Looks good chronic! Helping me a lot.

Question, you say what to do in regards to Gflops and LinX, however what do we do if for example our second pass is higher than our first; but our third pass is in between the two. Do we raise? Do we lower?

I have passes that come out like this;

54.4562
54.4994
54.4654

StormTheGeek

Looks good chronic! Helping me a lot.

Question, you say what to do in regards to Gflops and LinX, however what do we do if for example our second pass is higher than our first; but our third pass is in between the two. Do we raise? Do we lower?

I have passes that come out like this;

54.4562
54.4994
54.4654

I would try 1 notch lower if I saw that and I was trying to get my vcore as low as possible. If one notch lower causes a crash or error put it back up.

XXChronicXX

happaflapp

OK, so I started to get a driver_irql_less_than_equal bsod yesterday upon boot up.  I haven't changed any settings since my initial OC of 3603 MHz.  It just happens upon boot.  Any Ideas?  I'm not OC'ing my RAM its under-clocked.  Is something off with the voltages?  Once in Win7 everything is great, stable and can play games and so on. 

Should I run linX and see if I'm dropping Gigaflops?  Or is it a tweak that needs to happen?  Because I really haven't started tweaking much.

hmmm sounds weird you would get that during boot but then be completely stable in windows. I would try this....
Write down EVERYTHING in your settings. Take out the cmos battery for 30 min. Put it back in and boot into windows without adjusting anything but the ram voltage and timings, leave the speed/mhz  alone. Once in windows restart and then apply all your overclock settings again and boot. See if it still happens.

It is sporadic though, that's the part I don't get.  Like it just booted straight into windows this time with no problem.  I can't seem to replicate it.  I wonder if one of the voltages are off a touch.  My Vcore right now is setting at 1.29v and the VTT fluctuates between 1.27v and 1.32v.  I didn't disable the my vdroop and left the Vcore on auto.  Could this be the culprit?

No, I haven't tried that yet Chronic.  What results would that yeild if I were already in windows?  Am I checking for stability?

XXChronicXX

I've corrupted my drivers so bad one time from changing IOH Vcore and IOH/ICH I/O while windows were open. I would be able to pass tons of testing but as soon as I play a game for a little while my computer would freeze and I would have to manually reboot. It had nothing to do with my OC settings. They were completely stable. I reinstalled windows and the problem was gone. Reinstalling the display driver did not work.

 
Ahh OK, no I didn't mess with any of those settings in Eleet or in bios.  I pretty much followed your recommendations and tested, then retested.   My rig is stable @ 3603 MHz in OS.  I'm pretty sure it had something to do with SLi.  I haven't had a single bsod *crosses fingers* since I reinitialized SLi.  I'm not ruling out completely other possibilities, but for now it seems to of been the culprit.  I don't leave my PC running so I have had at least 6 cold boots since then, and no problems.
 
I do have another question for you.  Is there anyway to bump the MHz up on my memory?  I would like to get it closer to 1600Mhz.  Right now my divider is set at 2.8.  Would it involve bumping up the bclk?  Is there any benefit to increasing the ram MHz?  From my prior experience and what I have read there is nominal gain to be had from OC'ing the ram.

Hello Chronic, thanks so much for writing this up. It's been very helpful in getting me to a 3.6GHz Stable OC. 

A couple questions for you though, I'm not using LinX but Intel Burn Test (which is Linpack, but just in a simpler UI). I'm getting all sorts of bouncing GFlop outputs, have a look:

Time (s) Speed (GFlops) Result
47.033 19.0053 3.631012e-002
47.575 18.7888 3.631012e-002
47.389 18.8625 3.631012e-002
47.750 18.7202 3.631012e-002
46.931 19.0466 3.631012e-002
45.841 19.4998 3.631012e-002
49.923 17.9052 3.631012e-002
45.321 19.7236 3.631012e-002
41.718 21.4270 3.631012e-002
45.896 19.4764 3.631012e-002

The 'Results' show stability, but by your definition the GFlops in a lowering status show instability? So I should bump the vCore more then, correct? I'm using a C0 Rev 920. I'm @ 20x180 w/ 1.275v for vCore set in the BIOS. vDroop off. VTT +100mv, CPU PLL +.025v

I'm using Corsair XMS DDR31600 9-9-9-24, which is 1.65v memory. However, the bios sets my vDimm @ 1.5v. Would upping this to stock voltage help with the fluctuating GFlops? 

I haven't dropped my QPI Link yet (Still @ 6.4 instead of 4.8), I'll give that a shot here in a second to see if that helps. 

Any suggestions would be helpful. 

Thanks again for writing this up. :D

XXChronicXX

Dont forget to set the amount of tests to run, or the amount of time to run. Its on the top right of linX. I like to do 3-4 tests while watching temps and GFLOPS. If the GFLOPS continue to decrease in descending order, that means your vcore is too low. 1 notch up should be enough to fix it.  You can usually tell after 3 or 4 tests.
Example of too low Vcore....

