The Corsair User Forums  

Go Back   The Corsair User Forums > Corsair Product Discussion > Memory

Notices

Reply
 
Thread Tools Rate Thread Display Modes
  #1  
Old 06-17-2017, 11:31 AM
halesw halesw is offline
Registered User
halesw's PC Specs
 
Join Date: Jun 2017
Posts: 4
POST ID # = 908382
halesw Reputation: 10
Default DDR4 Compatibility ???

When I go to the website to check best mem modules for my board the CMD32GX4M4B3000C15
isn't listed but states its designed for X99 series boards. My computer tends to shut off randomly during gaming and I have stress tested on numerous occasions. When running mem test or AIDA 64 nothing fails.
Voltage Values:
CPU Core 1.237 V
CPU VRM 1.888 V
CPU Cache 1.034 V
+3.3 V 3.328 V
+5 V 5.040 V
+12 V 12.192 V
+3.3 V Standby 3.424 V
VBAT Battery 3.296 V
VCCIO 1.240 V
VCCSA 1.216 V
PCH Core 1.044 V
DIMM AB 1.341 V
DIMM CD 1.337 V
Battery 27.020 V
Battery Input 120.000 V
GPU1: GPU Core 0.831 V
GPU2: GPU Core 0.843 V
Reply With Quote


  #2  
Old 06-17-2017, 01:49 PM
c-attack c-attack is offline
Registered User
c-attack's PC Specs
 
Join Date: Jun 2014
Posts: 3,972
POST ID # = 908407
c-attack Reputation: 56
Default

Are you using the XMP timings? When you enable XMP, does it change the strap to 125? (you'll notice your multipliers for CPU and memory become drastically different)

Usually step 1 on Asus is to set the primary timings manually and then leave secondary and tertiary on Auto. The board will fill those values from a pre-programmed table from Asus and they are likely a little softer and certainly designed for your motherboard. Also, the VCCIO and VCCSA values are usually overblown on Asus defaults. I am OK with 1.24 on the VCCIO, but that is a lot of System Agent voltage (SA). I am only at 0.96v@3200MHz. You may want to manually change that offset to something between +0.10 to 0.15 offset in the AI Tweaker column.

You haven't specified your CPU clock speed. 1.23v seems too much for stock and perhaps you have targeted 4.0GHz. This should not be a limitation on the memory at this point, but it would be helpful to know for the discussion.
Reply With Quote


  #3  
Old 06-18-2017, 12:48 AM
halesw halesw is offline
Registered User
halesw's PC Specs
 
Join Date: Jun 2017
Posts: 4
POST ID # = 908495
halesw Reputation: 10
Default DDR4 compatability???

As of right now, my system is running at 3.8 GHz a minor 12% boost from stock. I used the one button click in the AMI BIOS. I have tried different OC settings but for this conversation, I typically run my rig at stock; 100 MHz strap 3.4 GHz. The crashes are intermittent. I have tried XMP in the past but it seemed as if the crashes were more frequent and yes under XMP the strap is set to 125 MHz. So I dropped back down to stock and things seemed stable until recently and nothing has changed other than typical software updates which I roll back just to make sure that these crashes aren't software related. I monitor the heat closely and while gaming the RAM is usually around 41-43 along with CPU temps hovering around the high 30's low 40's. GPU's pop right up to 80 and hold.

I am going into the BIOS now and will manually make some adjustments based on your post and report back. Thank you for the reply.
Reply With Quote


  #4  
Old 06-18-2017, 07:41 AM
c-attack c-attack is offline
Registered User
c-attack's PC Specs
 
Join Date: Jun 2014
Posts: 3,972
POST ID # = 908529
c-attack Reputation: 56
Default

So is it also crashing when at the default settings? 2133 MHz and JEDEC 15-15-15-36 timing?

One possibility is the 3000 MHz setting is a natural fit for the 125 strap. In the early days of X99, it was forced onto you when you selected that memory speed. 100 might work at 3000, but it might also cause these issues. If it is crashing at 2133, then we are likely looking for something else. Generally, BW-E is more flexible with memory than its predecessor, so it may be feasible. The downside to the 125 strap is you lose the ability to use adaptive voltage. The 2800 frequency is also a 125 strap. This makes 2666 and 3200 the more desirable frequencies. For quite a while I ran 2666 with tighter timings and it was equal to my 3200 speed in benchmarks. It is only in the last 6 months with recent BIOS upgrades the 3200 pulled away. It may be possible to stretch your settings to 3200 with looser timings to keep the 100 strap. If not, 2666 tightened down may be appealing as well.

