Confused about possible bad memory?

[essential]

I have frequent crashes on my new build (see sig). Seems to only crash when it's completely idle for a while. I thought it was several different things along the way. The last thing I investigated was bad RAM.

I have a asus maximus vii impact motherboard (with latest BIOS) with 16gb of corsair dominator platinum (2x8gb) (CMD16GX3M2A1600C9). The RAM is rated for DDR3 1600, Timing 9-9-9-24. It defaults to 1333 but I load the XMP profile and it runs at the rated timings, nothing is overclocked.

I made a memtest86 bootable USB and ran the full tests (see below). Obviously it looks like the RAM has completely failed (it showed over 2 millions errors), however the machine runs and is stable all the time except idle. If it was that bad I'm not sure the machine would even run?

I then booted back in to Windows and ran 5 copies of HCI Design Memtest simultaneously, each for 2047mb of RAM. It ran overnight ... no errors found, each ran over 1000% coverage. Shouldn't it have found something?

Did I run Memtest86 incorrectly for it to find so many errors? Something doesn't seem right.

Memtest86 Summary:

Report Date - 2014-11-02 00:19:09
Generated by - MemTest86 V5.1.0 Free (64-bit)
Result - FAIL

System Information
EFI Specifications - 2.31
CPU type - Intel Core i7-4790K @ 4.00GHz
Measured CPU Speed - 4000 MHz [Turbo: 4080.0 MHz]
# Logical Processors - 8
L1 Cache - 64K (204927 MB/s)
L2 Cache - 256K (62152 MB/s)
L3 Cache - 8192K (44406 MB/s)
RAM - 16431M (19147 MB/s)

Result summary
Test Start time - 2014-11-01 18:25:26
Elapsed Time - 5:52:26
Memory Range Tested - 0x0 - 41FE00000 (16894MB)
CPU Selection Mode - Single (CPU #0)
# Tests Passed - 4/44 (9%)
Lowest Error Address - 0x2E01F248 (736MB)
Highest Error Address - 0x400D3D70 (1024MB)
Bits in Error Mask - 0000080000000800
Bits in Error - 2
Max Contiguous Errors - 1
Test # Tests Passed - Errors
Test 0 [Address test, walking ones, 1 CPU] - 4/4 (100%) 0
Test 1 [Address test, own address, 1 CPU] - 0/4 (0%) 3984
Test 2 [Address test, own address] - 0/4 (0%) 3967
Test 3 [Moving inversions, ones & zeroes] - 0/4 (0%) 80435
Test 4 [Moving inversions, 8-bit pattern] - 0/4 (0%) 326047
Test 5 [Moving inversions, random pattern] - 0/4 (0%) 411504
Test 6 [Block move, 64-byte blocks] - 0/4 (0%) 231339
Test 7 [Moving inversions, 32-bit pattern] - 0/4 (0%) 1287670
Test 8 [Random number sequence] - 0/4 (0%) 176867
Test 9 [Modulo 20, ones & zeros] - 0/4 (0%) 41307
Test 10 [Bit fade test, 2 patterns, 1 CPU] - 0/4 (0%) 8599

Last 10 Errors
[Data Error] Test: 10, CPU: 0, Address: 3FB6B30C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3F14B30C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3EFA730C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3EC1F38C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3E4D338C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3E41F38C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3E14B30C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3DEF338C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3DE3F38C, Expected: FFFFFFFF, Actual: FFFFF7FF
[Data Error] Test: 10, CPU: 0, Address: 3DB2F38C, Expected: FFFFFFFF, Actual: FFFFF7FF

 

[VirtualLarry]

Did you use Memtest86, or Memtest86+ ? Memtest86 is horribly broken, in my experience. Memtest86+ is actively being developed.

 

[essential]

I ran Memtest86 ... the one by passmark. I'll try again with Memtest86+ then.

 

[Elixer]

You can also try MS's tool... http://windows.microsoft.com/en-us/windows7/diagnosing-memory-problems-on-your-computer
That works on 7 and higher.

And yeah, I agree with VirtualLarry, memtest86 hasn't been reliable in years.

What are the crashes you are getting (check your event log for info), it could also be PSU, or CPU (or...)

 

[BonzaiDuck]

Just as a note, there is something I discovered about XMP settings on some/certain motherboards. I had used manual settings on the sig rig since I built it. After getting more of an "education" from certain readings, I decided to use the XMP profiles.

