• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

SR-2 Optimization Thread

It probably is voltage related, but I have a lot of USB devices, and I don't want the IOH over 1.4 volts. I will try turning the clock back up and using the connector after this current WU has finished.

I have a SB Matx coming, and I will be moving most of my stuff to that, so the SR-2will not be encumbered with so much junk and can render and fold in peace.
 
I can verify the issue on mine with multiple keyboards and mice, yet other devices have no issues
 
I sometimes notice the cursor of my wireless mouse is a little jumpy, but it's usually while running F@H and other stuff (like Youtube or Flash games) so I'm not sure it's a hardware/overclock issue or not. I do have the extra floppy power connector hooked up. I think there's been some discussion of this issue in the EVGA SR-2 forum.
 
Kendrak said:
If the NICs are powerd off the same loop that the USB (and thus this floppy power header) are on, it might fix the network drop issue.
Click to expand...

I believe Tobit already tried this with the NIC issue, and had no luck. Also per Tobit, the NIC issue appears to be heat-related. That does make sense from my experience since I didn't start running into issues with one of mine until I sandwiched it between two others in my rack.
 
So i noticed something new last couple of days.

i have been till now unsuccessful in disabling NUMA and for my dual quad cpu's my ppd was around 87K and once in a while went to 90K.

so while playing arnd with the memory i changed the MCH strap to "DRAM ratio" from 1600 based on the MIBW template. and WOW my PPD shot to 97K from 87K - a 10K jump and my NUMA is still enabled.

am i missing something here or is it just a fluke??
 
F@H is very sensitive to RAM speed. The only reason not to run at DRAM Ratio is if you need to slow down your RAM in order to get a higher overclock. You will get better PPD with faster RAM, and you should also get an extra PPD boost if you could disable NUMA.
 
juanP said:
So i noticed something new last couple of days.

i have been till now unsuccessful in disabling NUMA and for my dual quad cpu's my ppd was around 87K and once in a while went to 90K.

so while playing arnd with the memory i changed the MCH strap to "DRAM ratio" from 1600 based on the MIBW template. and WOW my PPD shot to 97K from 87K - a 10K jump and my NUMA is still enabled.

am i missing something here or is it just a fluke??
Click to expand...

All you really did was tighten you memory timings. A 10K ppd jump is more than I would have expected, but I guess its not completely outrageous.

What I want to know is, what kind of speeds are you running your quads at to get even 87K ppd with NUMA enabled? Mine at 4.14 GHz with NUMA disabled only do 92K. Yours must be flying.
 
my dual 5620 are set at 4008mhz set at 211 x 18, uncore 18x, mch - dram ratio.
my memory is set at - 8-9-8-24- 1T . this is my 24/7 rig. my vcore - 1.34 and vtt - 1.35, NUMA disabled
 
Last edited:
Hey guys. After another week or more of testing things, I feel I'll have to stick with sitting at 200 BCLK / 3.6Ghz (3.8Ghz with turbo on). I've been able to run at 210, even pass crazy Linpack CPU benchmarks at 210 but anything with graphics and I/O simply seems to lock the system, leading me to believe it's a NB/SB or board issue rather than the processors themselves. Only could even complete half my graphs at 4.0 since not all games and such would want to run. I've tried loose timings, tight timings, MCH straps and Uncore all over the place, and many different voltages but all to no for sure rock solid stability.

In the meantime though, I finally popped in this 2nd 580 that's been a fun time as this is RMA card #3 now... Had one quick lock in SCII but I also saw afterwards upon reboot that the card was well past 90C so guessing it's heat related (I had forgotten to put the fan speed higher). Now that that's fixed, it's been doing game benchmarks, folding etc so far without any problems for now. Good news is I'm even under my UPS limit :)

I don't know. I mean I could tweak more (and will probably at least post on OCN again with my stats for testing at 210) but to spend the bulk of my OC'ing time trying to bump 10 BCLK more as opposed to getting almost 70 without massive headache just seems to me like I've hit that point of diminishing returns and a limit. Still though... plenty happy with the system being at 3.8! Not bad for my first OC IMO.

