Karhu vs TM5 vs HCI Memtest + 9900KS anomaly? - Overclock.net - An Overclocking Community
Forum Jump: 

Karhu vs TM5 vs HCI Memtest + 9900KS anomaly?

Reply
 
Thread Tools
post #1 of 13 (permalink) Old 05-22-2020, 12:37 AM - Thread Starter
New to Overclock.net
 
Kargeras's Avatar
 
Join Date: May 2010
Posts: 179
Rep: 1 (Unique: 1)
Karhu vs TM5 vs HCI Memtest + 9900KS anomaly?

Hello everyone,

I'm trying to determine if my 64GB RAM kit is 24/7 stable, rig is in the signature.
The configuration I'm trying to find memory stability is is:

9900KS + Gigabyte Aorus Master + BIOS F11e
MCE: Disabled
vCore: 1.2v Adaptive+75mv
CPU ratio: 50
AVX : 1
Uncore ratio: 45
XMP: Enabled (no overclock)

Memory kit: G.Skill Ripjaws 4x16GB DDR4 3200 Mhz - [email protected] (F4-3200C14Q-64GVK)
DDR Voltage: 1.35v
DDR Training Voltage: 1.35v

VCCIO: 1.2v
VCCSA: 1.25v

Intel Speed Step: Disabled

AC Loadline: Power Saving
LLC: Standard
EIST: Enabled
CSTATES: Auto


This particular config passed any CPU stress test I threw at it: Prime95 v29.8 build 6 (Small with or without AVX, Large with AVX, FFTs in place 112k-112k no AVX), AIDA FPU, AIDA all + GPU, OCCT, Blender for 4h+, LinX (7/10 residuals match).

For years I've been using the HCI Memtest and found it to be very reliable.
With 9900KS I heard word around that people were using some other tools for the exact same purpose, namely Karhu and TM5.
Gave both of them a try and noticed the following, when Uncore multiplier is 45:

1. Karhu passes 8775%+

2. TM5 + extreme .cfg by anta777.cfg reports 1 error per cycle

3. HCI Memtest Pro 5.0 reported 1 error at 34%?

4. HCI Memtest Pro 7.0 reported error at 153.6%?

If Uncore is 43, there are no errors in any of the 3 variants (Karhu, TM5, HCI Memtest Pro v5.0 or v7.0)
That I get.

What I don't get is why for Uncore 45 I pass Karhu 9000% (well beyond their recommendation) and fail the other 3.
I'm not really comfortable with the idea of HCI failing. And that particular idea is backed up by TM5 too, albeit with a special .cfg which is supposed to hit harder.
I haven't tried the default .cfg for TM5.

Is Karhu too soft on the memory maybe?

Also, another thing that I don't get is the 50+ and even 51+ multiplier-like CPU Effective Clock values applied to some cores in Karhu and both HCI versions.
I marked those occurrences with X in the attached images, first tab.

I'm thinking that I get errors in HCI because of those freak CPU Effective Clock values but then again, I get them in Karhu too and it ran 9000% with 0 errors.

Any input is appreciated, thank you kindly.
Attached Thumbnails
Click image for larger version

Name:	01 - 1.2vADA+75v - BIOS F11e - AVX 1 - UNC 45 - XMP 1 - IO1.2+SA1.25 - LLC STD - EIST 1+CSTATES .jpg
Views:	44
Size:	639.7 KB
ID:	348084  

Click image for larger version

Name:	02 - 1.2vADA+75v - BIOS F11e - AVX 1 - UNC 45 - XMP 1 - IO1.2+SA1.25 - LLC STD - EIST 1+CSTATES .jpg
Views:	16
Size:	730.0 KB
ID:	348086  

Click image for larger version

Name:	03 - 1.2vADA+75v - BIOS F11e - AVX 1 - UNC 45 - XMP 1 - IO1.2+SA1.25 - LLC STD - EIST 1+CSTATES .jpg
Views:	31
Size:	663.3 KB
ID:	348088  

Click image for larger version

Name:	04 - 1.2vADA+75v - BIOS F11e - AVX 1 - UNC 45 - XMP 1 - IO1.2+SA1.25 - LLC STD - EIST 1+CSTATES .jpg
Views:	32
Size:	653.8 KB
ID:	348104  


Intel 9900KS + Thermal Grizzly Kryonaut + Kraken X62 AIO @ 2x140mm (~1800RPM) front push
Aorus Master @ BIOS F11e + Seasonic Prime Titanium 1000w PSU
G.Skill Ripjaws 4x16GB DDR4 | 3200Mhz @ [email protected]6v (F4-3200C14Q-64GVK)
Phanteks Eclipse P600S | front lid half-open | top lid closed
Fans 1x140mm (~1200RPM) front intake | 2x140mm top exhaust (silent ~600RPM) | 1x140mm (silent ~700RPM) back exhaust
ALL fans RPMs (including Kraken X62's) are stated during maximum load.

