cancel
Showing results for 
Search instead for 
Did you mean: 

Problems achieving stable XMP settings using i7-8700k, Maximus X Apex, G.Skill 4400

Softkore
Level 7
My build:


  • Intel i7-8700k
  • 2x G.Skill Trident 8GB DDR4-4400
  • Asus Maximus X Apex


So I had an issue with a faulty DIMM slot the other day, and received my new motherboard via RMA. I rebuilt the system, put both of the RAM sticks in, updated the BIOS, and turned on XMP. Everything worked fine.

I played Destiny 2 for about two hours or so, cranked up to max and 4k resolution, with no issues at all. Then my buddy and I downloaded Intel XTU and ran some tests. My buddy was a little worried about the temps we were seeing (which I later found out are within normal ranges for my CPU, around 80C under load).



When googling about the temps, I found info on the "multicore enhancement" setting which is effectively a CPU OC setting - which is turned on by default for this motherboard (set to "Auto" but did seem to be OCing the CPU to 4.7Ghz). We turned the setting off in the BIOS, and then couldn't POST. I tinkered with the XMP setting, figuring that turning off the multicore enhancement while XMP was still turned on was causing an issue. With XMP turned off, we were able to POST with multicore enhancement also off.

I played with the RAM timings manually, but was never able to get the computer to POST with "multicore enhancement" turned off, and my RAM speed set to DDR4-4400. After reading about the temps more, I decided that having multicore enhancement turned on alongside XMP would be a perfectly stable system (temperature-wise). I decided to restore the default BIOS settings, but I noticed that they were not resetting XMP - so I cleared the CMOS. Then I turned XMP back on.



At this point, my computer should be in the same state as when it was working fine. However, at this point the system would consistently freeze on the Windows login screen, or shortly after logging in. Now my settings seemed extremely unstable even while idling, even though the exact same settings seemed perfectly stable earlier today while under load. After screwing with the settings and voltages for some time, nothing seems to resolve the issue.

Most recently I was able to achieve a seemingly stable boot when setting the RAM frequency to 4000 instead of 4400. However, I was previously able to use 4400 without any issue (used one stick of 4400 for 2 days with no issues, and used two sticks of 4400 for ~2 hours with no issues). At this point I'm just trying to get my system stable with the default BIOS settings (including multicore enhancement on Auto) and the standard XMP profile at 4400 frequency. I'm really confused about why those settings were stable under load earlier, but are no longer stable even while idling.

I was able to run a 5 minute stress test (I know that's not long but when it's unstable it seems to crash pretty much immediately) at 4000 frequency. Just now booted fine at 4300 and completed a 5 minute test at that frequency as well. I also just completed a Windows Memory Diagnostic test with no issues/errors found.
16,017 Views
12 REPLIES 12

Softkore
Level 7
My build:


  • Intel i7-8700k
  • 2x G.Skill Trident 8GB DDR4-4400
  • Asus Maximus X Apex


So I had an issue with a faulty DIMM slot the other day, and received my new motherboard via RMA. I rebuilt the system, put both of the RAM sticks in, updated the BIOS, and turned on XMP. Everything worked fine.

I played Destiny 2 for about two hours or so, cranked up to max and 4k resolution, with no issues at all. Then my buddy and I downloaded Intel XTU and ran some tests. My buddy was a little worried about the temps we were seeing (which I later found out are within normal ranges for my CPU, around 80C under load).



When googling about the temps, I found info on the "multicore enhancement" setting which is effectively a CPU OC setting - which is turned on by default for this motherboard (set to "Auto" but did seem to be OCing the CPU to 4.7Ghz). We turned the setting off in the BIOS, and then couldn't POST. I tinkered with the XMP setting, figuring that turning off the multicore enhancement while XMP was still turned on was causing an issue. With XMP turned off, we were able to POST with multicore enhancement also off.

