but the CPU fan still won't slow down automatically :(

    this is a screen on Speedfan and Coretemp.

    so, have i totally screwed up my pc? o.o
  22. Definitely some wierd issue then. For now I'd say poke around in your BIOS at start up. Load the defaults from there and double check what the fan settings are. It's entirely possible that the defaults are to run the fan at max and you need to set it to use proper scaling with the temp.

    My guess is after you got the overclock failed sign you told it to load the defaults and that's when the fan speed got screwed up. The reason it said that is simply that the mobo crashed, it doesn't really know why so it assumes it was an overclock failure. Generally speaking if you didn't change any settings or know that the crash was unrelated to any cpu overclock, then don't load defaults just tell it to "enter setup" (I assume that's the other option) and then save and exit without changing anything.

    But yeah, this time go poke around in there and look for the fan speed control. I also have an Asus mobo so it's probably similar, it's in a "hw monitor" sub menu.
  23. aw crap... i did let it load defaults :O

    i'll look around then, and tell you what i find.

    what i already did find the last time i poked around there was that the "intel Speedstep Technology" was already enabled, which, if enabled, would let the OS control the fan speed... could it be because i loaded a custom fan profile for my GPU, and that that screwed up my CPU fan profile?

    anyway, i'll report on you :p
  24. No, EIST is related to automatically adjusting the clock speed. Mostly it's used to downclock the CPU at idle to save power. If you look at CPUZ it usually shows only a 9x multiplier at idle but under load it goes up to 21x or whatever. It also is needed to use Turbo Boost, which requires the use of a few different multipliers.
  25. YAY! FOUND IT! well, the fan speed control anyway. it's under "Hardware Monitor", and then Q-fan control or something similar. i saw it earlier, but i didn't change it then because i didn't want to screw my pc up any more than it already was :p

    so now there's just the "no drives detected" issue, which isn't really an issue as windows loads fine, but i'd still like that fixed.

    do you have any idea what it could be? i have everything set to IDE, because i got a BSOD when i set the storage control to AHCI. though when i first got it it didn't work because it wasn't in AHCI.... any ideas on that?

    and my last question: is it (relatively) safe to start OC'ing again? because now i know that it's safe at 1195 MHZ. though i'll try again with some MSAA.

    oh, btw

    THANKS WOLFRAM!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! you are my hero :P i seriously thought i'd ruined my whole pc.

    and now i did learn my way around BIOS, so it's actually pretty good that this happened! but now we have to just fix that last HDD issue, but still, THANKS! =D
  26. If you installed Windows with it set to AHCI then that's what it should be. If it gives you a BSOD again it's because Windows isn't loading the right drivers. There's a fix for it, but you have to poke around in the system registry.
  27. allright. i'll look into it. as for OC'ing, how long should i test if i think i have the right frequency? i just got artifacts and a crash with my mem at 1180 MHZ, with furmark at 1680x1050 with 8x MSAA. is that too much? because i could reach 1205 MHZ, but then again, that was at 1024x768 with 0x MSAA...

    edit: apparently the temp matters a lot, even when it's low... i don't get it over 65-70*C, but it ran fine at 1175 MHZ, then i let it pause for 2 minutes, started furmark again and got a crash anyway.... do you have this 'problem' too wolfram? or is this just common in OC'ing?
  28. You're still at stock core speed and voltages right?

    8xMSAA can be pretty hard on it. Try it at 4xMSAA and see what happens. Maybe start back down at 1150 like this. I find if the test is stuttering a lot it's likely to crash, especially if the stutters are seemingly random or last different amounts of time. The wierd thing about GDDR5 ram is that instead of crashing right away, if there's an error it can just resend the data. You end up with a lower framerate at a higher speed setting. The key is to find that point where you get stability and best framerate.
  29. If you have a no name PSU then I suggest you don't overclock anything. I had a Diablotek 550W PSU that fried my GPU while refusing to allow it to even overclock a little bit. The stupid thing gave it bad voltages and fried it.

    Of all the components to skimp on when overclocking, the PSU is not it. I happened to have a UD3P board so most of my system survived but I had to RMA the GPU multiple times as a result of said frying.
  30. well i'm not going to overvolt anyway, and i can expect even my no-name PSU to work properly at normal voltages... right?
  31. Generally yeah, but one issue is that they might not have enough amperage available.
  32. oh... is there any way to check?

    looked at my PSU again, it can do 650W =D
    i also found the model number; it's an "Advance EA4G-650"
    on all +12v rails it can do 18A, on the +3.3V and 5V rails it can do 27A, and on the -12V and "+5Vsb" rails it can do 0.8A and 2.5A respectively.

    i'm just typing what it says on the tin, i have no idea what i'm talking about :p

    so what rail do i need to check to see if it has enough amperage? Google isn't being very helpful by showing lots of guitar amps, lol.
  33. You should be ok, from what I've read a 5850 is ~150W so that's ~12A. With overclocking you're looking at maybe... 15A? So 18A is enough although it's not a big margin.
  34. you are amazing :p

    well ok, but i just got a BSOD saying "page fault in non paged area". i just turned automatic restart off after this BSOD, so should it happen again i'll have time to read it. the crashdump is empty for some reason, but i just tried to open it with notepad and not with anything else. I wasn't overclocking btw, just regular PC use, not even gaming.

    i hope it doesn't happen again, and i hope it does happen again because then i'll be able to see it, lol. anyway, could i have (partly)fried my GPU memory? internet says this could happen because there is damaged memory in either the HDD/RAM/GPU mem. i'll run a memtest and lay off the OC'ing for some time, i want to keep my system running :/
  35. Could be any number of things. If it persists, start worrying but if everything is stock and no issues for, say, the next week then it was just a random thing. Windows isn't perfect, remember!
  36. allright tnx!

    though i have another question: last time i tried to get a final memory clock, when it was hot it keeps running at 1195 MHZ, but when pause furmark and let it cool down to idle temps, then fire it up again, i still get a crash. how do i now know when my clock is stable? because crashing everytime can't be good for any piece of hardware...
  37. Acutally I had a similar thing happening... AFAIK it's a heat issue with the VRM/VRAM so if you're only using Afterburner you won't be able to see those temps. GPUZ can read them tho, usually. Generally they don't have as good cooling as the core itself.
  38. i got another bluescreen... and i think its my HDD. i can now hear it spin-up at boot, which i never could. also the boot was VERY slow, it took it like 3-4 seconds to go from the end of "welcome" to actually showing my background o.o

    i don't know if you can do anything with this, but this is what windows says:

    Gebeurtenisnaam van probleem: BlueScreen
    Versie van besturingssysteem: 6.1.7600.
    Landinstelling-id: 1043

    Aanvullende informatie over dit probleem:
    BCCode: 3b
    BCP1: 00000000C0000005
    BCP2: FFFFF88004B02817
    BCP3: FFFFF88006DB2900
    BCP4: 0000000000000000
    OS Version: 6_1_7600
    Service Pack: 0_0
    Product: 256_1

    Bestanden die helpen bij het beschrijven van het probleem:

    i just left for guitar lessons for about 1 1/2 hour, and left my HDD defragging, because it was time anyway. when i got back my screen was just blank o.o
    i think i screwed something up >.<
    and i'll stop OC'ing whatsoever, at least until i get a new PSU >.>

    update: it's the display driver that's causing those last BSODs! my pc still works! i'm still looking into it though :p
Ask a new question

Read More