Bad bios chip, and the likely hood of degrading CPU

[SeEnCreaTive]

Built an 1600x system with an Asrock x370 Taichi

 

Originally booted up fine, worked for the day, started the overclocking process. Easy 3.9ghz at 1.375. But as soon as memory was touched, it wouldn't post. It got worse and worse and worse, then just wouldn't post all together. Quick call with newegg, bad board, RMA'd in a jiffy.

 

New board, now RGB actually works (didn't on the old board apperently), I can overclock the ram no problem now. But now I can't seem to get over 3.7ghz at 1.375, I bumped the voltage up a single notch to 1.38 and I have it happily at 3.8ghz with XMP ram. Can't seem to get 3.9 now, and 4.0 is just out of the question without dumping in voltages. Originally on the new board 3.9 was alright, started crashing, bumped it down to 3.8. Passed all tests, ran at that for a week, and started crashing. That's when I bumped the voltage up a slight bit.

 

Its a Noctua u14 so temps are more then alright. 56 degrees under Prime95, Ambient of 19C. 

 

What would be the likely hood of a bad board degrading processor as such? Give the evidence that seems to be the case...

[SquintyG33Rs]

5 minutes ago, SeEnCreaTive said:

Built an 1600x system with an Asrock x370 Taichi

 

Originally booted up fine, worked for the day, started the overclocking process. Easy 3.9ghz at 1.375. But as soon as memory was touched, it wouldn't post. It got worse and worse and worse, then just wouldn't post all together. Quick call with newegg, bad board, RMA'd in a jiffy.

 

New board, now RGB actually works (didn't on the old board apperently), I can overclock the ram no problem now. But now I can't seem to get over 3.7ghz at 1.375, I bumped the voltage up a single notch to 1.38 and I have it happily at 3.8ghz with XMP ram. Can't seem to get 3.9 now, and 4.0 is just out of the question without dumping in voltages. Originally on the new board 3.9 was alright, started crashing, bumped it down to 3.8. Passed all tests, ran at that for a week, and started crashing. That's when I bumped the voltage up a slight bit.

 

Its a Noctua u14 so temps are more then alright. 56 degrees under Prime95, Ambient of 19C. 

 

What would be the likely hood of a bad board degrading processor as such? Give the evidence that seems to be the case...

it's not always a question of degrading the CPU. the board can introduce instability on its own. with some bad solder joints or other things that add ripple to the voltage. make sure you're actually at the same settings. things like LLC and the frequency the VRM is driven at is often forgotten. but there's a good chance you can change them

[johndms]

25 minutes ago, SeEnCreaTive said:

Built an 1600x system with an Asrock x370 Taichi

 

Originally booted up fine, worked for the day, started the overclocking process. Easy 3.9ghz at 1.375. But as soon as memory was touched, it wouldn't post. It got worse and worse and worse, then just wouldn't post all together. Quick call with newegg, bad board, RMA'd in a jiffy.

 

New board, now RGB actually works (didn't on the old board apperently), I can overclock the ram no problem now. But now I can't seem to get over 3.7ghz at 1.375, I bumped the voltage up a single notch to 1.38 and I have it happily at 3.8ghz with XMP ram. Can't seem to get 3.9 now, and 4.0 is just out of the question without dumping in voltages. Originally on the new board 3.9 was alright, started crashing, bumped it down to 3.8. Passed all tests, ran at that for a week, and started crashing. That's when I bumped the voltage up a slight bit.

 

Its a Noctua u14 so temps are more then alright. 56 degrees under Prime95, Ambient of 19C. 

 

What would be the likely hood of a bad board degrading processor as such? Give the evidence that seems to be the case...

I agree with the previous reply, it may be a matter of different settings, such as LLC. It's possible the two different boards could have shipped with different bios versions, thus different settings applied with Auto/Default. So it may not have been something you did. I would recommend updating the bios to the most recent version, though.

 

If you're unfamiliar with "CPU Load-Line Calibration" (LLC), it basically counters voltage drops (vdroop) when the CPU is under load allowing you a potential higher overclock at a lower voltage. ASRock does the numbers backwards from other boards. LLC1 is the highest setting, LLC5 is the lowest. I'd recommend trying LLC2 or LLC3 and seeing if you can achieve the overclock you previously had. Note: Even though the Taichi is the best board possible for overclocking, I'd still recommend avoiding LLC1 as extreme LLC can be dangerous.

[SquintyG33Rs]

yeah like he elaborated. avoid the latest latest bios is it says it adds support for the upcoming APU because those bring loads more problems than they fix (because of the added features for the APU).

[SeEnCreaTive]

On 11/5/2017 at 6:14 PM, johndms said:

I agree with the previous reply, it may be a matter of different settings, such as LLC. It's possible the two different boards could have shipped with different bios versions, thus different settings applied with Auto/Default. So it may not have been something you did. I would recommend updating the bios to the most recent version, though.

 

If you're unfamiliar with "CPU Load-Line Calibration" (LLC), it basically counters voltage drops (vdroop) when the CPU is under load allowing you a potential higher overclock at a lower voltage. ASRock does the numbers backwards from other boards. LLC1 is the highest setting, LLC5 is the lowest. I'd recommend trying LLC2 or LLC3 and seeing if you can achieve the overclock you previously had. Note: Even though the Taichi is the best board possible for overclocking, I'd still recommend avoiding LLC1 as extreme LLC can be dangerous.