I played with the RAM timings manually, but was never able to get the computer to POST with "multicore enhancement" turned off, and my RAM speed set to DDR4-4400. After reading about the temps more, I decided that having multicore enhancement turned on alongside XMP would be a perfectly stable system (temperature-wise). I decided to restore the default BIOS settings, but I noticed that they were not resetting XMP - so I cleared the CMOS. Then I turned XMP back on.



At this point, my computer should be in the same state as when it was working fine. However, at this point the system would consistently freeze on the Windows login screen, or shortly after logging in. Now my settings seemed extremely unstable even while idling, even though the exact same settings seemed perfectly stable earlier today while under load. After screwing with the settings and voltages for some time, nothing seems to resolve the issue.

Most recently I was able to achieve a seemingly stable boot when setting the RAM frequency to 4000 instead of 4400. However, I was previously able to use 4400 without any issue (used one stick of 4400 for 2 days with no issues, and used two sticks of 4400 for ~2 hours with no issues). At this point I'm just trying to get my system stable with the default BIOS settings (including multicore enhancement on Auto) and the standard XMP profile at 4400 frequency. I'm really confused about why those settings were stable under load earlier, but are no longer stable even while idling.

I was able to run a 5 minute stress test (I know that's not long but when it's unstable it seems to crash pretty much immediately) at 4000 frequency. Just now booted fine at 4300 and completed a 5 minute test at that frequency as well. I also just completed a Windows Memory Diagnostic test with no issues/errors found.

Softkore wrote:
My build:


  • Intel i7-8700k
  • 2x G.Skill Trident 8GB DDR4-4400
  • Asus Maximus X Apex


So I had an issue with a faulty DIMM slot the other day, and received my new motherboard via RMA. I rebuilt the system, put both of the RAM sticks in, updated the BIOS, and turned on XMP. Everything worked fine.

I played Destiny 2 for about two hours or so, cranked up to max and 4k resolution, with no issues at all. Then my buddy and I downloaded Intel XTU and ran some tests. My buddy was a little worried about the temps we were seeing (which I later found out are within normal ranges for my CPU, around 80C under load).



When googling about the temps, I found info on the "multicore enhancement" setting which is effectively a CPU OC setting - which is turned on by default for this motherboard (set to "Auto" but did seem to be OCing the CPU to 4.7Ghz). We turned the setting off in the BIOS, and then couldn't POST. I tinkered with the XMP setting, figuring that turning off the multicore enhancement while XMP was still turned on was causing an issue. With XMP turned off, we were able to POST with multicore enhancement also off.

I played with the RAM timings manually, but was never able to get the computer to POST with "multicore enhancement" turned off, and my RAM speed set to DDR4-4400. After reading about the temps more, I decided that having multicore enhancement turned on alongside XMP would be a perfectly stable system (temperature-wise). I decided to restore the default BIOS settings, but I noticed that they were not resetting XMP - so I cleared the CMOS. Then I turned XMP back on.



At this point, my computer should be in the same state as when it was working fine. However, at this point the system would consistently freeze on the Windows login screen, or shortly after logging in. Now my settings seemed extremely unstable even while idling, even though the exact same settings seemed perfectly stable earlier today while under load. After screwing with the settings and voltages for some time, nothing seems to resolve the issue.

Most recently I was able to achieve a seemingly stable boot when setting the RAM frequency to 4000 instead of 4400. However, I was previously able to use 4400 without any issue (used one stick of 4400 for 2 days with no issues, and used two sticks of 4400 for ~2 hours with no issues). At this point I'm just trying to get my system stable with the default BIOS settings (including multicore enhancement on Auto) and the standard XMP profile at 4400 frequency. I'm really confused about why those settings were stable under load earlier, but are no longer stable even while idling.

I was able to run a 5 minute stress test (I know that's not long but when it's unstable it seems to crash pretty much immediately) at 4000 frequency. Just now booted fine at 4300 and completed a 5 minute test at that frequency as well. I also just completed a Windows Memory Diagnostic test with no issues/errors found.



If pushing things as far as you are then you need to test memory stability more thoroughly. Even on the Apex, 4400 is going to be tough. You're on the very edge of what the memory controller is able to achieve. You've done well in fact to get as far as you have.

