TridentX F3-2400C9D-8GTXD Downclocked / 3770K @ 4.7GHz / Bsod 124 Issues

[Arrow0309]

Hi, here I am again cause this 3770K / F3-2400C9D-8GTXD combo it's just running me insane.

After I got "stable" before with the Xmp defaults of the ram at 4.6Ghz (never tried a 24h prime95 run though):

http://www.gskill.us/forum/showthread.php?t=12487

Now I've got my cpu delidded for a couple of months with a nice, 19°C temp drop, and so, tried some further oc (4.7-4.8Ghz) with the same, 2400 cl 9-11-11-31-1T ram and 1.65v vram / 1.20v vtt

The result: Linx 1h or so stable (up to 28000 problem size) but giving me randomly the bsod 124 otherwise

Now forget the Xmp timings, and 2400Mhz, I recently lowered everything in order to get a nice, offset 4.7GHz daily so I set the ram manually at 2133, cl 9-11-10-27-1T with only 1.5v and default vtt (1.05v) and seemed stable with the vcore vdropped at 1.296 il LINX 0.6.5, latest (offset, LLC medium):



But they're still 124 bsods, randomly

For the last two days I tried lots of combinations of the vcore (offset), LLC, and vtt, even changing the ram settings to 9-11-10-28 @ 1.55v, nothing, there always end with the bsod 124

Not sure what to think about, faulty cpu, faulty ram?
Could any one give me a hand please?

I'm using the stability test of Aida64 because it's one serious stress test, tried about 25-30 different bios settings this morning and I'm always getting the 124 bsod (after 2-3 min or even instantly)

What am I doing wrong?

Forgot to mention, I was using both F12a bios (one of the stablest) and the latest F12j mod (using right now)

[emissary42]

If i remeber correctly a bsod 124 most of the time is cpu voltage related, either vcore itself or vccio/pll.

[supahos]

Quick search shows lots of causes for this, most common seem to be jacked up memory, jacked up gpu, jacked up cpu (voltage), ... if you havent already run memtest on the sticks individually, if they all pass and onboard video is an option try using it to run your stress test, if neither of those work (or maybe even first) downclock your cpu, it might not be stable, just because you can get that high doesnt mean it'll be stable at that speed


Edit: after rereading ill almost bet you money either your processor isnt stable at thatbspeed or you messed up a voltage on the new oc

[Arrow0309]

Quick search shows lots of causes for this, most common seem to be jacked up memory, jacked up gpu, jacked up cpu (voltage), ... if you havent already run memtest on the sticks individually, if they all pass and onboard video is an option try using it to run your stress test, if neither of those work (or maybe even first) downclock your cpu, it might not be stable, just because you can get that high doesnt mean it'll be stable at that speed


Edit: after rereading ill almost bet you money either your processor isnt stable at thatbspeed or you messed up a voltage on the new oc

I already tried running memtest 86+ in the past, no errors
What about the igpu, I didn't get you quite well, however I'm only using the dedicated video card (HD7950), should I simply disable the internal graphics (right now in auto) from bios?

[Arrow0309]

However, I just tried something else at 4.7ghz, no cpu saving features, fixed vcore at 1.30v, LLC Turbo, ram at 1.50v and every other voltage at default.

Much better, lower full load vcore, here's quite an hour of Aida64's stress test:



But then I had to leave and when I got back home I found a new windows session with the bsod 124 error message


Now what?

[supahos]

no, pull your video card, and run off the onboard graphics just to see if that is the issue, also could try updating any and all drivers you can, or if you recently updated (say a video driver) roll it back. Another note that ram is designed to run at 1.65v isn't it? Don't skimp it out, it won't hurt it to run at rated voltage especially with it not overclocked

[Arrow0309]

no, pull your video card, and run off the onboard graphics just to see if that is the issue, also could try updating any and all drivers you can, or if you recently updated (say a video driver) roll it back. Another note that ram is designed to run at 1.65v isn't it? Don't skimp it out, it won't hurt it to run at rated voltage especially with it not overclocked

No, I haven't recently install new drivers except for the ME firmware and MEI drivers
And I've never used the integrated graphics, never installed any intel graphics driver.
And the 124 bsod (randomly, or when I'm playing games) is something that I'm getting for quite a while.

About the memory (gonna try a memtest86+ session this evening), you mean 1.5v are not enough for 2133 with 9-11-10-28-1T? This memory could do 2400 9-11-11-31-1T or even more with 1.65v but you have to bump your vtt as well, I don't want it no more, it has the Samsung IC's that could do 1866 with 1.35v, do you really mean I should increase it to 1.65 even downclocked?

