SOLVED New build hard boots during gameplay

[Distractor_Beam]

I'm going to go into detail, but I hope that doesn't deter you from reading.

The problem:

Computer Hard boots during gameplay. I can browse the internet and watch movies all day, but after about 10-15 minutes of playing either World of Warcraft or Modern Warfare 2, the computer just clicks off and then does it's normal boot procedure (of course with the standard windows has closed unexpectedly prompt).​

What I'm using:

AMD Phenom II X6 1100T Black Edition Thuban 3.3GHz, 3.7GHz Turbo 6 x 512KB L2 Cache 6MB L3 Cache Socket AM3 125W Six-Core Desktop Processor HDE00ZFBGRBOX with stock cooling.
http://www.overclockers.com/amd-phenom-ii-x6-1100t-review

ASUS Crosshair IV Extreme AM3 AMD 890FX SATA 6Gb/s USB 3.0 Extended ATX AMD Motherboard
http://www.asus.com/Motherboards/AMD_AM3/Crosshair_IV_Extreme/

CORSAIR Enthusiast Series TX850 V2 850W ATX12V v2.31/ EPS12V v2.92 80 PLUS BRONZE Certified Active PFC High Performance Power Supply
http://www.corsair.com/power-suppli...d-850-watt-high-performance-power-supply.html

EVGA SuperClocked 012-P3-1573-AR GeForce GTX 570 HD w/Display-Port (Fermi) 1280MB 320-bit GDDR5 PCI Express 2.0 x16 HDCP Ready SLI Support Video Card
http://www.evga.com/products/moreinfo.asp?pn=012-P3-1573-AR
Driver listed in device manager as: 8.17.12.7061 Date: 4/7/2011

CORSAIR Vengeance 8GB (2 x 4GB) 240-Pin DDR3 SDRAM DDR3 1600 (PC3 12800) Desktop Memory Model CMZ8GX3M2A1600C9B. 2 sets for a total of 16GB.
http://www.corsair.com/vengeance-8gb-dual-channel-ddr3-memory-kit-cmz8gx3m2a1600c9-279.html

3 mechanical hard drives
1 ssd for the operating system.
1 Blue ray drive

Windows 7 Ultimate 64bit​

What I've done:

• Eliminated imho the video card as a problem.
• Eliminated imho drivers as a problem.
• Eliminated imho temperatures as a problem.

What I'm going to do:

• Turn off Aero
• Run with 1 stick of Ram and see if it crashes. If it does, run with a different stick and see if it crashes.
• Underclock video card.
• As a very last result, (that I'd rather lie and say I did than actually do, but may do anyway) pull the Coolermaster 650W PSU from my girlfriends computer and see if this computer performs any better with it.

The long version.

At this point if I had to take a guess I'd say that it's the PSU.

I put the computer together 2 days ago and spent the rest of that day downloading and installing the software that I was going to be using. I started with the packaged video and chipset drivers. I downloaded the newest drivers from Asus as well as EVGA and installed them. During all of this I was watching AVI files and browsing the internet with no problem. After I got the new drivers installed I played a little Eve online (not a graphic intensive game) to see how it all looked. Everything went smoothly.
The next day World of Warcraft finished updating so I took a crack at it. It played very nicely with no hiccups, artifacts, frame rate drops or problems of any kind. After about 10 minutes of gameplay my computer shut off and started booting back up. I was at the time also watching avi files in my second monitor as well as browsing the internet during wait times (dungeon finder).

I considered it a fluke and attempted again with the same result.
At this point I just played on the internet to see if it would do it again and after an hour I had no problem. So I closed all applications and played JUST WoW by itself. Again it hard booted.

I deleted the chipset and graphics driver from the device manager and reinstalled them from the newest downloaded set. After a reboot I had the same result.
I used the Nvidia driver installer "clean install" option to do a D&R of the graphics driver with the same result.

Always hard crashes between 5-15 minutes.

I did some research online to see if other people were having troubles with my components and found 2 possible matches.

One says that this Asus motherboard gets a BSOD due to the Ethernet driver while watching youtube.
http://www.hardwarecanucks.com/foru...s-resolved-intel-82583v-gigabit-ethernet.html
I wasn't getting a BSOD, nor was I having trouble while browsing, but I applied the update anyway.

