Computer shuts down under load.

[ccjcc81]

Hey guys, my computer keeps crashing to black screen. It happens when I am playing iRacing and FSX. It does not happen at idle or low load. There are no warnings or errors, just crashes to black screen, no video output. Power is still on, but I cannot turn the computer off by holding power button, and reset button does nothing. I have to throw the power supply switch or unplug it to turn it on. When it crashes, all the LEDs in the case stay on, and all the fans inside stay on. There are no beeps. The only indication of a crash other than the black screen is the joystick LEDS turn off. The joystick is plugged into the front USB slot.

Computer specs:
Asus M4A89TD PRO/USB3
AMD 1100T Black Edition X6 3.3GHz
Radeon HD6870
12GB Corsair DDR3 1600
Rosewill 700W power supply
Western Digital CB 1TB 7200rpm
Windows 7 X64; Fresh install, crashes occurred before and after install

Fresh driver installs all around, and Windows fully updated. I even tried reverting to an older GPU driver. Nothing overclocked. Case is clean and dust free. All power cables are secure, and I have reseated the GPU. All fans are working. According to Core Temps, CPU temp stays below 30C at load, and according to CCC, GPU temp stays below 50C.

Interesting symptom; when I run FSX, right before the crash, when cycling through views, I lose textures. Because of that, I thought it might be a memory error, so I ran windows memory diagnostic. I ran it successfully 2 times, but the third time I got black screen during test. I have 3 sticks of mem, so I pulled one out and ran FSX. Still crashed. Pulled another out and put first one back, crash. Did that with third, crash. Then I moved two remaining sticks from memory slot 1,2 and put them in memory slots 3,4, still crashing. I don’t think it’s the memory, and if it’s the mobo I don’t think it’s a bad slot. I ran a disc check and no bad sectors were found on HDD.

Based on this, what would you guys do?

Thanks guys.

 

[unksol]

Use gpuz and enable the log so you can see what your gpu temp was at the crash point. You can also stress it and monitor temps with furmark

I'm just gonna say core temps is wrong. 30C is a very good idle temp. Unless its in a freezer. Get AMD s temp monitor or use the motherboard sensor and run prime 95 for a few hours and watch the temps. Want to stay below 60C.

Run memtest 86+ too check for memory problems.

Those 3 should pretty effectively test your components "individually" and subject them to high stress. Another possibility is a bad PSU that gets overloaded with everything under load

 

[smorizio]

start with checking that your mb bios is up to date. sometime there some ram and or bug fixes. if someone has another power supply i would try that next. it may be that the power supply not holding up under load. I would also use cpu-z and gpu-z make sure the ram and video card being read right. with the ram make sure the timing set right. if it not the power supply it could be power issue on the mb..under load it heats up and then crashes.

 

[ccjcc81]

Use gpuz and enable the log so you can see what your gpu temp was at the crash point. You can also stress it and monitor temps with furmark

I'm just gonna say core temps is wrong. 30C is a very good idle temp. Unless its in a freezer. Get AMD s temp monitor or use the motherboard sensor and run prime 95 for a few hours and watch the temps. Want to stay below 60C.

Run memtest 86+ too check for memory problems.

Those 3 should pretty effectively test your components "individually" and subject them to high stress. Another possibility is a bad PSU that gets overloaded with everything under load

I'm confident in the temps. When I built the unit I lived in an apartment with really bad air conditioning. I built the rig with a hyper 212 and 3 scythe utra-kaze fans that push 250CFM. I have since moved to a place with proper air conditioning, so even cooler. I'll check the temps later anyway. I'll download Memetest later too.

Thanks for feedback.

Edit to add: the Scythe Ultra Kaze fans push 133CFM, not 250.

 

[ccjcc81]

start with checking that your mb bios is up to date. sometime there some ram and or bug fixes. if someone has another power supply i would try that next. it may be that the power supply not holding up under load. I would also use cpu-z and gpu-z make sure the ram and video card being read right. with the ram make sure the timing set right. if it not the power supply it could be power issue on the mb..under load it heats up and then crashes.

Bios is up to date. I'll try your suggestions and post back. Thanks.

 

[jaquith]

Start by looking at the Event Manager, errors like 41/63 can be all sorts of things. IF that shutdown only happens @ load then isolate the CPU from the GPU(s). Use only Stock settings i.e. no OC'ing of the CPU and/or GPU(s); in the BIOS Load Optimized Defaults; on ASUS it's F5 then F10 to Save & Exit = Yes.