On my board, changing to XMP and leaving all the timings set to "Auto" results in the RAM configuration with the proper timings and command-rate = 1. Yet the RAM's published spec gives those timings with command-rate = 2. And as we all know (or should know), running at the lower command-rate could mean either a need to bump up VDIMM voltage or VCCIO (or equivalent).

You could either manually adjust the Corsair RAM to fit the printed spec (after selecting XMP), or see what happens with voltage adjustments (I would recommend looking at VCCIO before anything) with more testing.

 

[essential]

I'm going to play more tonight. I disabled "Hybrid Sleep" in power settings and it made it through the night with no crashes. I've made it over 5 days with use, but like I said it seems to crash when idle for a while, almost like it's crashing while trying to go into a sleep mode. Hybrid sleep was the only thing still enabled.

All the critical error in the event log have to do with Kernel Power issues, and I hooked in to stable power, which is why I think it's an altered stay (such as the Hybrid Sleep) which when it tried to enter that mode it caused the problem.

At first all the crashes also had "DipAwayMode.exe" crashing which is something with the Asus Suite, but I disabled that from starting with the computer and the crashes still occurred. The machine had been completely idle since about 2pm eastern yesterday, so about 20 hours full idle, so disabling Hybrid Sleep might have done something.

I'm going to run the other memory tests tonight, and I'll look in to the BIOS memory settings beyond the XMP selection.


Log Name: System
Source: Microsoft-Windows-Kernel-Power
Date: 10/17/2014 11:48:50 AM
Event ID: 41
Task Category: (63)
Level: Critical
Keywords: (2)
User: SYSTEM
Computer: Desktop
Description:
The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.
Event Xml:
Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event"
System>
Provider Name="Microsoft-Windows-Kernel-Power" Guid="{331C3B3A-2005-44C2-AC5E-77220C37D6B4}" /
EventID>41</EventID
Version>3</Version
Level>1</Level
Task>63</Task
Opcode>0</Opcode
Keywords>0x8000000000000002</Keywords
TimeCreated SystemTime="2014-10-17T15:48:50.616450300Z" /
EventRecordID>1272</EventRecordID
Correlation /
Execution ProcessID="4" ThreadID="8" /
Channel>System</Channel
Computer>Joe-Desktop</Computer
Security UserID="S-1-5-18" /
/System
EventData
Data Name="BugcheckCode">0</Data
Data Name="BugcheckParameter1">0x0</Data
Data Name="BugcheckParameter2">0x0</Data
Data Name="BugcheckParameter3">0x0</Data
Data Name="BugcheckParameter4">0x0</Data
Data Name="SleepInProgress">6</Data
Data Name="PowerButtonTimestamp">0</Data
Data Name="BootAppStatus">3221225684</Data
/EventData
/Event

 

[BonzaiDuck]

I'm going to play more tonight. I disabled "Hybrid Sleep" in power settings and it made it through the night with no crashes. I've made it over 5 days with use, but like I said it seems to crash when idle for a while, almost like it's crashing while trying to go into a sleep mode. Hybrid sleep was the only thing still enabled.

All the critical error in the event log have to do with Kernel Power issues, and I hooked in to stable power, which is why I think it's an altered stay (such as the Hybrid Sleep) which when it tried to enter that mode it caused the problem.

At first all the crashes also had "DipAwayMode.exe" crashing which is something with the Asus Suite, but I disabled that from starting with the computer and the crashes still occurred. The machine had been completely idle since about 2pm eastern yesterday, so about 20 hours full idle, so disabling Hybrid Sleep might have done something.

I'm going to run the other memory tests tonight, and I'll look in to the BIOS memory settings beyond the XMP selection.


Log Name: System
Source: Microsoft-Windows-Kernel-Power
Date: 10/17/2014 11:48:50 AM
Event ID: 41
Task Category: (63)
Level: Critical
Keywords: (2)
User: SYSTEM
Computer: Desktop
Description:
The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.
Event Xml:
Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event"
System>
Provider Name="Microsoft-Windows-Kernel-Power" Guid="{331C3B3A-2005-44C2-AC5E-77220C37D6B4}" /
EventID>41</EventID
Version>3</Version
Level>1</Level
Task>63</Task
Opcode>0</Opcode
Keywords>0x8000000000000002</Keywords
TimeCreated SystemTime="2014-10-17T15:48:50.616450300Z" /
EventRecordID>1272</EventRecordID
Correlation /
Execution ProcessID="4" ThreadID="8" /
Channel>System</Channel
Computer>Joe-Desktop</Computer
Security UserID="S-1-5-18" /
/System
EventData
Data Name="BugcheckCode">0</Data
Data Name="BugcheckParameter1">0x0</Data
Data Name="BugcheckParameter2">0x0</Data
Data Name="BugcheckParameter3">0x0</Data
Data Name="BugcheckParameter4">0x0</Data
Data Name="SleepInProgress">6</Data
Data Name="PowerButtonTimestamp">0</Data
Data Name="BootAppStatus">3221225684</Data
/EventData
/Event

Just coming up with a fresh thought here.