 

 

Turns out on this board (X370 Tiachi), even with manual multiplier settings and setting voltages, it insists in changing those voltages through an auto function. Kept running it at 1.408. So I flipped it over to manual offset, and magic. Just shortly after I posted this.... 4.0 ghz at 1.38v. That cleared prime95 for 20 minutes, and AIDA64 for 20 minutes. Pretty darn good if I do say so myself. I have 0 clue why that happens. I don't think it board switches the voltages on the fly

 

That worked for about 4 days, no issues. 5th day, just didn't post. Posted once, restarted just to see what would happen, and that was that wouldn't post again. So pressed the CMOS reset, pull up my last profile and reapply it, but bumped the voltages up a single notch, now 1.392. LLC I have set also bumped up to Level 3. (AsRock boards are backwards, Level 1 or off is the most aggressive). Voltages still droop a tiny bit, but haven't had any issues under any sort of load.

 

The concerning part is again today, didn't post. Throwing a code 68 like always. Hit the power button again, started up just fine. Not sure. Its just so inconsistent. PSU is a very nice EVGA 650w P2. Once its on and running I can throw what ever I'd like at it and be fine. Its always just on posting.

 

 

Edit: I shut it down to play with my loadline calibration. I lowered it to level 4. I can't get it to post at 4.0ghz at all now.

EDIT 2: Put it back up to 3, I also bumped Vcore to 1.375 from 1.35, offset up to 0.03125, for 1.406v total. Seems alright so far, Cinebench: 1340, Passmark: 14668

 

EDIT 3: 4.0ghz is now gone. 3.9ghz is the new norm I guess.

[johndms]

4 hours ago, SeEnCreaTive said:

 

 

Turns out on this board (X370 Tiachi), even with manual multiplier settings and setting voltages, it insists in changing those voltages through an auto function. Kept running it at 1.408. So I flipped it over to manual offset, and magic. Just shortly after I posted this.... 4.0 ghz at 1.38v. That cleared prime95 for 20 minutes, and AIDA64 for 20 minutes. Pretty darn good if I do say so myself. I have 0 clue why that happens. I don't think it board switches the voltages on the fly

 

That worked for about 4 days, no issues. 5th day, just didn't post. Posted once, restarted just to see what would happen, and that was that wouldn't post again. So pressed the CMOS reset, pull up my last profile and reapply it, but bumped the voltages up a single notch, now 1.392. LLC I have set also bumped up to Level 3. (AsRock boards are backwards, Level 1 or off is the most aggressive). Voltages still droop a tiny bit, but haven't had any issues under any sort of load.

 

The concerning part is again today, didn't post. Throwing a code 68 like always. Hit the power button again, started up just fine. Not sure. Its just so inconsistent. PSU is a very nice EVGA 650w P2. Once its on and running I can throw what ever I'd like at it and be fine. Its always just on posting.

 

 

Edit: I shut it down to play with my loadline calibration. I lowered it to level 4. I can't get it to post at 4.0ghz at all now.

EDIT 2: Put it back up to 3, I also bumped Vcore to 1.375 from 1.35, offset up to 0.03125, for 1.406v total. Seems alright so far, Cinebench: 1340, Passmark: 14668

 

EDIT 3: 4.0ghz is now gone. 3.9ghz is the new norm I guess.

I have one of the B350 ASRock boards and I've learned not to shut it off or there's a good chance it'll reset my overclocks. I've gotten in the habit of just putting it to sleep mode (not hibernate). Many claim bad experiences with Sleep, but I've been doing it several times a day for six months and have had absolutely no negative side-effects. Regarding my overclocks, however, I've noticed that when I push my 3000 rated ram kit to 3200, it tends to reset more often. If I leave it at 2933, even after shutdown it tends to keep my settings.

 

I think your Taichi has LLC options for both CPU and SOC. SOC is generally used to increase stability for ram overclocks, but it couldn't hurt to set both LLC options to Level 3 (even Level 2 may be okay). I'm wondering if the ram overclock may be the part failing before post as that'll wipe both cpu/ram overclocks. But that doesn't explain why you can't do 4GHz now, though. As a test, try lowering your memory frequency from 3200 to 3066 for example (I don't know what your kit is) and see if you still experience the overclock resets after shutdown/restart. One frequency setting below XMP.

 

http://www.asrock.com/MB/AMD/X370 Taichi/index.asp#BIOS - You should be able to use 3.20 safely, if you aren't already. I'm going to avoid the next batch of bios updates (agesa 1007) for a while as they'll contain a new re-work of the bios for the new and future AMD processors. Rumor seems to be that it'll bring a lot more issues than it'll fix. But the agesa 1006 updates should be fine and may improve your stability.

 

For the record, I was able to get 1321 Cinebench at 3.95GHz, but I needed 1.45v for that to be somewhat stable. My board doesn't have LLC.

[Binder]

This inconsistent posting sounds like oc instability ...especially if its gone with llc/core clock tweaking.

BTW, what's your soc voltage ?