CPU @ 100% Load - AIDA64 Extreme (stress uncheck GPU) or Prime95 (use small FFT's)
GPU @ 100% Load - 3DMark11 with CPU @ Stock settings

IF the problem happens in either case then chances are a bad PSU, but to be certain F8 and run Memtest86+ for 2-4 full passes is the best way to rule-out a memory issue.

Event Manager:

 

[unksol]

I'm confident in the temps. When I built the unit I lived in an apartment with really bad air conditioning. I built the rig with a hyper 212 and 3 scythe utra-kaze fans that push 250CFM. I have since moved to a place with proper air conditioning, so even cooler. I'll check the temps later anyway. I'll download Memetest later too.

Thanks for feedback.

Yea, def check the temps. Especially since you moved. With a heavy heatsink it can rock loose moving it. AMD is way Better than intels weak pins, so you're probably ok, but still. Others have also had great suggestions so maybe not the highest priority but still possible

 

[ccjcc81]

Ok, running memtest86 now to rule out mem issues. Downloaded and installed cpu-z and gpu-z, and set both to log, but I may need some help reading the log files.

I'll try to figure event viewer out.

Thanks for all the great advice guys.

 

[ccjcc81]

Update: The rig didn’t survive its first memtest. When I walked away it was at about 25% overall, and no errors showing. 10 minutes later I come back to a black screen. I restarted it.

INTERESTING ANOMOLY: The first time it ran the memtest, the CAS timings were 7-7-7-20 at 533MHz. That’s after I loaded setup defaults in the bios. The second time I ran it, after the crash, the timings were 8-8-8-20 at 533MHz. I aborted the test and restarted it to see if it starts the test at 7-7-7-20 and I just didn’t notice it. Nope, it started again at 8-8-8-20 @ 533MHz. I aborted it, loaded bios defaults again, and ran it again, and it’s running at 533MHz 8-8-8-20 now. Is there a problem that would cause random memory timings? How could it change?

When I bought my memory, it said it was DDR1600, but it’s running at DDR 1066. Why? And on the website it says the SPD Speed is 1333, but Tested Speed is 1600. What does SPD stand for?

 

[smorizio]

http://en.wikipedia.org/wiki/Serial_presence_detect
with asus mb the new gui bios there three ways ram speeds will change.
you clicked on the overclock icon. in the ram speed you set the default speed to x speed.
you set the ram speed to xmp profile under the ai part of the bios gui.
with cpu-z the last line on the right is the xmp speed and voltage. mb wont use that line undless in the dram sped setting you turn it on.

 

[ccjcc81]

I didn't make any ram speed changes between setting bios defaults, and I've set bios defaults before every test now. Last night it crashed again during a memtest. When I unplugged it and booted it back up, it said overclocking failed, press F2 to load setup defaults, even though no attempt at overclocking was made. It says that from time to time when the system crashes, but not every time. I loaded setup defaults, and started memtest again, and oddly enough, it is running the ram at 533MHz at 7-7-7-20. Something has changed the memory speed again.

Edit to add: Last night I ran memtest 3 times. First time crash to black screen, second time passed 1 test with no errors, started running another test, crash. Running again right now.

 

[ccjcc81]

It’s getting worse now. I was rummaging around in the bios looking for memory timings and it crashed 4 times. Each time it would crash I would restart into bios and load system defaults first thing. I gave up and let it boot to windows and it lasted less than a minute. I’m going to wait a while and try again later. I’m also going to go get some thermal paste and reseat my heat sink.

Do these problems seem like memory problems? What would you guys do now?

Thanks again guys.

 

[jaquith]

First off you don't want to run @ DDR3-1600 on a non-FX AMD without an OC of the FSB -> 240MHz or (Black Edition only) CPU-NB Frequency -> 2400MHz AND an increase in the (CPU IMC) CPU-NB Voltage -> 1.20v. The Default SPD depends on the RAM kit and it's encoded SPD/JEDEC Frequency & Timings. XMP is an Intel only setting.

Typically IF the SPD is being misread by the BIOS you'll get a 'OC Failure ... Press F1' error and if you do then the solution is to manually set the RAM's Frequency, CAS Timings, Voltages and if needed the OC.

The probable reason for the various CAS Timings and perhaps Frequency is because you're Mix-Matching two kits or have a Tri-Channel kit installed. There's no way to equal 12GB in a fully matched Dual Channel kit.