. . . Uh-Hu-uh. . . An Asus board -- the Maximus line for Z97, I think . .

Your Maximus and my sig-board are two generations apart, but BIOS menus and settings would more likely have the Maximus including the P8Z68 BIOS' as a subset of features. Hopefully the cross-identification will be easy even if the names have changed a bit.

There should be a Digi+VRM submenu, or as follows as first two priority items:

VRM Phase Control -- with settings "Standard," "Optimized," "Extreme," and "Manual." [Per my own mobo's BIOS] Manual is described as "phase number" by "Current(A) step -- assumed to be amperage.

Digi+VRM Duty Control -- offering "T-Probe" Thermal Balance or "Extreme" Current Balance

If you don't have these both set to "Extreme," try it, and watch for any more BSODs. I think you described these as being intermittent? How frequently do the BSODs, resets or freezes occur? What are the error codes from the BSODs if they remain visible long enough for you to read them? I believe it is a hexadecimal expression in this pattern: 0x00000009C, 0x0 . . . . 124, etc.

For yourself, you may want to open up a spreadsheet and keep dated crash data in a chronological list.

See what happens with that. There are other causes of these idle, intermittent crashes, and enough had been written about them back in Sandy-Bridge days that I'm thinking that you'd want to try the two Digi+VRM settings. However, as I say again below, these occurred under otherwise stable overclocks which could run all day with the maximum temperatures and clock speed under LinX or IBT.

Still need to find out if CMD=2 eliminates or increases the "inter-crash" interval.

I should've asked this first: You didn't indicate any overclocking. Have you done any overclocking?

My old processor had "issues" with Sleep states with overclocking. They were resolved with the third BIOS revision of my Z68 board. Maybe you want to turn all sleep states off to see if that eliminates the problem.

And -- sorry to say -- you'll need to have patience for doing these things one at a time.

Also, someone else might know more. It could always be defective RAM, but it could be some other hardware problem -- even a matter of BIOS settings. In any case, RMA'ing RAM is a fairly easy thing to do, even if the owner pays shipping.

 

[essential]

I could not find VRM Phase Control but I did find VRM Duty Control, it&#8217;s set to T-Probe.

I&#8217;ve never seen the blue screen &#8230; it just restarts and when I see it again it&#8217;s just on the Windows desktop so I know it restarted cause nothing else is open except what runs with the computer. It&#8217;s very stable when I&#8217;m using it, it&#8217;s only ever crashed when it&#8217;s been idle.

When I load the XMP profile, basically everything else is set to Auto, nothing else even set to any numbers. Command rate is 2 ... not sure what it's supposed to be, maybe 1 like yours.

I have nothing overclocked, even the XMP profile is just running the rated speeds.

I think the hybrid sleep was causing the issue but I&#8217;d like to test the RAM with memtest86+ &#8230; it just won&#8217;t boot to it, tried a USB and DVD. Memtest booted fine from USB, with this the USB isn&#8217;t recognized in the boot menu and when I select the DVD it just boots in to BIOS. I&#8217;ve changed RAID to ACHI, and UEFI to legacy OS, and same issue. Nothing will boot to Memtest86+ that I can figure out. I assume it&#8217;s my XP941 messing things up but who knows.

 

[BonzaiDuck]

I could not find VRM Phase Control but I did find VRM Duty Control, it’s set to T-Probe.

I’ve never seen the blue screen … it just restarts and when I see it again it’s just on the Windows desktop so I know it restarted cause nothing else is open except what runs with the computer. It’s very stable when I’m using it, it’s only ever crashed when it’s been idle.

When I load the XMP profile, basically everything else is set to Auto, nothing else even set to any numbers. Command rate is 2 ... not sure what it's supposed to be, maybe 1 like yours.

I have nothing overclocked, even the XMP profile is just running the rated speeds.

I think the hybrid sleep was causing the issue but I’d like to test the RAM with memtest86+ … it just won’t boot to it, tried a USB and DVD. Memtest booted fine from USB, with this the USB isn’t recognized in the boot menu and when I select the DVD it just boots in to BIOS. I’ve changed RAID to ACHI, and UEFI to legacy OS, and same issue. Nothing will boot to Memtest86+ that I can figure out. I assume it’s my XP941 messing things up but who knows.