The other was a bit more frequent with the GTX570 and Windows 7, but was just about the GPU resetting and not an entire hard boot. The idea behind it was that the GTX570 was running unstable and that you should underclock it or increase the voltage to make it stable.
http://www.overclock.net/nvidia/905272-gtx-570-crash-crysis.html
http://www.tomshardware.com/forum/312452-33-windows-crash
http://www.hardwarecanucks.com/forum/troubleshooting/41216-gtx-570-crash.html

It wasn't until I read the last one that I agreed that a stock video card shouldn't need to be modified to run properly, so I called EVGA tech support.
​

The first person I talked to immediately diagnosed it as my power supply but for thoroughness had me boot to the bios where I checked my temperatures and voltages:
47C for CPU and 53C for Northbridge
I can't remember what the 3 volt and 12 volt were but they didn't throw up any red flags to him. We waited a few minutes and noted that the CPU temperature had only climbed to 48C
His final thought was to run an EVGA program called OC scanner with artifacts on for an hour and see how well my computer holds up.
http://www.evga.com/articles/00530/
It basically just ran the hell out of my GPU and would record any errors. The temperature on my GPU steadily climbed until it hit 86C, which of course scared the pants off of me, but the fan didn't have to climb past 57% and the temperature never rose another degree. It never recorded any artifacts, and held steady for 45 minutes. Then it froze and I was forced to hold the power to reboot it.

The 2nd tech support representative also diagnosed it as my PSU and had never heard of an 1100T. He asked if it was a dual core or a quad core. I explained it was a 6 core. He then asked if I had my 8 pin CPU power connected or 4. I was using 4 and here's why.
http://www.playtool.com/pages/psuconnectors/connectors.html#eps4plus4
My TX850 has a 4X4, but only 4 of the pins actually matched the mobo with squares and D's. The other 4 pins were all D's and I was nervous about plugging in plugs that didn't match.
The tech support representative explained that a 4 pin is good for a dual core, but you need an 8 pin for a quad core and especially for a 6 core. So I bit my lip and shoved the extra 4 pins in. I wasn't comfortable with the fit, but it eventually snapped into place. With the 8 pins now connected, I got the same result.

The third technical support representative asked if I had another power supply to test the video card out with. I do, but I really didn't want to pull the PSU out of my other computer, so I pulled the video card out of the other computer and put it in this one. After deleting the drivers and installing the new 9800GTX drivers, I got the same result. I didn't bother contacting EVGA again as I was now fairly certain that the video card was not the problem.

Today after work I decided that I had better test this computer out with another game, seeing as how the only time I'm having the problem is when I'm playing WoW. So I installed Modern Warfare 2, and got the same result.

I got some good advice from a friend on this forum about checking the voltage recommendation on the RAM. Most RAM uses 1.5V while some require 1.6V. The Vengeance RAM I have came with no documentation on the subject, but what I found online states that it is designed for 1.5V.

I have set the RAM in the bios to its' correct (1600MHz) setting. I didn't even bother to let the mobo auto detect it.
When I first put it together I booted it with each stick of RAM individually before I put all 4 in and all 4 booted the computer fine. I ran the windows memory diagnostic and it reported no errors.

I wrote this entire thread without the computer having any trouble.
​

To those of you that read this whole thing, I thank you from the bottom of my heart. To those that offer advice, I am your loyal servant.

 

[RollingThunder]

Beamer,

If that Corsair power supply is functioning properly with proper voltage readings under load using a DMM, it will run two of those cards.

You should have a 20+4 pin and a 8-pin 12v power connectors, you need them both connected. (Black and yellow wires)

http://www.newegg.com/Product/Product.aspx?Item=N82E16817139022&cm_sp=Cat_Power Supplies-_-YTVideo-_-YT17-139-022

Magnify the photo of the jumbled cables and just to the left of the 20+4 pin is an 8-pin 12V cable. You need that one connected to the motherboard too.

The board may come with a cover over part of the 8-pin connector exposing only 4 pins. Remove the cover to access all 8 pins.

 

[Distractor_Beam]

Thank you for your reply.