I guess final for me is:
BCLK 200
Freq -> BCLK = 200 / PCIe = 100 / Multi = 18x
QPI = 4.8G (Auto) / Memory = 1,066
Strap = DRAM / UCore = Auto
CPU Options -> (D)isabled (E)nabled (A)uto = D, E, E, D, E, E, A, D, E, E, D, E
Memory -> 8-8-8-20-8-59-4-4-4-20-1(T)-0(Auto)-55-57-59
Volts-> Vcore 0/1 = 1.31250 (All) / VTTcore 0/1 = 1.375 (All)
DIMM = 1.65 (All) / All other settings = Auto
Signals -> -80, -15, -31, -31, -31, -31, A, A, -31, All rest on Auto

I believe IOH is at 1.35 as it reports 1.32 in Eleet and I know a few things there are 0.03v off. All good,
 
Last edited:
Hi guys,

been reading and reading...and I am getting dizzy I have been reading MIBW's many good settings but I still seem to have system freeze or the mouse go all funny, as if it had exploded, pixels spread out. BTW i am using a tablet Wacom L.

I am currently @ 3.8 20x190, I have a pair of Corsair A70 coolers, and 6x 8gb Kingston/Hynix ECC ram.

What are the latest best settings for these CPU's I would really appreciate some help, please

thanks all, this is one amazing thread going on here! WOW!

Henrik
 
Hi Henrik - can you post some settings - what ram timings, what vcore, vtt etc?

- my first thought is that with 48 GB of RAM you will face some issues I didnt. I would get HCI Memtest Pro and test your ram first. (run heaps of instances from a shortcut with a /t2048 flag until you only have about 1-2G free ram - run for 200% at least)

You might try MCH strap of 1600 to loosen up, and be at 1.375v vtt if you are not already... more tomorrow..
 
Hi MIBW,

thanks very much for getting back to me. I have been trying different things all day yesterday. your settings, Shimano's settings default, and now I am getting Stop Error 50! PAGE_FAULT_IN_NONPAGED_AREA

So I think I need to re-install win7 x64. perhaps check and reseat my CPU's with fresh tim, and check my memory modules are seated properly. Test the memory as you suggested, getting that program now.

How do I embed images, such as screen shots? (i know noob?)

thanks

Henrik
 
Is this getting errors once you have loaded optimised defaults?

(is this an overlclocking problem - or problem at stock?)

I would guess memory problem before reinstalling anything...

Upload screenshots to the website of your choice - can't host them here - then embed using the yellow icon and paste the direct link in there - if all too hard just include a link to a gallery and we will work it out.
 
yes, I was running everything at Optimised defaults, mainly because I have some work to be done and didn't want it to freeze on me or stall in the middle of doing thing.

ATM it won't even repair...I boot up and i get the choice to repair or normal, both..gives me stop 50!

thanks very much for your help its much appreciated

Henrik
 
Ouch. I would try loading optimised defaults, but manually set MCH strap to 1600, and try to boot.

Next keep the above and set vtt to 1.375 volts.

Next try setting command rate to 2T.

Next try setting MCH to 1333, then reboot - go in and see what the memory timings are set to, and then manually set them to fix their value. (except back to back cas delay - leave on 0 = auto) Then set MCH strap to 1600 and try to boot to windows.
 
thanks.....ohhhh, I didn't like that Ouch! ...breaking out in sweat now!
 
Don't worry - we will get there, only a matter of how much pain.:p

But very likely due to ram...
 
tried them all, and I get to the windows 7 progress bar and then i get the error stop 50

Pain alright! ;-)

ok, now what do i do next?
 
hmm, my initial reaction is to pull things out and slowly add in one by one.

got the memTest program! but atm can't really use it ;-(

thanks Mate, for hanging in there with me, I very much appreciate it
 
Pull out all your memory and start with just two sticks. If it won't boot, try two diffeent sticks and so on. If you can't get a combination to boot, disable one of the processors and try to boot with a single stick.

But before you do that, did you disable NUMA in the bios. If you did, re-enable it and try to boot first.
 
I tried Safe mode no luck

I was trying to apply your settings and Shimano's setting, trying to see if I could stop it from freezing, it seems to happen after a little while working in Photoshop CS5 and Bridge anywhere from 5min to 45min.
still didn't seem to make much difference, so I set it to default and left it last night, this morning it had frozen. The last thing I looked at last night was my raid controller...it was performing a Patrol read to all the drives. Honestly do not know what that is. This is using LSI MegaRaid Storage Manager. I was looking at Eleet too and temps were in the high 40's to low 50's for the hotter cpu.