Why not HCI Memtest-64? I'm still using the version I got when I built this rig. I put it on a CD and boot to that. You should be able to put the USB thumb drive in top priority in the BIOS boot menu, but you're saying it isn't recognized?

I went through troubled times early this year to catch my own "intermittent reset" [sometimes BSOD] problem. I'd already planned hardware upgrades, so I started making them earlier than planned.

It would be nice to know the source of your problem so you could either fix it yourself or RMA the item if you can't fix it.

 

[Elixer]

I think the hybrid sleep was causing the issue but I’d like to test the RAM with memtest86+ … it just won’t boot to it, tried a USB and DVD. Memtest booted fine from USB, with this the USB isn’t recognized in the boot menu and when I select the DVD it just boots in to BIOS. I’ve changed RAID to ACHI, and UEFI to legacy OS, and same issue. Nothing will boot to Memtest86+ that I can figure out. I assume it’s my XP941 messing things up but who knows.

What program did you use to write the iso to the USB flash drive ?
Use this, http://rufus.akeo.ie/ then, select your boot menu options to boot from USB.

 

[essential]

Why not HCI Memtest-64? I'm still using the version I got when I built this rig. I put it on a CD and boot to that. You should be able to put the USB thumb drive in top priority in the BIOS boot menu, but you're saying it isn't recognized?

I went through troubled times early this year to catch my own "intermittent reset" [sometimes BSOD] problem. I'd already planned hardware upgrades, so I started making them earlier than planned.

It would be nice to know the source of your problem so you could either fix it yourself or RMA the item if you can't fix it.

I used HCI Memtest in Windows, I ran 5 simultaneously overnight for 2047 each, no error found, each had over 1000% coverage. I didn't know you could boot in to HCI, didn't see that on their website.

The Memtest86 was able to be seen on the thumbdrive as a boot device, the Memtest86+ is not, the system sees the Memtest86+ boot CD in my ROM drive, but also can't boot in to it for some reason, when I select it it just goes in to BIOS. When I try to force boot to the ROM it just stays in BIOS.

What program did you use to write the iso to the USB flash drive ?
Use this, http://rufus.akeo.ie/ then, select your boot menu options to boot from USB.

Memtest86+ has a USB installer on their site, you run it and pick what drive you want to use and it does the rest, this is the process I used for Memtest86 as well, they provide the installer. I will try Rufus though, didn't think of that last night.

 

[BonzaiDuck]

I used HCI Memtest in Windows, I ran 5 simultaneously overnight for 2047 each, no error found, each had over 1000% coverage. I didn't know you could boot in to HCI, didn't see that on their website.
. . .

I just thought of something. See -- my idle-level resets/BSODs would occur once a week to ten days, sometimes skipping a month. I reset the BIOS overclock parameters to stock, but it would still happen. I swapped hardware as part of a budgeted upgrade, but it was still occurring.

I also discovered that my USB3 front-panel hub -- connected to the motherboard -- wasn't working correctly - "Not enough resources" would show in device manager.

I went into BIOS and took inventory of everything I didn't need or wasn't using which was "Enabled," and disabled eSATA, the extra onboard Marvel SATA controller, and the Asmedia USB3 controller. The problem simply "went away."

The mobo manual (for my P8Z68 ASUS) shows how you can use one of these "add-on" feature items, and then others won't work. There is a limited number of PCI-E lanes, shared IRQ's, etc.

eSATA on these boards -- "Silicon-Image"??-- separate controller. But what is "eSATA?" It is SATA with special external ports for the appropriate cable. You can get eSATA off the main onboard SATA controller by investing in $15 in parts and cables. For USB3, you can get a Hoo-Too PCI-E card, which still has the 20-pin port for internal connection along with two other external ports at the PCI-E slot.

Long and short of it -- as I said -- find out what you don't need or aren't using, and disable those things in BIOS.

UPDATE: Oh, gee . . . . you have an ITX Maximus VII. Not likely a lot of extra controllers or features . . . .

 

[essential]

Was never able to get memtest86+ working, but when removing one of the RAM sticks and using Windows Memory Test, it passed fine, taking that stick out, replacing with other in same slot, same test, immediately found errors and crashed before finishing, this is what happened with both sticks in as well.

So, finally confirmed a bad RAM module. RMAed with Corsair as I was just outside my Newegg window, should have a replacement later this week.