I've seen someone running it like that with a 5Ghz oc:

http://www.techpowerup.com/forums/sh...9&postcount=16

[supahos]

You don't have to run 1.65v, but it is the rated spec for it Running less Voltage will certainly not make things more stable, I don't know but it just kind of sounds like you delidded, overclocked more, then are having issues. Not sure if its the Overclock, The delidding, or possibly the video card. I'm saying to remove the video card, and use the onboard video. A failing video card can cause that error as well, easiest way to isolate that issue is use onboard video.


every chip is different, you can't assume because one guy can get 5 you can get close, there have also been several have major trouble getting 4.6 stable.

[supahos]

And I know its fun, as I've OCed my stuff nearly all, just for the hell of it, but I do kind of have to ask why you're going so far out on the edge? You are only running 8gb of memory, so you're not doing cutting edge 3d modeling or video editing. Why not go back to a stable setting (seemed you found that back at 4.6mhz) and just run with it, There is no way gaming or anything short of rendering massive scenes you'd ever see a difference. It's your system and obviously that is just my opinion, but consider it?

[Arrow0309]

And I know its fun, as I've OCed my stuff nearly all, just for the hell of it, but I do kind of have to ask why you're going so far out on the edge? You are only running 8gb of memory, so you're not doing cutting edge 3d modeling or video editing. Why not go back to a stable setting (seemed you found that back at 4.6mhz) and just run with it, There is no way gaming or anything short of rendering massive scenes you'd ever see a difference. It's your system and obviously that is just my opinion, but consider it?

You're surely right, 4.6ghz is enough too, let's just say I have my reasons to stabilize the 4.7ghz daily too.
I have delidded my cpu and I have a decent air cooling, also 8gb of ram instead of 16gb could only improve to the overclock.
Did you see the max temps I'm getting, 68°C max after 1h of stress test?
Man, I saw lots of notebooks getting at almost 100° C at stock clocks during gaming sessions

OK, I did another memtest 86+ session with more than 2 pass and got no errors at these clocks as above, 4.7Ghz fixed with 1.3v and ram at 2133 @ 1.5v (although the oc clocks are not recognized by the memtest86+):



So, the last thing I have to do is trying to lower some more the cpu vcore and / or increase the ram voltage (and maybe lower the cpu pll as well)
But then, supposing I'm giving 1.65v of vdram with the default vtt (1.05) wouldn't I get this way a bigger gap between the two voltages? Or is not applying anymore this rule?

[supahos]

Glad to hear you got it all stable, even if its at 2133. I don't think the spread between Vdram and Vtt is that bit of an issue but you can bump vtt a tad if you want. I know some crazy people out there running 1.5v vcore 24/7 which sounds nuts to me but what do I know? I figured you had some sort of "reason" to max oc your cpu, was just pointing out that if you want to max your ram out someitmes you have to give up a bit of the cpu overclock for stability. Looks like you have a severly overpowered rig that should chew up anything you throw at it though either way.

Enjoy the hell out of it!


Edit: Wonder why it says "intel Core 2nd gen" in memtest before saying you have a 3770k... Dunno but thought my memtest said 3rd gen. I dunno and I guess prime will give you the final confirmation on stability.

[Arrow0309]

I checked out all the windows startup app and processes, unplugged the internet, returned to offset mode, C1e and Eist enabled, Pll Overvoltage and C3/C6 disabled, Dvid +0.100 (stress vcore is 1.284-1.296), LLC Medium, Cpu Pll 1.675 and ram 1.575. Running aida64's stability test
Bsod 124 after about 10'

Trying now with Dvid +0.90 (1.284 mainly under load), Pll 1.645 and ram 1.545 and the same, default 1.050 for the vtt (vccio)
Running aida64's stability test, 68°C core max ...

30' and running ...

[Arrow0309]

But in the end it still got me the 124 nasty bsod, after 70-75'

Trying with the linx 0.6.5 right now, in order to see if the max load (vdropped) vcore is good enough, only 1.260v during linx

[Arrow0309]

The Linx 0.6.5 run fine, without any issue, the gflops are even better than before and I even got improved temps




I think I'll stick around with these voltages for a while, see how's working and run a 12h prime95 (v26.2) session eventually

[supahos]

Sounds like a plan. Hopefully this will work for you. Hope any of what I said helped, that or I just wasted both of our time :P