I'm not really sure how to test out a PSU while running with a multimeter. I have a basic understanding and ownership of an analog mm, but I don't have a digital one. I know that doesn't help.

I currently have the 20+4 pin connected as a 24 pin.

The 4+4 I should note doesn't look precisely like it does in the Newegg picture. The pins are the same, but it doesn't snap together as it appears to show on Newegg. They slide aside each other. It's deffinately the 8 pin CPU connector and I have all 8 connected. It's also worth noting that the 8 pin in the picture does match my 8 pin in that the bottom 4 pins that should have 2 square and 2 D's are actually 4 D's. That threw me off originally, but it is properly connected as of yesterday. You were correct that the bottom 4 pins on the mobo did have a cover.

 

[Distractor_Beam]

I had a small breakthrough due to the advice of my friend Conumdrum.

Prime95 killed it in less than 20 seconds.
Although I'm not certain exactly as to what that means.

 

[trents]

My guess is the CPU or the motherboard are overheating. You gave us great detail about the hardware except . . . there's no info about the case and its ventilation, i.e., number, size and placement of case fans. And there was also a comment in one of the posts about a jumbled mass of cables and wires. That doesn't sound good to me from a ventilation standpoint. And what is your room temp? I note you are in Las Vega. Get's pretty hot their without AC.

 

[Distractor_Beam]

My original test that killed it was conducted using "Blend (tests some of everything, lots of RAM used)"

I just did another test using "small FFT's (Maximum FPU stress, data fits in L2 cache, RAM not tested much)"
This ran for 20 minutes without any problems.

I'm going to run "In place Large FFT's (maximum heat, power consumption, some RAM tested)" and report back.

 

[Distractor_Beam]

Thank you for the reply.

My guess is the CPU or the motherboard are overheating. You gave us great detail about the hardware except . . . there's no info about the case and its ventilation, i.e., number, size and placement of case fans. And there was also a comment in one of the posts about a jumbled mass of cables and wires. That doesn't sound good to me from a ventilation standpoint. And what is your room temp? I note you are in Las Vega. Get's pretty hot their without AC.

I'm using a HAF X
http://www.coolermaster.com/product.php?product_id=6653
I also have the final extra 200mm fan installed.

thermostat set to 72 but it's probably a few degrees warmer in here.

 

[Distractor_Beam]

My last Prime95 test concluded with a hard boot in less than the amount of time it took for my girlfriend to complain about me not taking out the trash. About 6 seconds.

I have to head to work now, but when I get back I'll continue where I finished. Any info for when I get back would be greatly appreciated.

 

[trents]

Thoughts:

1. What are your core temps running when you are playing games? Please download, install and have open HWMonitor when you are running your game. This program will record max CPU temps and max core temps. Overheating can be caused by several things, including improper application of thermal paste or improper seating of the CPU heat sink.
2. Have you checked in bios to see what your CPU core voltage is set to? If "Auto" it may not be applying enough voltage to make the CPU stable. Set it manually to the appropriate amount, probably around 1.35v at stock settings. Same for ram. Set the ram voltage manually to factory recommended amount, usually about 1.5 volts.

 

[RollingThunder]

And there was also a comment in one of the posts about a jumbled mass of cables and wires.

Trents,

Those wires were in the NewEgg product slideshow advertising. Not meant to imply Beamer's looked like that.


Beamer, you can use that analog MM. Plug it into an unused molex yellow hot wite and black grounded for a 12v reading.

 

[bda1967]

Have you tried running memtest86+ on your RAM?

 

[Distractor_Beam]

I would first like to curb any thoughts about my jumble of wires.

And if you think that's bad you should see my other computer. I know I'm not the best at it, but I did make an honest attempt.

Beamer, you can use that analog MM. Plug it into an unused molex yellow hot wite and black grounded for a 12v reading.

Thank you for your reply.

This was actually great advice. I had to dust off the old mm, but after recalibrating it and hooking it up I was happy to see it still worked.

The first thing I did when I got home was to power on, run Prime95, and see what happens. It crashed in 3 seconds. The next thing I did was read these posts.