There will be a few more things to manually change in the AI Tweaker/Extreme Tweaker column in the BIOS. Obviously you will need to manually select 3000 from the DRAM frequency drop down menu. Enter the 4 primary timings into the DRAM sub-menu. 125 strap back at the top. It may or may not tdo this automatically. Then return to the Extreme Tweaker menu and scroll further down to reach DRAM voltage (1.35v for AB & CD), System Agent Voltage (offset +0.10-0.15v), and VCCIO can stay on Auto. Your SA voltage is unique for each CPU specimen, so I can't give you an exact value. My current 5930K is really low. My older 5820K was more moderate, but still the 0.15 offset was plenty for 3200.

I don't think heat is an issue. It's pretty hard to get DDR4 hot without putting it in a poisoned environment. If I had to guess, the odds favor one of the key settings not being correct for the current level. This is really easy to do when switching back and forth between XMP on and off. Also, the XMP setting is just an overclock preset, like the ones that come with the motherboard (TPU) or through the BIOS. Which reminds me, even if you want to use the preset CPU overclock, I suggest not using the TPU physical switches on the motherboard. Those are meant to be flipped and then you don't touch anything else. Doing so after the fact usually results in crashed. That same TPU level can be selected directly through the BIOS or I can walk you through a manual. Either way, pre-defined presets are universal and not always optimal for your unique set-up.
Reply With Quote


  #5  
Old 06-18-2017, 02:13 PM
halesw halesw is offline
Registered User
halesw's PC Specs
 
Join Date: Jun 2017
Posts: 4
POST ID # = 908562
halesw Reputation: 10
Default

Originally when I OC it was using XMP and nothing else. I didn't realize at that time that my H100i cooler had failed and destroyed my chip; 2 chips and 1 board later (long story) I decided stock was sufficient for my needs. I have been running at stock now for more than 6 months with only a handful of crashes; probably once a week. To be honest, I never tried manually configuring the RAM only the GPU's until you mentioned it. I attached an AIDA/64 benchmark report from this afternoon; this includes the settings that you mentioned in the last post.

Stress test went fine, however, once I open my game then login to the actual character the computer crashes. With the previous setup, I could play for a few hours before the computer would crash.
Attached Images
File Type: png cachemem.png (94.2 KB, 27 views)

Last edited by halesw; 06-18-2017 at 02:40 PM.
Reply With Quote


  #6  
Old 06-18-2017, 02:45 PM
halesw halesw is offline
Registered User
halesw's PC Specs
 
Join Date: Jun 2017
Posts: 4
POST ID # = 908564
halesw Reputation: 10
Default

Here is the benchmark report using stock settings. This should directly answer your previous question about 2133 at 15-15-15-36 and the infrequent system crashes.
Attached Images
File Type: png cachemem norm.png (93.7 KB, 21 views)
Reply With Quote


  #7  
Old 06-18-2017, 04:39 PM
c-attack c-attack is offline
Registered User
c-attack's PC Specs
 
Join Date: Jun 2014
Posts: 3,972
POST ID # = 908579
c-attack Reputation: 56
Default

So it crashes even at 2133 standard timings. That is not super promising. It suggests you have some basic module issues or this is an entirely different undiagnosed problem. Sporadic crashes are often memory related, but can also be PSU or motherboard as well. It is probably easier to confirm or rule out the memory first.

Does it only crash when loading up a game or other large program? Do you get random freezes? Or is it a straight BSOD? MS makes it hard these days, but see if you can get a code next time. You can also set the BIOS in the Boot menu not to automatically restart on crash so you can copy the code.

You need to run a specific memory module test like HCI or Memtest86. HCI can be run from Windows. You open one instance for each physical and virtual core (12 little windows for 6800K). You will also need to set the memory size for each instance when you open it. You are trying to target around 90% of your total available system memory, so likely around 2048 MB per window. This does not need to be exact and if there is a module problem at 2133, errors will pop pretty fast. Keep in mind this is effectively a whole system stress test and your PC will behave as such. Go for at least 200%, which will likely take 1-2 hours.

Most people know Memtest and it has been around for ages. You need to launch it from a USB drive or ISO file on CD/DVD. When X99 launched, Memtest was not ready and I got used to using HCI. I can't really tell you the pros and cons, other than the way in which the test is run. Memtest is likely easier for most people to use, even if it does require a bootable USB to run. Again, if there is a serious issue the error will fly left and right. Run this with everything at the default settings.

EDIT: I might be interested to hear your CPU/board story at some point. We are both on X99 Strix and I see the "board killed my CPU" stories around often enough. I had considered them the usual we don't care if 10% of our products fail attitude, but do wonder if is more than that. You don't see the same stories on the Deluxe I/II or prior Pro model.

Last edited by c-attack; 06-20-2017 at 10:10 AM.
Reply With Quote


Reply

Thread Tools
Display Modes Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not 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

Forum Jump


All times are GMT -4. The time now is 08:52 PM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2017, vBulletin Solutions, Inc.