i9 9900K overclocking - confused, what to do with AVX, LLC and OCCT stress test fail - Page 2 - Overclock.net - An Overclocking Community
Forum Jump: 

i9 9900K overclocking - confused, what to do with AVX, LLC and OCCT stress test fail

Reply
 
Thread Tools
post #11 of 17 (permalink) Old 05-12-2020, 03:00 PM - Thread Starter
Zireael
 
Wolfsblvt's Avatar
 
Join Date: Jun 2015
Location: Germany
Posts: 56
Rep: 1 (Unique: 1)
Hmh, even with playing around more and more with the values, I can't seem to get those OCCT tests to work.
So I guess I should mark the 5.1 GHz overclock testing as failed and not possible for me?
LLC didn't make a difference here.

Let's see how low I can get the voltage and temps with 5.0 GHz then. If I can keep the named benchmarks stable there, I guess I will stick to it and just eat the performance penalty.
Guess that's what you get for getting a flow-restricted cooling loop.

Wolfsblvt is offline  
Sponsored Links
Advertisement
 
post #12 of 17 (permalink) Old 05-12-2020, 04:30 PM
Facepalm
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 7,370
Rep: 441 (Unique: 306)
Quote: Originally Posted by Wolfsblvt View Post
Hmh, even with playing around more and more with the values, I can't seem to get those OCCT tests to work.
So I guess I should mark the 5.1 GHz overclock testing as failed and not possible for me?
LLC didn't make a difference here.

Let's see how low I can get the voltage and temps with 5.0 GHz then. If I can keep the named benchmarks stable there, I guess I will stick to it and just eat the performance penalty.
Guess that's what you get for getting a flow-restricted cooling loop.
Why are you trying to pass OCCT? What is your use criteria for your computer? Are you crunching AVX2 prime numbers for a living?
There are different levels of stability. If your computer is used for medical or scientific grade work, then yes, you should probably pass everything you can throw at it. But if you're just gaming and rendering, all you really need to pass is 8 hours of Realbench 2.56 (zero CPU Cache L0 errors in HWinfo64) and 4 hours of Cinebench R20 (3600*4 seconds, or longer) loops with zero cpu cache L0 errors. Maybe throw in some Blender Classroom loops also.

Don't get hung up on passing the most intense torture tests out there unless your system actually needs platinum level stability. If everyone did that, most 5+ ghz overclocks would be completely failing.

[email protected] ghz, RX Vega 64, 32GB DDR4, Gigabyte Aorus Master, Seasonic Platinum 1000W, Corsair 760T
Alt: MSI GT73VR Throttlebook with 7820HK @ 4.7 ghz, GTX 1070 MXM TDP mod to 230W, 32 GB RAM

Quote: Originally Posted by sakete View Post
Well, I want you to know I have an academic degree in speculation.
Falkentyne is online now  
post #13 of 17 (permalink) Old 05-12-2020, 05:27 PM
New to Overclock.net
 
criskoe's Avatar
 
Join Date: Aug 2010
Posts: 398
Rep: 5 (Unique: 4)
I definitely have to agree with Falkentyne on this. Chasing stability on these unrealistic stress tests is a waste of time. I’ve been down that path not that long ago myself and wasted hours of time doing it. I let my obsession take over. I was obsessed getting that platinum level of stability at 5ghz falkentyne talks about. And while it was a nice feeling for a short while when I obtained it. In the end It was kinda stupid. Lol. It’s really easy to get caught up in it. If you can avoid those obsessions you will be much better off. After all was said and done it got me no where really other then seeing what my cpu was capable of handling to the max. Looking back I see there was really not much value in that cause after days of running night long tests and getting platinum stability, I ended up being able to lower my voltage by a crazy amount afterwards and I was still stable during normal tasks I do. Which in turn resulted in much lower temps.

Unless you find “fun” in running stress tests constantly and pushing your OC to the brink I’d just stick with the suggested tests Falkentyne suggested.