Thoughts:

1. What are your core temps running when you are playing games? Please download, install and have open HWMonitor when you are running your game. This program will record max CPU temps and max core temps. Overheating can be caused by several things, including improper application of thermal paste or improper seating of the CPU heat sink.
2. Have you checked in bios to see what your CPU core voltage is set to? If "Auto" it may not be applying enough voltage to make the CPU stable. Set it manually to the appropriate amount, probably around 1.35v at stock settings. Same for ram. Set the ram voltage manually to factory recommended amount, usually about 1.5 volts.

Thank you for your reply.

I pulled out the mm and got it set up. I looked online for recommended voltages on my 1100T and my RAM. Strangely enough I couldn't find anything that said this is the correct Voltage on my 1100T, nor did it say anything useful in the packaged manual. I simply saw volts that stated between 1.0 and 1.45 recommended. Then, of course, people that were pushing it to 1.59 and above. My RAM I did get a correct recommendation of 1.5V.

What I did was boot to my BIOS and noticed first of all that the method of setting voltage was completely different than what I'm used to. I usually get a list of defined volts and I choose one, but in this Bios it actually had a real time display of what it's running at (which was constantly changing, not by much, but changing nonetheless) and the ability to add or subtract from it. Everything here was set to auto except my DRAM frequency which I had set to 1600MHz to match what I have.

My CPU was reading between 1.310 - 1.312
My DRAM was reading between 1.660 - 1.667

Now I've never had a strong stomach for voltage correction as I don't know a heck of a lot about it, and I don't have a lot of money to replace my equipment if I screw it up, so what I did was against advice, but worth a try for the moment, and that was to set my DRAM frequency also to Auto just to see what happens.

With HWmonitor now installed and running I set my multimeter up in a location where I could easily see it. Then I ran the "In-Place Large FFT's" program on Prime95. I noted that the starting temperature was 36C on the CPU and 29C on the cores. With a hawks eye on the multimeter I was waiting for anything, ANYTHING! It should happen at any second. But it didn't. At this point I knew it was mocking me because it knew I was watching it.

This went on for about 20 minutes and the temperature on the CPU very slowly rose to 57C and the cores to 51C. This I know is particularly hot, but it is also a torture test so I expect them to be higher than normal operations. I decided at this point that I must have used the wrong Prime95 program so I used one that I knew had crashed it before and that was "Blend".

Again, 20 minutes went by and nothing. Prime95 would state it's running its' tests and report Self test 1024k passed over and over, but no crashing. The temperature did rise to an all time high on the CPU of 59C and 53C cores.

I gave up at this point and took off the multimeter (that never moved from 12V in the least) and put the case side panel back on, which strangely enough actually dropped the temperature on the CPU to 54C and the cores to 50C (using it as it's meant to be used?)

So what changed? The only thing I did different was set the DRAM frequency from 1600MHz to auto, and watch it to make sure it behaved.

Finally, with Prime95 still running, I took a leap of faith and wrote this entire post.... and it's still purring fine. I hope to the heavens it stays this way.

Any reason why a manual setting of 1600MHz would cause this?

Have you tried running memtest86+ on your RAM?

Thank you for your reply.

After I'm done with this post, this will be my next step.

 

[Conumdrum]

Hey buddy! I got a PSU tester I'll bring into work on Sunday.

Good post and you got lots of replies already, glad it's working out!

 

[Conumdrum]

On a side note for all you air guys, Beam has a Corsair A70 he'd like to use but has the ram height problem. I know it's not a megalehem etc, but's it's pretty decent. No reason not to use it.

I told him he can cut some of the huge ram fins to help with the clearance issue.

Your thoughts?

 

[RollingThunder]

On a side note for all you air guys, Beam has a Corsair A70 he'd like to use but has the ram height problem. I know it's not a megalehem etc, but's it's pretty decent. No reason not to use it.

I told him he can cut some of the huge ram fins to help with the clearance issue.

Your thoughts?

Never needed to do this but if he doesn't care about a warranty, why not?

 

[trents]

Any reason why a manual setting of 1600MHz would cause this?



Thank you for your reply.

After I'm done with this post, this will be my next step.