Q - What exact 'Corsair' kits or sticks do you have installed? ; Please provide links.
Q - What DIMM Slots are the sticks installed?
Example:
CPU: | 2GB | 4GB | 2GB | 4GB |
CPU: | 4GB | 4GB | 4GB | ----- |

 

[ccjcc81]

Corsair XMS3 4GB DDR3
http://www.corsair.com/en/cmx4gx3m1a1600c7.html

I used to have it set up this way:
CPU: | 4GB | 4GB | 4GB | ----- |

But after all the troubleshooting I left it like this:
CPU: |-----| 4GB | 4GB | 4GB |
I have to remove a fan from the heat sink to get to the 1st slot. Crashes both ways.

As noted above, I have tried removing a stick to have a dual channel setup, and have tried 8GB in these configs:
CPU: | 4GB | 4GB | ----- | ----- |
CPU: | ----- | ----- | 4GB | 4GB |

 

[jaquith]

Well as I assumed and a tad worst case that I also assumed you do NOT have a kit and instead individual sticks and of that Corsair's most problematic series.

For Dual Channel:
CPU: | 4GB | ----- | 4GB | ----- | ; Blue DIMM slots

Non Dual Channel:
CPU: | 4GB | 4GB | 4GB | ----- |

CPU-z for CMX4GX3M1A1600C7
See -> https://lh5.googleusercontent.com/-fwyLYNcy1nc/T8ezUdi3StI/AAAAAAAAAYo/bqHfUwXCV3I/s407/SPD.JPG

I'm going to post (2) versions of proper BIOS settings, and if using (3) stick IMO use DDR3-1333 Frequency but I'll post up to DDR3-1600. Keep in mind the RATING is for (1) ONE stick and not as a pseudo kit.

1. BIOS DDR-1333 (my recommendation):
F5 ; Load Optimized Defaults
DRAM Frequency -> DDR3-1333
DRAM Timing Configuration /Enter
DRAM CAS -> 9
DRAM RAS to CAS -> 9
DRAM RAS PRE -> 9
DRAM RAS ACT -> 24~26 ; 26 is listed in the SPD but 24 is more normalized JEDEC
-
DRAM Command Rate -> 1T~2T ; 1T should be fine but as a mix and IF errors use 2T
-
DRAM Voltage -> 1.55v~1.60v ; use x3 sticks I'd up the voltage by +0.05v, the RAM is Rated to 1.65v
F10 ; Save & Exit = Yes

2. BIOS DDR-1600 (Best as a Kit and IMO in Dual Channel):
F5 ; Load Optimized Defaults
AI Overclock Tuner -> DOCP
DRAM OC Profile -> DDR3-1600
DRAM Timing Configuration /Enter
DRAM CAS -> 7 ; failure use 8
DRAM RAS to CAS -> 8
DRAM RAS PRE -> 7 ; failure use 8
DRAM RAS ACT -> 20~24 ; 20 is listed in the SPD (XMP) but 24 is more normalized JEDEC
-
DRAM Command Rate -> 2T
-
DRAM Voltage -> 1.65v~1.70v ; use x3 sticks I'd up the voltage by +0.05v, the RAM is Rated to 1.65v
CPU/NB Voltage -> 1.20v or +0.10v
F10 ; Save & Exit = Yes

 

[AntiZig]

Edit: nevermind me

 

[ccjcc81]

Well as I assumed and a tad worst case that I also assumed you do NOT have a kit and instead individual sticks and of that Corsair's most problematic series.

Lol. What I don't know about memory would fill a warehouse. So it's better to buy it in kits then? I'm going to set the memory to your recommened specs right now. But, this problem only just started occurring. I've had this rig for nearly a year now. Could timings and speed really be causing the problem I'm having?

ALSO: Just ran Prime95 small FFTs. Crashed after a short while. I glanced at the cpu-z report and found that under monitoring, it showed the cpu temp at 32C and the mobo temp at 30C. I tried to post a copy of the report, but the computer crashed again. I'll see if it will stay on long enough now.

 

[ccjcc81]

Well, I changed the settings to your recommended specs, ran prime95 and crash. Whats next?

As soon as the baby wakes up I'm going to radio shack to get some paste to reseat the cpu heat sink so we can rule that out.

Thanks again.

 

[jaquith]

Which Settings? DDR3-1333 (recommended) or the OC DDR-1600??