I would work with 4000Mhz testing the memory with HCI Memtest Pro. Set one instance per thread allocating 90% of available RAM across all instances. Those sticks should work well with the profiles available within the Tweaker menu.
13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

I played around with the settings a bunch more, and kept lowering the voltages - trying to find the lowest voltage which would also boot and complete a stress-test.

Weirdly enough, I was able to get everything booting again with the same XMP settings that had originally worked, and had then stopped working... So basically the results I get from the exact same settings seems EXTREMELY inconsistent, but I have heard you kind of need to "hone in" on your clock speed and voltages, so maybe that's why it works for me now? No idea.

Anyways, I booted and completed a stress test with these settings:

XMP On, DDR4-4405 profile
DRAM Frequency: DDR4-4400MHz (set by XMP)
DRAM Voltage: 1.4000 [1.392] (set by XMP)
VCCIO Voltage: Auto [1.376] (set by XMP)
System Agent Voltage: Auto [1.368] (set by XMP)
Maximus Tweak: Mode 2 (set manually)
DRAM CAS# Latency: 19 (set by XMP)
DRAM RAS# to CAS# Delay: 19 (set by XMP)
DRAM RAS# ACT Time: 39 (set by XMP)
DRAM Command Rate: 1N (set by manually)

69632

Now I'll actually try to use this for an extended period of time, put it through its paces in some games in 4k, and perhaps perform a 6-hour stress test tonight.

So I'm at 5.0GHz with the full DDR4-4400 frequency. Everything seems stable. Assuming that the overclock is rock solid...

I'm seeing some thermal throttling when stress testing. However, I did some more tests under "realistic" loads. I opened a bunch of YouTube videos in Chrome, had Discord running, and was downloading and installing Witcher 3 on Gigabit fiber. Then I played Destiny 2 in 4k on ultra settings. My framerate was around 55fps solid, and my CPU temps maxed out at 85C. Once the Witcher 3 install completed it was back down to 70ish.

While I *am* seeing thermal throttling in stress tests, it appears that temps are looking fine under realistic loads, from what I've read of Coffee Lake temps. Keep in mind I'm on air cooling (Noctua NH-D15).

Softkore wrote:
So I'm at 5.0GHz with the full DDR4-4400 frequency. Everything seems stable. Assuming that the overclock is rock solid...

I'm seeing some thermal throttling when stress testing. However, I did some more tests under "realistic" loads. I opened a bunch of YouTube videos in Chrome, had Discord running, and was downloading and installing Witcher 3 on Gigabit fiber. Then I played Destiny 2 in 4k on ultra settings. My framerate was around 55fps solid, and my CPU temps maxed out at 85C. Once the Witcher 3 install completed it was back down to 70ish.

While I *am* seeing thermal throttling in stress tests, it appears that temps are looking fine under realistic loads, from what I've read of Coffee Lake temps. Keep in mind I'm on air cooling (Noctua NH-D15).




https://rog.asus.com/forum/showthread.php?73665-Our-preferred-memory-stress-test
13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Menthol
Level 14
Softkore, welcome to the ROG forums and congratulations on purchasing the best hardware available

4400 mhz is at the very upper end of compatibility and the CPU is going to be the limiting factor more so than your Apex, can you please list all your specs and any settings in bios you have changed, also be aware that unstable memory will corrupt and OS faster than anything
Dram voltage, VCCIO, and VCCSA voltages will be need to be adjusted to achieve, at least I believe so, I don't have a 4400mhz kit to test, the fact that it booted and ran at all by only enabling XMP surprises me and tells me you may have a good CPU, VCCIO and VCCSA will probably need to be around 1.25 to 1.35 volts and 1.35 is about max for daily use, also raise dram volts a little over manufacturer specs, 0.2 to .05
I have a 4226mhz kit and around 1.25 VCCIO, 1.275 VCCSA are needed for stability
There is a memory preset in the memory timing section of the bios Raja's 4000 or 4133 at 1.4 volts that works very well for me, the timings are tighter that the factory timings, even though it may be a little lower frequency, it gives better memory bandwidth and system performance than the higher frequency

Hey Menthol, thanks so much for your response and for the warm welcome. I've had a few replied to my post (crossposted in other forums) but yours is the only one that seems to have taken into account all of the information that I provided.