Last edited by Kargeras; 05-24-2020 at 11:14 AM.
Kargeras is offline  
Sponsored Links
Advertisement
 
post #2 of 13 (permalink) Old 05-22-2020, 07:29 PM - Thread Starter
New to Overclock.net
 
Kargeras's Avatar
 
Join Date: May 2010
Posts: 179
Rep: 1 (Unique: 1)
Nobody?

Intel 9900KS + Thermal Grizzly Kryonaut + Kraken X62 AIO @ 2x140mm (~1800RPM) front push
Aorus Master @ BIOS F11e + Seasonic Prime Titanium 1000w PSU
G.Skill Ripjaws 4x16GB DDR4 | 3200Mhz @ [email protected]6v (F4-3200C14Q-64GVK)
Phanteks Eclipse P600S | front lid half-open | top lid closed
Fans 1x140mm (~1200RPM) front intake | 2x140mm top exhaust (silent ~600RPM) | 1x140mm (silent ~700RPM) back exhaust
ALL fans RPMs (including Kraken X62's) are stated during maximum load.
Kargeras is offline  
post #3 of 13 (permalink) Old 05-22-2020, 08:06 PM
Facepalm
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 7,836
Rep: 455 (Unique: 316)
Quote: Originally Posted by Kargeras View Post
Nobody?
Your cache is unstable at x45. Could be an issue with the IMC signal rails. Try increasing VCCIO, which powers the L3 cache and the pre-input buffers of IMC, by 50mv. If that doesn't help, increase both VCCIO and VCCSA by 50mv.

[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 offline  
Sponsored Links
Advertisement
 
post #4 of 13 (permalink) Old 05-22-2020, 08:20 PM - Thread Starter
New to Overclock.net
 
Kargeras's Avatar
 
Join Date: May 2010
Posts: 179
Rep: 1 (Unique: 1)
I was beginning to lose hope on receiving an answer, having put the same question in different threads.

Oh, thank you very much for your input.

Intel 9900KS + Thermal Grizzly Kryonaut + Kraken X62 AIO @ 2x140mm (~1800RPM) front push
Aorus Master @ BIOS F11e + Seasonic Prime Titanium 1000w PSU
G.Skill Ripjaws 4x16GB DDR4 | 3200Mhz @ [email protected]6v (F4-3200C14Q-64GVK)
Phanteks Eclipse P600S | front lid half-open | top lid closed
Fans 1x140mm (~1200RPM) front intake | 2x140mm top exhaust (silent ~600RPM) | 1x140mm (silent ~700RPM) back exhaust
ALL fans RPMs (including Kraken X62's) are stated during maximum load.
Kargeras is offline  
post #5 of 13 (permalink) Old 05-22-2020, 08:31 PM - Thread Starter
New to Overclock.net
 
Kargeras's Avatar
 
Join Date: May 2010
Posts: 179
Rep: 1 (Unique: 1)
I forgot to mention that I did play with VCCIO and VCCSA today, after posting here.
I either lost matching residuals in LinX or simply failed LinX altogether.

I'm growing more and more confused

Intel 9900KS + Thermal Grizzly Kryonaut + Kraken X62 AIO @ 2x140mm (~1800RPM) front push
Aorus Master @ BIOS F11e + Seasonic Prime Titanium 1000w PSU
G.Skill Ripjaws 4x16GB DDR4 | 3200Mhz @ [email protected]6v (F4-3200C14Q-64GVK)
Phanteks Eclipse P600S | front lid half-open | top lid closed
Fans 1x140mm (~1200RPM) front intake | 2x140mm top exhaust (silent ~600RPM) | 1x140mm (silent ~700RPM) back exhaust
ALL fans RPMs (including Kraken X62's) are stated during maximum load.
Kargeras is offline  
post #6 of 13 (permalink) Old 05-22-2020, 08:39 PM
Facepalm
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 7,836
Rep: 455 (Unique: 316)
Quote: Originally Posted by Kargeras View Post
I forgot to mention that I did play with VCCIO and VCCSA today, after posting here.
I either lost matching residuals in LinX or simply failed LinX altogether.

