New custom built computer : can't play any game, faulty hardware?

[Jhkh]

Hi everyone! I'm here again asking and not helping (newbie inside!) as I ran into an issue I'm not able to solve. I don't like making requests without a good reason, so I ran many tests before coming here.


I just finished assembling a new setup for a friend, and it's the first "gaming" modern computer I ever put together :


2x 4Go Kingston PC15000 cas10 HyperX Fury black
AMD FX 6300 Black Edition
MSi 970A-G46
Cooler Master hyper 212 EVO
MSI GTX 660 OC 2Go
Corsair CX Bronze 500W
Western Digital Blue 7200 1To
All of this inside a Xigmatek Recon USB 3.0 Edition. (Cheap case)


A memory stick was defectuous and caused errors during windows install, so the system now runs on one 4Go stick (tested ok with 2 pass on memtestx86).


It runs nice and quiet with windows 8.1. However, almost every game my friend tried crashed after some time with an error or just quit (League of Legends, Battlefield3, counter strike GO). So I tried installing windows 7 but I ran in the same trouble. Only Call of Duty Modern Warfare (first one!) worked full settings as it's a very old game and won't use much resources.
The drivers are up to date, as is the bios of the motherboard.


I tested the graphic card with furmark (65°C dixit speedfan on 720p test).
I also tested some video encoding with MEgui and the system freezes (no BSOD) after 5 to 7 minutes at full CPU load even if it seems to be quite cool, but I don't trust the readings I get with speedfan and HWmonitor (see below the pic of the test, readings in HWM are weird. For example +5V and +12V are completely out of range in the programs, but ok with a multimeter connected to PSU (5,1V and 12,2V)).
http://i.imgur.com/K6uJDaO.png


I checked windows 7 event manager and found an error when games are crashing : nvstreamsvc.exe not found.


I'm now waiting for the replacement set of ram from Kingston, and thinking about setting manually the timings, voltage and speed (actually set to auto in msi bios), but I still want your view over the facts as I'm getting confused on which part(s) I should suspect in the end. I hate such erratic and undefined behaving, so I'm all into your hands for the solution


P.S : I will run a test with OCCT and linX as soon as I've received the new ram but it may take two weeks as my local store ran out of stock.


One more thing : I won't drop this thread and will tell you how things will go, good or bad

 

[Smoove910]

with the error 'nvstreamsvc.exe', I'll assume you have an NVidia graphics card?

Regardless, what I briefly read says the nvstreamsvc.exe service typically spawns several more instances, which inherently eats up all of your RAM. This would be indicative of your games running like shit... or at least that's what it's sounding like to me.

You can download and install sysinternals tool and look at the Rammap to see if it has spawned.
http://technet.microsoft.com/en-US/sysinternals/bb842062

Solution: Disable the nvstreamsvc.exe service and see if it resolves the issue.

 

[Jhkh]

Thanks Smoove910, I will have a look at that, this may be easier than I thought.
I'm sorry, I made the assumption that I could keep the working ram module, but as they were bundled together I had to return both today so I can't test anymore for now: . I will come back to you when I have the new ones, I hope you won't mind if it takes some time.
My bad :ninja:

 

[Ketchup]

In Windows Services you should see an entry for nvstreamsvc.exe (Nvidia Streaming Service). My first thought would be to turn it off, set to manual or disabled, and see if the problems go away.

 

[John Connor]

I have read in Newegg reviews that people were setting their RAM manually instead of auto as that was causing problems, but it could very well be that service running. I have a GTX 560 TI and never noticed this service. I'll have to check that out. If I do have the service I haven't had this kind of trouble.

Check this out! https://esports.geforce.com/forums/topic/672846/geforce-experience/nvstreamsvc-exe-memory-leak/

 

[Jhkh]

Check this out! https://esports.geforce.com/forums/t...e-memory-leak/

The symptoms are very similar even if the memory wasn't full when the crash happened, but it may as well be the problem. I'm really looking forward for the moment I get my ram, it would be great if the hardware isn't involved.

