i5 3570K OC Adventure

Discussion in 'Die-hard Overclocking & Case Modifications' started by Matt26LFC, Aug 13, 2012.

  1. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    OK guys, I've got my rig all put back together with the new parts, actually got it all together Sunday, but had Graphics Cards issues, my Watercooled 6950's today whilst playing Crysis hit 100C lol which is clearly wrong so I spent time today figuring out what was wrong, I solved the problem, now load temps in Crysis 2 didn't go above 39C which is how I like it :)

    Anyway I've upgraded from

    i7 930
    Asus P6X58D-E
    12GB Corsair Vengeance 1600MHz

    To

    i5 3570K
    Gigabyte GA-Z77X-UD5H
    8GB Corsair Dominator Platinum 1866MHz

    I'll try take some photos of my system and post them. Until then I'm going to try my hand at Overclocking the i5 3570K. I'm totally new to Overclocking SB/IB style as my previous chip was i7 930, from what I've seen/read it should be easier to OC these :)

    First screen shot is everything set to stock with a quick Prime27.7 run with AVX for 20mins

    [​IMG]

    OK, temps look good. Rooms ambient was approx 23C-24C

    BTW I'm using a Full Watercooling Loop to cool my system, consisting of an XSPC Raystorm CPU block, two Heatkiller Full Cover blocks for the GPUs, XSPC D5 Pump/Res, XSPC RX360 with Push/Pull 1200RPM Fans + RX240 with 20mm Shroud and 1650RPM Fans in Pull.

    Perhaps I'll do some SuperPi runs and 3dmark runs at stock before I OC to try and get some idea of performance increase from the OC. Then I'll shoot for 4.5GHz and run them again after I've gotten 4.5GHz Prime stable.

    Ok thats it for now, I'll get SuperPi and 3dMark downloaded and installed.

    SuperPi @ Stock settings

    [​IMG]

    3dMark06 Run @ Stock

    [​IMG]

    3dMark Vantage Run @ Stock

    [​IMG]

    3dMark 11 Run @ Stock

    [​IMG]

    Ok, that'll do for tonight, its gone midnight here now. I'll crack on with the Overclocking 2morrow, might even take some pictures of the UEFI BIOS showing my settings.
     
    Last edited: Aug 14, 2012
  2. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    Haven't had much time today to play, though had a little play

    UEFI Shots of BIOS settings for a 4.2Ghz OC

    [​IMG]

    In this shot you can see I set the Multi to 42 and set the XMP Profile for the RAM to run 1866MHz

    [​IMG]

    Here you can see I've disabled all Power Saving features of the chip and the Turbo feature as I just want a Fixed Voltage and Clock Frequency at all times.

    [​IMG]

    As for Memory, I've left all settings to the Profile, not playing with this yet. Since the RAM is designed for the settings its set at I'm hoping it won't inter fear with the OC.

    [​IMG]

    These are settings I've copied from Sin0822 at OCN, though I think he has LLC @ Extreme whereas I've gone one notch lower to Turbo.

    [​IMG]

    All other Voltages aside from Vcore are on Auto, CPU PLL is something I may lower as its been said it can help with temps otherwise for now its at 1.8v.

    Also not sure CPU Vtt and IMC are all about, thought they where the same thing, so need to research whats what there before I play.

    I've also turned off the Onboard IGP.
     
    Last edited: Aug 14, 2012
  3. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    Stress Test Result

    Ran Prime95 27.7 with AVX Custom Blend using 80% of my system RAM for 30Minutes.


    [​IMG]

    Now I know its only half hour, however this is just a small step in the direction of hitting 4.5GHz for 24/7 usage. Pleased with temps so far, 4.5GHz should be OK for my cooling setup I think.

    Gonna try for 4.3GHz next...

    Next Stress Test Result

    Same as above, only difference is its 4.3GHz and I upped the Vcore to 1.19v

    [​IMG]

    Hottest core broke the 70C barrier, however I'm still good to go.

    One thing thats confused me is the VID reading in CoreTemp, its gone up and its also much higher than what I've set in BIOS and what CPUz is reading, so not sure what thats about either.

    Anyway, onto 4.4GHz

    4.4GHz failing with BSOD 124. Not sure what this means for IB. Upped Vcore to 1.21v as 1.19 & 1.20v failed Prime instantly!!

    1.21v 1.22v also failed, lasted about a minute longer than previous two attempts.