Current
(13 items)
CPU
9900KS
Motherboard
Asus Maximus XI Extreme
GPU
Nvidia 1080Ti FE
RAM
Corsair Dominator Platinum RGB
Hard Drive
Samsung 970 EVO
Power Supply
Evga 1200 P2
Cooling
Ek Magnitude
Cooling
EK D5 Kinetic D5 TBE
Cooling
2X EK 360 PE
Cooling
EK 1080ti Full Cover Block
Cooling
AC High Flow USB
Cooling
10X Corsair 120mm ML Pro Fans
Case
Lian Li 011 Dynamic XL
▲ hide details ▲

Last edited by criskoe; 05-12-2020 at 05:33 PM.
criskoe is offline  
Sponsored Links
Advertisement
 
post #14 of 17 (permalink) Old 05-12-2020, 05:43 PM - Thread Starter
Zireael
 
Wolfsblvt's Avatar
 
Join Date: Jun 2015
Location: Germany
Posts: 56
Rep: 1 (Unique: 1)
I guess you guys are totally correct...

I am trying to pass those tests because I've read dozens and dozens of threads and sources during the last days on how to stability test and what is considered stable, and many people recommend different stuff and all.
There is no scientific stuff I need to do. Gaming mostly, game development and some rendering in the future. So CPU stability is an requirement, but not in the edge-case area.
I shouldn't worry so much on passing every test... It's just that my last pc crashed regularly with BSOD and I want to prevent that from happening.

Guess I will throw away the idea that the Linpack test has to pass. I can't even get it to work on 5.0 GHz, so it's a total waste of time.

I am following your advice now, @Falkentyne , (again), you seem so helpful through all the different forums and threads I have found.
Starting with the Cinebench r20 tests, because they've proven to weed out instable OC's quite fast for me, and than I will evaluate again if 5.1 GHz is possible for me, otherwise I'll go down to 5.0 GHz and try this one.
Got both softwares installed already, so will be quick done tonight and tomorrow during my workday.

Will get back with feedback on how it went (:

Wolfsblvt is offline  
post #15 of 17 (permalink) Old 05-12-2020, 05:58 PM
Facepalm
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 7,370
Rep: 441 (Unique: 306)
Quote: Originally Posted by Wolfsblvt View Post
I guess you guys are totally correct...

I am trying to pass those tests because I've read dozens and dozens of threads and sources during the last days on how to stability test and what is considered stable, and many people recommend different stuff and all.
There is no scientific stuff I need to do. Gaming mostly, game development and some rendering in the future. So CPU stability is an requirement, but not in the edge-case area.
I shouldn't worry so much on passing every test... It's just that my last pc crashed regularly with BSOD and I want to prevent that from happening.

Guess I will throw away the idea that the Linpack test has to pass. I can't even get it to work on 5.0 GHz, so it's a total waste of time.

I am following your advice now, @Falkentyne , (again), you seem so helpful through all the different forums and threads I have found.
Starting with the Cinebench r20 tests, because they've proven to weed out instable OC's quite fast for me, and than I will evaluate again if 5.1 GHz is possible for me, otherwise I'll go down to 5.0 GHz and try this one.
Got both softwares installed already, so will be quick done tonight and tomorrow during my workday.

Will get back with feedback on how it went (:
I would start with Realbench 2.56. RB 2.56 tends to have more violent transients, which can really weed out an overclock if your voltages are too marginal. And it also loads your GPU at the same time, so it's a good test of your +12v as well. Please do NOT rely on RB to show "instability detected." On hyperthreaded chips, you will usually get L0 errors or BSOD's long before realbench shows any instability.

CB R20 has smoother transients and a better sustained test load. You can make it loop in the settings.

AIDA64 Stress FPU is another good test to work on. These three tests have similar current loads so passing all three is a good sign. Prime95 small FFT with AVX disabled puts a similar load as Stress FPU. However Prime95 AVX disabled does not test AVX, so this should be a baseline for basic SSE2 stability.

Cinebench R20, AIDA64 Stress FPU and Realbench 2.56 all test AVX instructions (AIDA64 may need a checkbox in the settings to test AVX, check your settings).

The one huge advantage of Prime95 is you can see exactly which core/thread fails as the workers are in order of physical and logical threads. CPU Cache L0 errors shown in windows event viewer show the APIC ID in the order of physical cores/threads with ID 0 and 1 being the first two threads for the first CPU core (on hyperthreaded chips), and so on.