If you set the ram speed to Auto now it may be running at 1333 instead of 1600 and that would be why it has stablized. The current generation of AMD CPU integrated memory controllers are rated for 1333. They may run stable at 1600, epecially with some extra CPU-NB volts, memory volts and relaxed timings. If you set the ram speed manually to 1600 and made no adjustments to these other things then its not mystery why you were not stable.

 

[trents]

RAM marketed as 1600 mhz. is really 1333 ram that will run stably (on some systems) at 1600 mhz when overclocked (more volts, relaxed timings or both). I know its hard to believe but this comes up all the time on the forum. I will attach a pic of my own "1600 mhz" ram JEDEC profile in CPU-z "SPD" tab to illustrate the point.

Because of this and because of the opinion by many that AMD CPU's don't benefit much from memory bandwidth in excess of 1333, the real value of 1600 mhz-rated ram may be that you can run it at tight timings in 1333 mode.

By t he way, I understand your reluctance to "over-volt" but this is a forum for overclockers. That's what we do here. And I think it is true that if you over-volt carefully and under control there is very little danger of frying stuff.

 

[Conumdrum]

Hehe, Trents. My rig over 2 years old hasn't been overclocked. But it's stupid big on watercooling. Just haven't had to OC yet...............

I will someday. I did come here at the beginning to overclock though...............m

 

[Distractor_Beam]

Hey buddy! I got a PSU tester I'll bring into work on Sunday.

Good post and you got lots of replies already, glad it's working out!

Yeah I noticed the people on this forum are very helpful.

I'm pretty sure that this point that it's not the PSU, but thank you very much for your offer.

On a side note for all you air guys, Beam has a Corsair A70 he'd like to use but has the ram height problem. I know it's not a megalehem etc, but's it's pretty decent. No reason not to use it.

I told him he can cut some of the huge ram fins to help with the clearance issue.

Your thoughts?

Never needed to do this but if he doesn't care about a warranty, why not?

Thank you for the reply.

For the purposes of anyone with the same problem searching for this issue I'm going to post this topic as a new Thread here:

http://www.overclockers.com/forums/showthread.php?p=6865318#post6865318

If you set the ram speed to Auto now it may be running at 1333 instead of 1600 and that would be why it has stablized. The current generation of AMD CPU integrated memory controllers are rated for 1333. They may run stable at 1600, epecially with some extra CPU-NB volts, memory volts and relaxed timings. If you set the ram speed manually to 1600 and made no adjustments to these other things then its not mystery why you were not stable.

Thank you for your reply.

That makes sense to me. I ran Memtest86+ as suggested by bda1967. I started it yesterday and after an hour it was at 47%. At that point it was time for me to snooze. 7 hours later, strangely enough it was at 51%. When I got home from work today, 22 hours after I began memtest86+, it was back at 27%. I noticed at this point that it was actually looping the program. It reported 0 errors. I can note that my DRAM did report in memtest as DDR1333, making you correct.

RAM marketed as 1600 mhz. is really 1333 ram that will run stably (on some systems) at 1600 mhz when overclocked (more volts, relaxed timings or both). I know its hard to believe but this comes up all the time on the forum. I will attach a pic of my own "1600 mhz" ram JEDEC profile in CPU-z "SPD" tab to illustrate the point.

Because of this and because of the opinion by many that AMD CPU's don't benefit much from memory bandwidth in excess of 1333, the real value of 1600 mhz-rated ram may be that you can run it at tight timings in 1333 mode.

By t he way, I understand your reluctance to "over-volt" but this is a forum for overclockers. That's what we do here. And I think it is true that if you over-volt carefully and under control there is very little danger of frying stuff.

Thank you for your reply.

I think then that what I need to do is read up more on these forums on how people were able to use their 1600 RAM efficiently, and stable. Although I imagine that tweaking it much might adjust the cutting of the fins idea.

I'm very fortunate to have found this forum and also to have gotten the attention of people that actually know what they're doing, and for that I thank you all very much. I can see I have plenty to learn from you fellas.

 

[RJARRRPCGP]

My last Prime95 test concluded with a hard boot in less than the amount of time it took for my girlfriend to complain about me not taking out the trash. About 6 seconds.

I would give it a Vcore bump.