    Trying 1.23v - Need to look up BSOD Codes for IB too, don't know whether more Vcore or something else needs looking at.

    1.23v Failed, though it lasted 15Mins so I've upped Vcore again to 1.24v too see what happens before looking into other things.
     
    Last edited: Aug 14, 2012
  4. BlackZero

    BlackZero Guest

    The 'vid' in core temp is just that, the vid. It is based around what the cpu is asking for and as these processors use dynamic or d-vid it will change with load. What you have been changing is the vcore.

    Regarding bsod 124, this can be caused by a number of different variable, mostly vcore but also effected by vccio and pll voltage. Try upping the llc and dropping the voltage set in bios to compensate for the overall increase. Also try increasing lowering the cpu pll voltage and vccio voltages. Reducing the cpu pll is instrumental with idle bsod 124.

    On a side note, these IB chips really do run hot considering you've got a 360 + 240mm rad cooling your system. Though these last few weeks have been quite warm too.
     

  5. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    Hey dude, hows you?

    Cheers for the input, I've had a bit more success, 4.4GHz at Vcore 1.24v went for 40mins before I stopped it

    [​IMG]

    Temps are starting to spike quite high, well one core is anyway, but for the most part they seem to sit around what you see in Real Temp and Core Temp for current temps. Not sure why they suddenly spike every so often.

    I'll go for 4.5Ghz, may change LLC to Extreme see if it lasts 30mins of Prime at the same Vcore, probably won't, who knows, and take it from there. If its successful at those settings I'll lower CPU PLL see what happens, maybe temps will drop a little bit, but yeah these do seem to run hot. Its also 25C in my room about 5C warmer than I like it! But at least I know it can do these frequencies and volts in the summer, looking forward to winter now :)

    Ok, 4.5Ghz Prime95 27.7 AVX 30Min Run

    Had to up the Vcore to 1.27v (I think, either that or 1.28v lol) LLC set to Extreme

    [​IMG]

    Temp really spiked on Core 1, hitting 83C! Cores 0 + 3 look good, quite a large difference in temps there, though I believe this is common for IB, again no idea why lol Perhaps I should muster up the courage to de lid this SOB and get shot of that ****e TIM Intel put in there!

    Well won't be doing much more tonight, I'll drop the CPU PLL down to 1.5v, which is the minimum I can set I think, and see what effect that has on temps/stability then probably call it a night.

    Ok, Same test as before with one change CPU PLL is at 1.5v

    [​IMG]

    As you can see from the result its made virtually no difference to the temperatures at all, so for now it seems that lowering CPU PLL does nothing for temps. Perhaps in a heavier OC it might make a difference, but at 4.5GHz its done pretty much nothing.

    I did notice however that I can drop CPU PLL to 1.2v, don't think I've read anywhere of someone doing that so I'll prob leave it at 1.5v for now, maybe try 2morrow see what happens.

    Obviously I don't consider this OC 24/7 stable yet so after maybe I've tried the really low CPU PLL I'll do some gaming to see how it holds up, and I guess I should really do like a 12Hr Blend, I may leave that for the weekend though would rather be near the computer when I do it as I didn't properly leak test when I put it all back together :) and throw in an IBT run.

    Will also redo those benchmarks, see what kind of boost I get and upload the new scores to HWbot, try pick up a couple more points, wanna break a 100 :)
     
    Last edited: Aug 14, 2012
  6. BlackZero

    BlackZero Guest

    Yeah, not bad Matt.

    Temps are pretty high for 4.5Ghz, especially without HT. A lot of it probably due to summer too I guess, but I'd take an average temp for all 4 cores though as that single core is quite a bit higher. My chip has a single core that runs 6-8c higher than the rest too, quite an annoyance.

    Still, those temps do seem a little high even for IB. You might want to double check your equipment and maybe redo the paste. Just a tiny drop in the centre of the cpu and allowing the heatsink to spread it out has always worked best for me.

    Also, regrading the PLL voltage. On my board I can go as low as 1.2v as well IIRC but my system refuses to boot at anything below 1.4625v. Figures around the 1.5v mark seem to work best for me. I have been running it at 1.5125v recently and this has proved to be quite stable, especially when using high levels of LLC with offset voltage as the idle voltage can be quite low.
     