I'm growing more and more confused
LinX is difficult to pass. And if you can't use x45 cache, but x43 is working, then it means maybe your CPU isn't capable of it. If it works at x43 but not x45, you need to either raise CPU Vcore, IO or SA voltages. There's nothing else you can do.
I am not a memory overclocker. I can't help you with memory overclocking or those memory stress tests you are using. Please ask these questions in the DDR4 memory thread if you can.

[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 offline  
post #7 of 13 (permalink) Old 05-22-2020, 09:23 PM - Thread Starter
New to Overclock.net
 
Kargeras's Avatar
 
Join Date: May 2010
Posts: 179
Rep: 1 (Unique: 1)
Quote: Originally Posted by Falkentyne View Post
LinX is difficult to pass. And if you can't use x45 cache, but x43 is working, then it means maybe your CPU isn't capable of it. If it works at x43 but not x45, you need to either raise CPU Vcore, IO or SA voltages. There's nothing else you can do.
I am not a memory overclocker. I can't help you with memory overclocking or those memory stress tests you are using. Please ask these questions in the DDR4 memory thread if you can.
I think I asked everywhere but there in the DDR4 24/7 thread, where it was most suite maybe.

And thank you very much for taking notice of my difficulties, your guidance here and many other places I find to be invaluable.

I will do as you advised and report back in the Z390 Aorus Master thread.
First, going back to fixed VID in BIOS as you suggested.

Intel 9900KS + Thermal Grizzly Kryonaut + Kraken X62 AIO @ 2x140mm (~1800RPM) front push
Aorus Master @ BIOS F11e + Seasonic Prime Titanium 1000w PSU
G.Skill Ripjaws 4x16GB DDR4 | 3200Mhz @ [email protected]6v (F4-3200C14Q-64GVK)
Phanteks Eclipse P600S | front lid half-open | top lid closed
Fans 1x140mm (~1200RPM) front intake | 2x140mm top exhaust (silent ~600RPM) | 1x140mm (silent ~700RPM) back exhaust
ALL fans RPMs (including Kraken X62's) are stated during maximum load.
Kargeras is offline  
post #8 of 13 (permalink) Old 05-24-2020, 11:09 AM - Thread Starter
New to Overclock.net
 
Kargeras's Avatar
 
Join Date: May 2010
Posts: 179
Rep: 1 (Unique: 1)
Oh wait, I had posted in the DDR 24/7 thread but nobody answered... yet.

Otherwise, in case anyone will be reading this, I'm still running on adaptive voltage but changed the DRAM voltage from 1.35v to 1.36v.
I'm just trying to obtain stability at standard XMP settings with this kit: G.Skill Ripjaws 4x16GB DDR4 3200 Mhz - [email protected] (F4-3200C14Q-64GVK).

Raising the voltage to 1.36v and now running HCI Memtest Pro v7.0, almost 100% coverage now and... no more weird behavior for the Core Effective Clocks.
It already passed TM5 + Extreme1 Anta .cfg with 200% coverage.

I was not able to do that before, with 1.35v.

And yeah, Karhu is way softer on the memory that it should be.

HCI Memtest remains the golden standard in Windows.
Probably TM5 + Extreme1 Anta is similiar if not identical.
Attached Thumbnails
Click image for larger version

Name:	x.jpg
Views:	7
Size:	736.0 KB
ID:	348970  

Click image for larger version

Name:	y.jpg
Views:	4
Size:	732.3 KB
ID:	348972  


Intel 9900KS + Thermal Grizzly Kryonaut + Kraken X62 AIO @ 2x140mm (~1800RPM) front push
Aorus Master @ BIOS F11e + Seasonic Prime Titanium 1000w PSU
G.Skill Ripjaws 4x16GB DDR4 | 3200Mhz @ [email protected]6v (F4-3200C14Q-64GVK)
Phanteks Eclipse P600S | front lid half-open | top lid closed
Fans 1x140mm (~1200RPM) front intake | 2x140mm top exhaust (silent ~600RPM) | 1x140mm (silent ~700RPM) back exhaust
ALL fans RPMs (including Kraken X62's) are stated during maximum load.
Kargeras is offline  
post #9 of 13 (permalink) Old 05-24-2020, 03:33 PM
Facepalm
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 7,836
Rep: 455 (Unique: 316)
Quote: Originally Posted by Kargeras View Post
Oh wait, I had posted in the DDR 24/7 thread but nobody answered... yet.