Here is my full setup:

Intel Core i7-8700k 3.7GHz 6-Core
Noctua NH-D15 82.5 CFM CPU Cooler
Asus ROG Maximus X Apex EATX LGA1151
G.Skill Trident Z 16GB (2x8GB) DDR4-4400
Samsung 660 PRO 512GB M.2-2280 SSD
EVGA GeForce GTX 1080 Ti 11GB FTW3 Gaming iCX
EVGA SuperNOVA P2 850W 80+ Platinum

And your advice seems to have paid off! I was able to boot and perform a benchmark and a short 5-minute memory tests at 4400 frequency using Raja's 4133 profile as you mentioned, as well as tweaking some voltage settings. After getting that to work, I loaded factory defaults, and tried to set up the RAM timings and voltages myself - as I noticed Raja's profile was set to 3700 frequency, and I still want to try to achieve 4400 (I can achieve 4300 seemingly fine).

I loaded the factory default settings, then manually tweaked the CAS timings and DRAM/CPU voltages. I was then able to boot into Windows and login fine, so I went ahead and performed a benchmark and 5-minute Memory Stress Test in Intel XTU. Then I ran a Cache and Memory Benchmark in AIDA64, and got these results:

69630

Do these results seem reasonable for my setup? After this I tried to restart the PC to note my final BIOS settings, and was having trouble POSTing outside of BIOS safe-mode (I think that's what was happening at least, I was getting a black screen and a q code of 55 when failing to boot). I went back into the BIOS and changed these settings:

MCH Full Check: Disabled
MRC Fast Boot: Enabled

Now I'm able to boot into Windows every time without issue. My final settings are as follows:

EDIT: Read my updates below, these settings were not stable!

DRAM Frequency: DDR4-4400MHz
DRAM Voltage: 1.4500
VCCIO Voltage: 1.3000
System Agent Voltage: 1.3250
Maximus Tweak: Mode 2
DRAM CAS# Latency: 19
DRAM RAS# to CAS# Delay: 19
DRAM RAS# ACT Time: 39
DRAM Command Rate: 1N
MCH Full Check: Disabled
MRC Fast Boot: Enabled

Everything is seemingly stable. I plan to run longer-term tests to make sure. Do you think these voltages are acceptable to run 24/7, and also under heavy load daily? Thanks so much for your help!

EDIT: Read my updates below, these settings were not stable!

Okay very shortly after I posted that my PC froze, so definitely not yet stable. I turned down the VCCIO, System Agent, and DRAM voltages. Going to test with these now, will update with exact numbers and my findings...

I played around with the settings a bunch more, and kept lowering the voltages - trying to find the lowest voltage which would also boot and complete a stress-test.

Weirdly enough, I was able to get everything booting again with the same XMP settings that had originally worked, and had then stopped working... So basically the results I get from the exact same settings seems EXTREMELY inconsistent, but I have heard you kind of need to "hone in" on your clock speed and voltages, so maybe that's why it works for me now? No idea.

Anyways, I booted and completed a stress test with these settings:

XMP On, DDR4-4405 profile
DRAM Frequency: DDR4-4400MHz (set by XMP)
DRAM Voltage: 1.4000 [1.392] (set by XMP)
VCCIO Voltage: Auto [1.376] (set by XMP)
System Agent Voltage: Auto [1.368] (set by XMP)
Maximus Tweak: Mode 2 (set manually)
DRAM CAS# Latency: 19 (set by XMP)
DRAM RAS# to CAS# Delay: 19 (set by XMP)
DRAM RAS# ACT Time: 39 (set by XMP)
DRAM Command Rate: 1N (set by manually)

69631

Now I'll actually try to use this for an extended period of time, put it through its paces in some games in 4k, and perhaps perform a 6-hour stress test tonight.