  7. ---TK---

    ---TK--- Guest

    Messages:
    22,104
    Likes Received:
    3
    GPU:
    2x 980Ti Gaming 1430/7296
    Quite toasty on water man
    Try CPU pll 1.71v. 0124 is usually vcore. 0101 too.
     
  8. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    I might be worth re-seating the block, when I did pop it on I did think maybe that was too much compound. So I may whip it off and try again, that'll be a weekend job, means draining the loop again :( shame my tubing runs are way to short for quick disconnects, they'd be real handy lol

    But yeah being 25C in this room isn't helping, get down to 20C and I'll be looking at my highest core being around 78C.

    btw I use Phoyba HeGrease, think its a good one, I'm sure Skinnee gave it a good review.

    Perhaps I won't bother with lowering CPU PLL any further it seems just as happy at 1.5v as it did at 1.8v so I'll leave it at that for now then.

    Well looks like I'll be busy this weekend, re-mount and massive Prime Run lol

    Anyway...

    New SuperPi 1M Score

    [​IMG]

    Just submitted this to the bot, picked up a couple points for my overall score, only about .1 points for the team though lol, not doing much bench stuff atm so I'll tweak and what not later and get better scores.

    Will run 3dmark 11 and do the other two 2morrow :)

    Ok, 3dMark 11 @ 4.5GHz

    [​IMG]

    Little disappointing to say the least, however I think running my two 6950s at stock is probably holding that score back a lot. Will have to give them an OC 2morrow see what happens.

    Ok 3dMark06 @ 4.5GHz

    [​IMG]

    Not bad a 5K increase, though when I OC the graphics cards hopefully I'll get more :)

    3dMark Vantage P Score

    [​IMG]

    Again, not bad, but thought I'd get more, perhaps more GPU power!

    Think I'll flick the BIOS switch on both cards as I have unlocked BIOS on BIOS2, so 1536Shaders and 880Mhz Core and 1375Mhz Memory clocks see what that gets me.
     
    Last edited: Aug 15, 2012
  9. BlackZero

    BlackZero Guest

    Yeah you should be close to the 12k mark with the GPUs overclocked, that would be a decent score. :)

    Don't think you need to drain the loop to change thermal paste, just detach the CPU heatsink and apply new paste.

    I have been using MX-2 thermal paste over the last few years, fortunately had enough stocked up to last me even after they stopped producing it. Though I think they may have started production again recently due to demand.

    Other than that I have been prolimatech pk-1 for my last few builds, has worked very well and doesn't cost a packet. I also have a small syringe of shin-etsu that I use occasionally but it is quite expensive and doesn't perform much better than the pk-1. I have never used the phobya thermal paste though, but from the reviews I have seen it performs quite well.
     
  10. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    Yeah, hopefully. My previous best was just over 10K, so really hoping to beat that.

    Just updated my previous post with new 3dmark scores.

    I might be able to clean off the thermal compound and reapply with out draining, the compression fittings are all 45Degree Rotary, so might be able to work something there. I'll remove the RAM and maybe be able to swing it around saves me draining.

    Yeah I've heard PK-1 is also very good, most of the top compounds seem to be very close to each other anyway, think I'll keep with the HeGrease, I think its the same stuff as GELID Extreme or something, can't remember.

    I've also got some Shin-Etsu, I used it on my GPUs when I put them under water, so far pretty good temps with it, well when I configure my loop properly I do lol
     

  11. Ilya

    Ilya Guest

    Messages:
    273
    Likes Received:
    1
    GPU:
    GTX 780 SLI @ 1200/7000
    I have mine running at 4.4ghz /w vcore on auto
    After 16 hours of blend test /w p95 avg temp was between 82-87
    on auto my vcore stays at 1.206-1.218
    Im using h50 as cooler and coolermaster scout (small case)
     
  12. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    I see, perhaps my temps aren't that bad then. Still I'll probably give my block a re-mount just incase.

    MSI Afterburner Issues

    Guys, I'm having a strange issue with MSI Afterburner, every time I try and unlock unofficial overclocking it reads my Core and Memory Frequencies as 0 and I can't change them!? GPUz still reads them as stock settings, but Afterburner is saying 0!? If I move the slider up and hit apply it just goes straight back to 0!?

    I've tried AB 2.2.3 and 2.2.2, I've also tried Catalyst 12.6 as I was running the 12.7 BETA, currently running 12.6 WHQL. When I edit the txt file (MSI Afterburnercfg) back to its original settings it reads the frequencies no problem, I can even adjust them to 840 & 1325, but once I unlock it just doesn't want to know

    Any ideas?
     