Otherwise, in case anyone will be reading this, I'm still running on adaptive voltage but changed the DRAM voltage from 1.35v to 1.36v.
I'm just trying to obtain stability at standard XMP settings with this kit: G.Skill Ripjaws 4x16GB DDR4 3200 Mhz - [email protected] (F4-3200C14Q-64GVK).

Raising the voltage to 1.36v and now running HCI Memtest Pro v7.0, almost 100% coverage now and... no more weird behavior for the Core Effective Clocks.
It already passed TM5 + Extreme1 Anta .cfg with 200% coverage.

I was not able to do that before, with 1.35v.

And yeah, Karhu is way softer on the memory that it should be.

HCI Memtest remains the golden standard in Windows.
Probably TM5 + Extreme1 Anta is similiar if not identical.
I spent all day yesterday (literally, ALL DAY) testing the Z490 Master and Maximus 12 Extreme, and I think I may know precisely what is going on with LinX residuals, and how to stabilize them, but first I need to report my findings to the Asus engineer. What I can tell you is that the Master and Extreme had exactly the same response to the tweaks, and I can also tell you that when dealing with LinX, EVERYTHING matters--your RAM, your timings, your RAM voltage, your RAM speed, your system agent and io voltage, gflops speed, AND your CPU voltage / AVX2 stability! What I do not know and can't determine is if my VCCIO/VCCSA findings for Z490 do the same as on Z390 or not as I've already found in my tests it was the opposite when dealing with LinX...

[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 offline  
post #10 of 13 (permalink) Old 05-24-2020, 07:04 PM
New to Overclock.net
 
Join Date: May 2020
Posts: 25
Rep: 0
Quote: Originally Posted by Falkentyne View Post
I spent all day yesterday (literally, ALL DAY) testing the Z490 Master and Maximus 12 Extreme, and I think I may know precisely what is going on with LinX residuals, and how to stabilize them, but first I need to report my findings to the Asus engineer. What I can tell you is that the Master and Extreme had exactly the same response to the tweaks, and I can also tell you that when dealing with LinX, EVERYTHING matters--your RAM, your timings, your RAM voltage, your RAM speed, your system agent and io voltage, gflops speed, AND your CPU voltage / AVX2 stability! What I do not know and can't determine is if my VCCIO/VCCSA findings for Z490 do the same as on Z390 or not as I've already found in my tests it was the opposite when dealing with LinX...
Lol care to share your "findings"? I have 10980xe which seemed easy to pass linpack at any given clock as long as enough vcore being supplied even when it is throttling compared with mainstream CPU. With mainstream CPU it really difficult to keep residual constant even at stock setting. I wonder if the implementation of cache vs mesh on HEDT is causing this issue on the mainstream but not on the HEDT.


As for OP. I believe you have to enable cache on Karhu RAMTest to be able to detect cache/uncore error with it. The default value is using cache too but not as stressful. You can also use Stress FPU option for mild core stability testing.

Upgrade from 9900K
(9 items)
CPU
Core i9 10980xe 4.8 all core @ 1.305v
Motherboard
Asus Rampage VI Extreme Encore
GPU
Asus Dual OC GTX1050
RAM
G.Skill Trident Z 4000 18-18-18-38 T1 @ 1.35v
Hard Drive
Samsung 970 Pro 1Tb
Power Supply
Corsair AX1600i
Cooling
EK Velocity, 1 x HWLab GTS 280, 1 x EK PE 360, 1 x EK D5 reservoir combo, DIY satin hard tube and reservoir
Case
Corsair 500D butchered ATX case to fit EATX board
Monitor
Asus MX27UC 3840 x 2160 60Hz
▲ hide details ▲
BlueEarth is offline  
Reply

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