One trick worth considering is noting the load (die-sense or VR VOUT) voltage requirement to pass Prime95 SSE2 small FFT, where if you go any lower, you become unstable. If any AVX test drops your vcore lower than this point, you are going to insta-crash. Also if an AVX test drops the vcore to the same point as Prime95 AVX disabled AND runs hotter, you are also going to crash. Keep that in mind.

[email protected] ghz, RX Vega 64, 32GB DDR4, Gigabyte Aorus Master, Seasonic Platinum 1000W, Corsair 760T
Alt: MSI GT73VR Throttlebook with 7820HK @ 4.7 ghz, GTX 1070 MXM TDP mod to 230W, 32 GB RAM

Quote: Originally Posted by sakete View Post
Well, I want you to know I have an academic degree in speculation.

Last edited by Falkentyne; 05-12-2020 at 06:05 PM.
Falkentyne is online now  
post #16 of 17 (permalink) Old 05-13-2020, 05:11 AM - Thread Starter
Zireael
 
Wolfsblvt's Avatar
 
Join Date: Jun 2015
Location: Germany
Posts: 56
Rep: 1 (Unique: 1)
Thanks for the writedown, makes sense. Haven't bought or tested Aida yet, but I may try if it works well for finding instabilities.

I was testing a lot Prime95 without AVX and I reached high temps of 93c max, but no instability after around 8h on my latest 5.1 GHz settings.

I followed your advice to start with RealBench and set memory to 55 GB of my 64 GB RAM. After a few hours it crashed with a BSOD multiple times.
I've analyzed the log, and - referencing this helpful reddit post about BSOD error codes I think the issue at the moment might not be vcore anymore but memory voltage.
I got IRQL_NOT_LESS_OR_EQUAL two times and PAGE_FAULT_IN_NONPAGED_AREA once, once was after 30 min, the other two were after a few hours.

So I guess I should try to upp VCCIO and SA now.
VCCIO 1.2v -> 1.25v
SA 1.1v -> 1.2v

Following the advice of this chart:


In this asus forum thread: https://rog.asus.com/forum/showthrea...for-24h-7d-use

I guess I can lower those a tiny bit later if it passes, but from what I read mine were a little bit too low for my 3600 MHz RAM sticks?

EDIT:
Another PAGE_FAULT_IN_NONPAGED_AREA. Hmh.


Last edited by Wolfsblvt; 05-13-2020 at 05:42 AM.
Wolfsblvt is offline  
post #17 of 17 (permalink) Old 05-13-2020, 06:45 AM
New to Overclock.net
 
reflex75's Avatar
 
Join Date: Aug 2016
Location: Paris, France
Posts: 132
Rep: 1 (Unique: 1)
Quote: Originally Posted by Wolfsblvt View Post
Thanks for the writedown, makes sense. Haven't bought or tested Aida yet, but I may try if it works well for finding instabilities.

I was testing a lot Prime95 without AVX and I reached high temps of 93c max, but no instability after around 8h on my latest 5.1 GHz settings.

I followed your advice to start with RealBench and set memory to 55 GB of my 64 GB RAM. After a few hours it crashed with a BSOD multiple times.
I've analyzed the log, and - referencing this helpful reddit post about BSOD error codes I think the issue at the moment might not be vcore anymore but memory voltage.
I got IRQL_NOT_LESS_OR_EQUAL two times and PAGE_FAULT_IN_NONPAGED_AREA once, once was after 30 min, the other two were after a few hours.

So I guess I should try to upp VCCIO and SA now.
VCCIO 1.2v -> 1.25v
SA 1.1v -> 1.2v

Following the advice of this chart:


In this asus forum thread: https://rog.asus.com/forum/showthrea...for-24h-7d-use

I guess I can lower those a tiny bit later if it passes, but from what I read mine were a little bit too low for my 3600 MHz RAM sticks?

EDIT:
Another PAGE_FAULT_IN_NONPAGED_AREA. Hmh.
My recommendation for 3600Mhz is vccio/sa 1.20v/1.25v (sa 0.05v higher than vccio)
But when checking for CPU stability, no RAM OC: keep default 2666Mhz for the RAM (no XMP) with vccio/sa -> 0.95v/1.05v
And remember OC is sensitive to temperature...
reflex75 is offline  
Reply

Tags
9900k , Intel , llc , overclocking

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off