Maximus VI Series UEFI Guide for Overclocking

[Nalizan]

Nalizan - The Luxmark crash has to do with your gpu's. your cpu is stable, if you can play your games all day long without crashing I wouldn't be too concerned with it.


JohnMainhard - Welcome to the ROG forum !

Are you getting 124 with a BSOD? If so this is a sign of an unstable system. Are you overclocking any of your hardware?

Seems you're right. I pulled the bottom card out of the system and then ran 4 hours on RealBench without issue. I will have to test with both cards individually and see if it is one, or the other, or if it is an issue with SLI, or the second PCI slot, or PSU etc. I may finish OCing the CPU with this single GPU in first though. Thanks for pointing me in the right direction.

[Nate152]

You're welcome

Yeah let me know if you find what is causing the Luxmark crash, you're on the right track for what to test.

[JohnMainhard]

Yes, of course i overclock my 4790k on M7F. With very cold water i did 90 Minutes prime95 custom run 4.8ghz @ 1.250V, 4.9ghz @ 1.310V and 5.0ghz @ 1.380V. But now water is little warmer and i want a stable 24/7 oc @ 4.8 ghz.

I set Vcore to 1.275V. Cache 4.5ghz @ 1.200V (tested with prime 512k and 576k für 90 minutes). VCCin is set to 1.86LLC3 but 448k prime test is crashing after 15 minutes. I dont know right setting to start at for sa/ioa/iod for my 4x4 2133mhz ripjaws z. 768k and 672-720k are not running for long time... longest 18minutes. and the clue is i always get 124bsod. No 101 that would mean vccin or something else... just 124.


Greetz JM

No one?


Greetz JM

[Nate152]

Hello JohnMainhard

The 124 BSOD means you need more cpu core voltage or you can lower the cpu speed.

[Nalizan]

I resolved the Luxmark crashing. Turns out it was not one of the video cards, but a driver issue instead. I have been struggling with getting a CPU overclock to be 100% stable since. I kept pushing vcore to try and stabilize 4.5Ghz. I got as far as 1.32v (around 1.376v under load). At this level I could run RealBench stress test for 8 hours straight using all the ram and it would not crash. However, later that same day, I used the benchmark feature in RealBench to see where I was at and it BSODed. I am ok with temps at this point, but it does seem like a ridiculous amount of extra voltage and temp for just 100mhz. For this reason I decided to just drop back to 4.4Ghz and see how low temp and volts I could get there.

Long story short, I am starting to notice a pattern with my testing. I can get a vcore amount that will run the stress test and benchmark reliably (multiple 2+ hour back to back stress tests or 2 rounds of looping the benchmark 10 times in a row) at first. But, if I let it cool back down and say just browse the net for a while or have the system just sit and run at idle then if I run the same test again later it will crater. I am starting to wonder if this could be an LLC or PLL issue or something that has to do with how much overshoot or how fast etc. I wanted to get your take on this since I seem to be chasing my tail here.

As a side note, after many hours of testing, I discovered that running the benchmark through multiple loops seems to ascertain whether a setting is going to crash quicker than running the stress test.

EDIT: One other question. When I BSOD, often it will show 101 on the actual blue screen but when I get back into Windows both blue screen viewer and the "windows has recovered" window show it as a 124. Which would it be in this case? I am assuming what I see on the actual blue screen window should be what it was, but I found it odd that BSV and windows think otherwise.

[Nate152]

Hey guy !

It sounds like you're on the edge of stability. You can try setting LLC to medium or medium high (4,5 or 6). The 101 BSOD is related to the cpu, if your temps are ok try giving it just a little more vcore.

The benchmark in realbench will require a little more vcore for the encoding test, that test is a little harsher than the stress test. If you can pass the stress test for 8 hours I would call that stable.

[Nalizan]

Final result 4.5Ghz with 4.4Ghz Cache. Ram 1866Mhz

XMP Mode
Enhanced Multi Off
Per Core 45 45 45 45
VIM 1.335v
Cache 1.261v
System Agent +.209v
CPU Digital I/O +.15v
CPU Analog I/O +.15v
CPU Input Voltage 1.96v
DRAM Voltage 1.52v

I ended up using AIDA64 as well to fine tune things. It was easier to be able to just stress the cache when I was tweaking that etc.
I used HCI Memtest with 8 instances running to test the ram.
Temps in the mid 80s

Realbench score : http://rog.asus.com/realbench/show_comment.php?id=8625
Not sure why the clock speed is wrong there. I guess I could turn off power saving features and run again.

[benparker]

Hi

Im new to PC building and have a Maximus VI Impact, 2x4GB ram 1866 cl10 and a G3258. Had two bsods yesterday while using dolphin emulator about 15 mins in. Once at 4ghz and 4.2 ghz, both times the only thing I did was changing the multiplier to 40 and 42. So the next step would be to manually set voltage rate? Also I read that the G3258 only supports upto 1333mhz speed so shall lower it to that or 1600mhz? And how do you make the timings tighter?

[kaufen]

Just finished building my computer and am ready to tweak it. I had some problems with BSOD, but that was my fault. A memory stick had come loose and was causing problems while loading windows. I finally noticed in BIOS that half my memory was gone so I took both out, put them back in and all is well!

[dw2222]

The cost of ivr system in india is suitable for gaming business?