One stick:
CPU: | 4GB | ----- | ----- | ----- |

BIOS DDR-1333 (my recommendation):
F5 ; Load Optimized Defaults
DRAM Frequency -> DDR3-1333
DRAM Timing Configuration /Enter
DRAM CAS -> 9
DRAM RAS to CAS -> 9
DRAM RAS PRE -> 9
DRAM RAS ACT -> 26
-
DRAM Command Rate -> 2T
-
DRAM Voltage -> 1.50v
F10 ; Save & Exit = Yes

Failure try a different stick

Failure the problem isn't the RAM.

 

[ccjcc81]

Which Settings? DDR3-1333 (recommended) or the OC DDR-1600??

One stick:
CPU: | 4GB | ----- | ----- | ----- |

BIOS DDR-1333 (my recommendation):
F5 ; Load Optimized Defaults
DRAM Frequency -> DDR3-1333
DRAM Timing Configuration /Enter
DRAM CAS -> 9
DRAM RAS to CAS -> 9
DRAM RAS PRE -> 9
DRAM RAS ACT -> 26
-
DRAM Command Rate -> 2T
-
DRAM Voltage -> 1.50v
F10 ; Save & Exit = Yes

Failure try a different stick

Failure the problem isn't the RAM.

I used your DDR3 1333 settings. just realized I forgot to manually change the dram freq to 1333...oops. trying again, next I'll try one stick at a time per recommendation.

 

[ccjcc81]

Ok guys, I’ve tried the memory every which way, and I don’t think it’s a bad stick. Here’s a recap of what I’ve done:

With memory: I’ve tested each stick individually with the memory settings recommended above. I used Prime95 to test, and I always crash to black screen. The length of time between crashes vary. Before I did that I also tested some other memory configurations, but the one stick at a time method was the most thorough. Are there any more memory tests, or have I ruled out memory now?

With heat: Throughout all the testing I have been monitoring temperatures with HWMonitor, Core Temps, and GPU Z. Temperatures have all been stable. My cpu temps have never gone above 40C, and this is evident in the CPU-Z logs. I have reseated my mammoth heat sink just to be sure. Temps were the same after reseating as before. GPU temps have stayed under 50C during all of my testing. I have not monitored the GPU temps under GPU stress tests, but if my system is crashing while GPU temps are at 50C, I don’t think it would tell us much if I did.

At one point during memory testing, my computer would crash whilst navigating menus in the BIOS. I hardly consider that “under load.”

What should I do now guys? Is there any way to determine if it is the mobo or power supply short of buying new ones? (no $$$)

Here’s another important fact that I probably should have mentioned earlier: I have 6 fans running in my system, not including the GPU fan. 3 of those fans are Scythe Ultra Kaze fans. These 3000rpm fans each push 133cfm of air. They are heavy duty. After my last computer died in what I believe was a heat related failure, I decided that there was no such thing as too much cooling.

So, could I have over-taxed my cheap power supply into an early demise?

Based on my (very) limited knowledge, I’m thinking this is either a PSU or mobo issue, but has anyone got any more ideas?

Thanks again guys.

 

[ccjcc81]

SOMETHING NEW HAPPENED. The last time it crashed to black screen, I just left it on. I didn't even bother getting up to unplug it. After 10 minutes, it restarted. It stayed on the mobo splash screen for a very, very long time. Then it gave me a no keyboard error, and gave me the windows boot menu options: Safe mode, normally...

Then it crashed before windows and started the process again...

 

[ccjcc81]

Great news!

I went and bought a new power supply. It seems to have fixed the problem. I've only been running Prime95 now for about 30 minutes, but that's about 25 minutes longer than I could get it to run before. Also, I've noticed in CPUID Hardware Monitor that my monster fans are running at 2900rpms now. I could swear that I saw that they were running at only 2200rpms earlier. It's too early to tell for sure, but things are looking good now. I'll get back with you guys tomorrow and let you know how it turns out.

 

[jaquith]

Yep I kind of assumes a PSU, but if all of the sticks were 'bad' @ DDR3-1333 then I started to be concerned you might have a bad MOBO.

Good Luck - please update.

 

[ccjcc81]

Well, I'm a happy camper. Computer is running better than it was before I started having problems. I didn't realize that my memory was running at 1066 before. Now it's at 1333, and I can actually see the difference. Windows boots faster, and FSX is running smoother. New PSU solved the problem, and, I went with a modular unit this time, and that alone would have been worth the upgrade even if the old one didn't go out. Much tidier.

I can't even begin to express my gratitude. Thank you all very much for generously sharing your knowledge to help me. I hope you all realize how much you are giving when you log on here and help newbs fix their computer problems. This problem was really stressing me out, and I had no clue what to do. And, I learned quite a bit. Again, thank you all.