TEST 1 GFLOPS  49.8758
TEST 2 GFLOPS  49.6852
TEST 3 GFLOPS  49.4286 ---------- I usually stop it right here if i see this. When I see test 3 doesn't jump up above test 1 or 2, I bump the vcore up 1 notch.

Could you please explain why you think that ? Differences in GFLOPS depend highly on other tasks (that includes OS) that run on your CPU at give time together with Linx. The more cpu is loaded by processes other than Linx, the less GFLOPS it shows. IMHO Linx GFLOPS fluctuations are more or less random, don't indicate anything wrong with OC and definitely don't have anything to do with Vcore. 

vardenkastav

XXChronicXX

Dont forget to set the amount of tests to run, or the amount of time to run. Its on the top right of linX. I like to do 3-4 tests while watching temps and GFLOPS. If the GFLOPS continue to decrease in descending order, that means your vcore is too low. 1 notch up should be enough to fix it.  You can usually tell after 3 or 4 tests.
Example of too low Vcore....

TEST 1 GFLOPS  49.8758
TEST 2 GFLOPS  49.6852
TEST 3 GFLOPS  49.4286 ---------- I usually stop it right here if i see this. When I see test 3 doesn't jump up above test 1 or 2, I bump the vcore up 1 notch.

Could you please explain why you think that ? Differences in GFLOPS depend highly on other tasks (that includes OS) that run on your CPU at give time together with Linx. The more cpu is loaded by processes other than Linx, the less GFLOPS it shows. IMHO Linx GFLOPS fluctuations are more or less random, don't indicate anything wrong with OC and definitely don't have anything to do with Vcore. 

Here is an example of a too low vcore.

49.8575
49.6328
49.3468
49.1463
48.8906
48.6085
48.4098
48.2081
47.9056
47.8472
47.4980

When your gflops dont do this. You are fine!
Don't worry if they look sloppy like this it has nothing to do with stability.

47.1231
45.8982
49.1235
48.4632
47.5643
48.2515
45.8392
49.3298

Only when they continuously drop.

Actually they were roughly 45 seconds per test. The '19' is the GFlops. But yes, I'll do the full memory test (was only doing I think 25% memory) Thanks, I'll up the vDimm to 1.65v, but am I still within the .45v range than between the mem controller and vDimm?


EDIT: I'm trying to set the QPI Link manually to 4.8, but my system is definitely not likin' it. I'll keep it on Auto for now (which is putting it @ 6.4). Anyone know if I'll see issues with the QPI Link staying @ 6.4? 

What is the stock VTT voltage? 1.2v? 

Some nice work in this thread.

Thanks guys. Glad I could help.

Ok Chronic. What do I do if my GFLOPS continually rise? I did 5 tests and every one was higher than the next. Do I lower or stay?

OK Chronic I have found out what exactly was causing my bsod's a week or ago.  And, I would like your thoughts as to why it would only effect boot-up while overclocked. 

The driver_irql_not_less_than_equal bsod I am getting is related to my printer.  It works fine at normal/stock clocks, but when I OC it causes the bsod.  The only reason I know this for sure is because I just reinstalled my printer drivers and it bsod'ed.  I couldn't even get into safe mode to remove the driver.  I thought it was SLi, but I was wrong.  I had uninstalled the printer drivers just before I re-initialized SLi.  Because the printer wasn't even available in the control panel.  I just never reinstalled them until just now.

So, I reset cmos.  Put my memory back to stock settings (7-7-7-20@1.65 v), and she booted right back into windows without a hiccup.  Printer is there and working, but if I OC it all hell breaks loose.  Any ideas?  Thoughts?  Could it be a setting I'm missing?  HP driver compatibility issue with Win7 x64?  Because this doesn't make any sense to me.  I'm going to uninstall the driver for now, and OC again.

Thanks for your time...again.
 
Oh my printer is a networked HP 3210 all-in-one.

StormTheGeek

Ok Chronic. What do I do if my GFLOPS continually rise? I did 5 tests and every one was higher than the next. Do I lower or stay?

Always try for lower. Im sure you have already
 
That GFLOP trick doesn't always work. Ive had Gflops staying level, but still crashed and ended up needing more vcore.
I just added a side note to the GFLOP section of my guide.
 

i7 Overclocking for dummies
Please Note: This Gflop trick is not consistent and has flaws, it does not always work this way.
Sometimes your gflops will look fine but you will still crash during a 100 tests with LinX. Increasing the vcore a little more will most likely allow you to pass all 100 tests.

I would still try increasing the ICH. Very weird how you found out its the wireless printer. I still am baffled by it. It seems weird that a simple printer driver would cause this, but hey, thats what we are here to learn for.

here is the guide from LinuxRouter
http://forums.evga.com/tm.aspx?m=5266