Overclocking when I don't want to.

MoridinUK

MoridinUK

MoridinUK

Associate
Joined
16 Jun 2014
Posts
396
Hey all so I seem to have an issue.

I was poking about in my old pc.

i7-7700k on an asus rog maximus VIII Gene.

It had 16gb of DDR4-3200 corsair vengeance ram. cl36
I upgraded to 32 gb corsair vengeance DDR4-3200 cl38

Both use XMP to get to their speeds. Basically just the same ram but twice as much.

I can't get the system stable.

The mainboard is using xmp to set the voltage and dram timings and has them right, I have the bios set to let the processor run at stock speeds 4.2ghz normal 4.5 boost.

It's boosting all the time according to HWMonitor.

I'm struggling what to do to try to make it stable, normal use seems ok and hyperpi runs fine, But prime95 (with AXV turned off) either stops due to rounding errors or once gave me a blue screen. I'm a bit lost as to what to try. The bios seems to want to do everything for me.

Any tips on benchmark software and monitoring software to figure this out?
Is memtest still a useful app?

The system was not 100.00000% stable when the old ram was in there, but it was very rare, it may be that I'm seeing that issue as I'm looking harder.

I also installed a new cooler and the system is running nice and cool, hyperpi can't get my processor over 52 degrees c...

I'm stumped, an actual fault on the memory is so rare and unlikely.
 
The cpu boosting part is normal, but you were crashing with your old ram and now crashing more often with your new ram, have you tried running without xmp to see if the crashing stops? it suggests the memory controller is a bit naff but hard to say with certainty.
 
Dg834man said:
The cpu boosting part is normal, but you were crashing with your old ram and now crashing more often with your new ram, have you tried running without xmp to see if the crashing stops? it suggests the memory controller is a bit naff but hard to say with certainty.
Click to expand...
I was crashing like once every week, and I had no reason to think it was ram.

I turned xmp off and set everything to auto. Prime95 got through more tests, cpu made it to a whooping great 64 degrees and then blue screen (a messed up one where you could see the blue screen and the information that the system was creating a report for microsoft, but it was all distorted and repeated.)

Tomorrow when I have more time I'll put the old ram back in and see if it can prime 95. I'd do it now but to get the fan off the cooler so I can reach the ram means removing the gpu (got to love mATX!) It's possible I've got some setting in the bios that's upsetting the cpu, I've been fiddling a lot today.

If I have the same issue with the old ram, then it's either the board or the cpu. I do have an i5-6600k and it's in another board, so I could fit it in a case and test the memory on that, but that still wont answer which component is the issue. I really don't want to swap the cpu's out again it's fiddly (though less so now I've got a better cooler in there)

Old config

i5-6600k (overclocked)
Dark rock pro 3
16gb 3200 corsair vengeance ram.
Occasional but rare crashes (assume software)

Changed to

i7-7700k (not overclocked)
Dark rock pro 3 (not happy with temps Prime95 would push cpu to 100 degrees)
16gb 3200 corsair vengeance ram.
Once a week cadence for inconsistent crash

Changed to

i7-7700k (not overclock)
Dark rock pro 5 (temps are nice now)
32 gb 3200 corsair vengeance ram
Crashes in prime95. not in hyper pi, crashed while my kid played slime rancher 2...
 
Last edited:
Clear the CMOS. If using two kits of memory install the cas 18 (?) kit first and set XMP. Test and then try the cas 16 kit at the cas 18 profile.
 
Last edited:
Quartz said:
Is the BIOS up to date?
Click to expand...
Yep latest!
jigger said:
Clear the CMOS. If using two kits of memory install the cas 18 (?) kit first and set XMP. Test and then try the cas 16 kit at the cas 18 profile.
Click to expand...
I'm using just one 2x16gb kit. :(
Quartz said:
Given how old the system is @MoridinUK might want to test and change the CMOS battery.
Click to expand...
Wouldn't I see setting being lost? Not sure how to test it but I could swap it out, it's not the original one though, I don't think.

Thanks all for the suggestions
 
MoridinUK said:
It's boosting all the time according to HWMonitor.
Click to expand...
Can you take a screen of hwinfo, 'cos it shows what things are enabled/displayed in the bios like EIST/speedshift, though always worth checking your windows power plan and be aware some monitoring software can pin your CPU.
 
Tetras said:
Can you take a screen of hwinfo, 'cos it shows what things are enabled/displayed in the bios like EIST/speedshift, though always worth checking your windows power plan and be aware some monitoring software can pin your CPU.
Click to expand...
I will do so when I next have it running I'm at work right now. I need to install HWinfo but no issue.
Quartz said:
Only if you remove the battery.



Use a multimeter. Red probe on top of the battery, black probe on a ground point.
Click to expand...

I don't have a multimeter and I also don't have a spare 2032 battery, though i think maybe I can pull one from somewhere else.


So I reset the cmos this morning, and put the old ram back in, left it on stock not xmp. It seemed stable and ran prime95 for about 40-45 mins before I got the crash with the messed up crash report screen. So that's not very encouraging. But suggests the ram itself is probably fine.

I could replace the cpu with the old i5-6600k or build a new pc with that cpu on a different board and check the ram in that, but in that situation I still don't know if the issue is a cooked board or cpu. (The i7-7700k is second hand from ebay with no returns, it's possible it got cooked or damaged, there is a strange mottling on the metal on the lid and one core runs a few degrees hotter than the others.)

Not really sure how best to proceed.
 
nah there is a little start up window that lets you check to get sensors only for _reasons_ on mine the sensors were checked.

I think you mean the summary here?

Right the bios is set to use auto for everything, I've no idea why there is XMP on some bits
 
The timings and speed for the ram is all wrong but I don't know what they should be as I only know the target timings when set to xmp (3200mhz, 16,18,36 1.35 volts)
Bios says it's running at 2133 mhz lower timings and 1.2 volts but hwinf isn't showing that
 
Tetras said:
Oh, I meant the main screen that shows the CPU area, not the sensor tab.
Click to expand...
nah there is a little start up window that lets you check to get sensors only for _reasons_ on mine the sensors were checked.

I think you mean the summary here?

Right the bios is set to use auto for everything, I've no idea why there is XMP on some bits
 
You must log in or register to reply here.
Back
Top Bottom