  13. BlackZero

    BlackZero Guest

    You have to use msi afterburner 221 Initially due to some needed files being removed by AMD and by AB in later releases. Once you've installed 221 you can then use later version too.

    Once that's installed create a shortcut and edit it's properties to add the "-xcl" switch at the end of the original file location.

    That should unlock MSI AB without the need to edit anything else.
     
  14. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    Cheers dude, didn't know that.

    Took me ages last night to get things working properly though lol I eventually got there though. Though some benchmark scores didn't go up by much even with the GPUs Overclocked

    If I remember rightly I got 3dMark11 upto about 10750 with 950MHz on my Cores and 1375MHz on my Memory, was hoping to break 11K with those settings. Oh well I'll keep playing at see how it goes.
     
  15. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    IBT Run v2.54

    [​IMG]

    Man those temps are up there, worse than under Prime!!!

    Reason I whacked this up is I want to know what people think of the GFlops? Do they look about right for 4.5GHz?
     

  16. BlackZero

    BlackZero Guest

    Yes Gigaflops are spot on for 4GB. Though I wouldn't bother with linpack for 1155, and especially not for IB.
     
  17. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    Cheers mate, only ran it to make sure my GFlops where all gravy ;)

    Gonna try remount the CPU block maybe this afternoon hopefully that'll sort out my temps problem. If not think I'm going to just have to assume I have an average chip, which would leave me a couple options, de-lid and throw on some of that Liquid Pro stuff or buy another one selling this one and hoping for the best.

    Just wish the 3770K's where cheaper, I've read there just the better binned versions.
     
  18. thatguy91

    thatguy91 Guest

    I have heard that raising the Bclk slightly can be beneficial for overclocking.

    For instance, 44x103 (4532Mhz) instead of 45x100 (4500Mhz).

    Obviously the first one is going to be faster, but I know of someone that claims that even 44x102 (4488) was faster than 45x100.

    Getting an answer from this person is next to impossible, but in saying that, he's usually right in everything he says.

    So, what are people's throughts on this? I know raising the Bclk also raises the PCI-E, but we're talking 2 or 3 percent here. (102 or 103).

    Reason I ask is I'm doing the upgrade to Ivy Bridge shortly, so would be good to see whether it's actually worth doing beforehand. I know people just say leave it at 100, its better to raise the multiplier, but that is not what I am asking, I know that! The question is, has anyone actually tried it and seen if there is any difference in benchmarks etc? I also realise raising to to 150Mhz etc is bad! but what is the consensus on other options, like 43x105 (4515Mhz)?
     
    Last edited by a moderator: Aug 17, 2012
  19. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,123
    Likes Received:
    67
    GPU:
    RTX 2080Ti
    Could be, I'm not sure, playing with the base clock means your messing with the PCI-E etc as you've already stated. Maybe going up 2-3Mhz is doable I don't know, I do think that 110Mhz+ is more the realm of Extreme Overclockers with Sub Zero cooling, probably not good a idea for us chaps with more conventional cooling. As for 150Mhz base clock, pretty sure thats not going to happen, way way to high.

    When I've improved my cooling I may play with the baseclock a little.

    On that note heres a Prime95 shot after I remounted my block

    Test is at Stock Not 4.5GHz

    [​IMG]

    Well as you can see from the pics temps are up, however I believe this can be attributed to the fact that its about 4C warmer than when I did my first stock test. So I'd say temps are really about the same.

    I may do another re-mount but drain my loop this time and while its drained I'll open up my CPU block to check for any gunk. I;m pretty sure I won't find anything as I cleaned it out when I put this rig together but I guess its worth a look.

    If the temps are still as they are I'll have to seriously consider de-lidding or buying another and selling this one, which just sucks lol
     
  20. BlackZero

    BlackZero Guest

    It seems yesterday was the hottest day of the year so you couldn't have picked a worse day to measure temps.

    The point to note here would be that at stock/lower overclocks the temps are going to be entirely limited by ambient conditions, especially considering it's only 77w of heat. Therefore it might be worth running the same test at stock and then at a higher overclock, this way you can compare the difference in temperature between both and use this as a more meaningful indicator of how well your loop and the chip is able to handle heat.

    I don't think it's going to get any hotter, hopefully.
     

Share This Page