I have turned off the beast and pulled my ram out to reseat it

they are Kingston KVR1333D3D4R9S/8GI 1.5V (It has HYNIX on it as well)
 
2x 8gb and I am into the GUI of startup repair, well that is progress ;-)
 
Last edited:
I also removed my RAID array to simplify it - went through repair and it restored to a previous point and has booted up ;-) yubii

i will now keep adding ram

I have also placed the case horizontal... temps are in high 20's on CPU #1 in Eleet
 
4x 8GB installed
ELEET
core speed 3063.1 Mhz x23 @ 133.2Mhz
Memory: 9 9 9 24 107 1T @ 665.9MHz 2:10
CPU0 CPU1
VCore 0.95V 0.92V
VTT 1.36V 1.34V
IOH vcore 1.08V
IOH PLL 1.73V

CPU0- the hot one 29 20 29 25 23 27 C
CPU1 - cooler 24 25 22 21 29 27 C
 
6x 8gb all install and booted into windows.
using your last settings

Thanks

Now, I will re-insert my Raid controller...it could be my downfall here

PS: Still running at default speed/turbo 3ghz x23
 
I would run memtest to be sure, although now it seems that other things may have been responsible - numa/raid etc.
 
Thanks MIBW for your help. I will run the memtest as you said, just to be sure.

I have just tried to push the system with a 23GB pano to stitch together, which is going fine, but at 2.9-3 Ghz is a lot slower then 3.8 Ghz ;-)

but I think it was more when I was pushing it with PS and bridge, loading lots of files and bridge had to access its cache

i'll post later today - I would really appreciate if you could help me along with a higher clock speed
that would be really great, thanks

Henrik
 
so far so good, no crashes - 15x Memtest running at about 80% and one at 370% NO ERRORS!! using 47.7 GB RAM of 48... cpu pumping at 60+% ...5 hours and counting ;-)
been working in PS and Bridge and it has been steady - temps at around 40-46 max on the hot cpu

I am very pleased with your help and the result so far. now fingers crossed when we start to OC it.

One thought, do you think some of these instabilities could be because of the CPU' coolers not connecting well enough to the CPU or that there could be some warping in the board and the RAM isn't connecting when the Case is in vertical position vs. the current horizontal position..eg.its lying on my desk ;-) I know I am desparetly stabbing out in thin air here.

Thanks again for all your help
 
Based on the temperatures you are seeing, i don't think that CPU cooler contact is the issue. I have heard reports of board flex issues that usually show up as memory issues. One way to find out - set the case up and run the tests again. Also look and see if one of the corners is pulling away from the MB tray.
 
Hi guys,

I spoke too soon yesterday, shortly after. I decided that I better copy my work to another set of drives, I had the Memtest going, then the machine froze up. Damn.

I restarted but pulled my RAID card out, and managed to boot up, after a few stop 50 errors.
manged to finish my job, started the Memtest again and it ran all night, at about 110%-127% had about 15 windows open. no errors.

I restarted the computer with the RAID card in, and I got error 50 again...and its been going on and on all morning. I have some data that I need for a meeting this avo, so i am a little stressed.

without the raid card, i have managed to get back in to windows a few times, sometimes it have had to run the repair, some of those times, it has both failed and other times it has completed it successfully. Its a bit inconsistant.

I am looking over the board to see if anything is looking odd, and it all looks fine. I have in the above, taken out all the ram and installed them again, inbetween trying to run.

I have no idea where this is coming from? something about a bad device driver showed up, but the error 50 does not refere to anthing specific, it does have (xxxxxxxxxxxxxxx,xxxxxxxxxxxxx,xxxxxxxxxxxxx,xxxxxxxxxxxx) I have paid attention to these numbers yet. but I have had a look online of others where there had been a reference to a specific file/driver.

I am feeling a bit lost, I have also reset the bios and entered the settings we went through yesterday but no difference.

buggar

thanks guys for the help and takin the time to help me, i do really appreciate it. thanks

Henrik
 
Sorry to hear that Henrik, but it looks like the RAID card/drivers. Unfortunately these are beyond my experience - I suggest you post ASAP on the EVGA SR2 forum and see if you can get some help.

Good luck with your meeting - we have all been there.
 
thanks MIBW,

called and cancelled, ain't no good stressed out with only half the job.

thanks
 
Good move - I learnt the less clients see of the technical headaches we deal with the better.
 
ohh yeah, but it was bad enough to cancel, little do they need to know that things can go wrong.

I am seriously thinking I might be better off re-installing and start on a fresh, thats only half a days work, where I could be going back and forth for days as I am at the moment scratching the walls for clues.

Now, if reinstalling does not help, then I am in trouble!
 
Any chance popping the raid card in another machine, putting a temp HDD in the SR2, copy data across, finish job?
 
You must log in or register to reply here.
Back
Top