 

[Jhkh]

Hi guys, I just got the new ram (sorry for the long wait). I installed the newly released update (wasn't available before) for nvidia drivers and desactivated streamer service, and it seems nothing is crashing so far, but I have only tried battlefield 3 for an hour (nice improvement anyway!).
As the origin of this isn't very clear, I will try again one thing after the other to figure out the culprit, so I will run windows 8.1 and other games/programs.
To be continued

 

[Jhkh]

Here are the news : crashes are back after a whole day!!

I looked at rammap before and after a crash, and the memory doesn't seem to be eaten by anything.
I also desactivated nvstreamsvc, and the errors in event manager are gone.

I changed the graphic card for a small nvidia GT610 but still games are closing without warning. The ram frequency, timings and voltage are set manually.

I ran OCCT (linX didn't work) and tested the cpu for an hour, and it ended without problem.
By the way, the motherboard seem to be adjusting frequency at will, as it goes between 1.5 to 4 Mhz depending on use.

I'm quite despaired to find a solution, and I'm thinking about paying the store to do the build again from the start with the hope they can do better than I did :'(

 

[hurrakan]

Maybe something to do with this?

http://www.tomshardware.co.uk/forum/331787-30-970a-compatibility-fx6300

 

[Jhkh]

Maybe something to do with this?

http://www.tomshardware.co.uk/forum/...ibility-fx6300

Alas, no, because bios is up to date. Bios version : E7693AMS V2.6

 

[hurrakan]

Several people in that thread say that this board does not work in Windows 8 when the BIOS is set to IDE mode. They say it has to be set to AHCI.

 

[Jhkh]

Several people in that thread say that this board does not work in Windows 8 when the BIOS is set to IDE mode. They say it has to be set to AHCI.

Gotta check that out.
For now, it can be either the PSU, the processor or the motherboard.
Some people around me say a 550W PSU is not enough for this gtx 660 OC together with this proc, and that the small graphic card cannot run the game for long. Others think the mobo or proc may be suspected.
For now I'll just unmount and head back for the store, as they told me their work can be supported under warranty if I let them do the job with my components.

 

[chusteczka]

When you say "crash", what is happening? Is the computer instantly shutting off, is it going to a blue screen error, is the application or game shutting down while Windows continues running?

Without more detailed information, my instincts question if the CPU or motherboard is overheating. Install CPUID HWMonitor, play a game or run Prime95 for a few minutes and watch the temps. Double check the motherboard chipset heatsinks for a plastic film that might be reducing heat transfer. Also double check the chipset heatsinks that they are firmly in place. It may help to remove them and replace the thermal paste.

Run memtest. Then run Prime95.

I get the impression you are learning (which is fine, no problem), so we need to return to ground zero and make no assumptions.

Is the 8-pin power supply connection fully inserted and connected to the CPU power connector to the top left of the cpu socket?

It might help to manually set the memory to a lower speed setting on AUTO in the bios while troubleshooting. This will ensure the problem is not caused by improperly set memory timings.

http://www.cpuid.com/softwares/hwmonitor.html
http://www.mersenne.org/download/

s/defectuous/defective/

 

[Ketchup]

Gotta check that out.
For now, it can be either the PSU, the processor or the motherboard.
Some people around me say a 550W PSU is not enough for this gtx 660 OC together with this proc, and that the small graphic card cannot run the game for long. Others think the mobo or proc may be suspected.
For now I'll just unmount and head back for the store, as they told me their work can be supported under warranty if I let them do the job with my components.

Assuming your PSU is not a dud, 550 should be fine. Couple things about your results though:

If you are running an old version of CPUID, that could account for the strange readings. I have also seen that AMD systems in general aren't the best at reading volts/temps. I never had a problem with mine, but I haven't had an 'current' AMD system in several years.

 

[Jhkh]

Some news : I handed back the graphic card to the store for them to check if the factory overclock is stable. After that they told me there is a problem and will swap it with a new one.

is the application or game shutting down while Windows continues running?

This is exactly what happens, I must have been messy in my description.

Install CPUID HWMonitor

I already posted this : http://i.imgur.com/K6uJDaO.png

Run memtest:

Check. New RAM is fine.

run Prime95 for a few minutes and watch the temps.

I ran OCCT during an hour and temps were fine, is Prime95 better for that?

Double check the motherboard chipset heatsinks for a plastic film that might be reducing heat transfer.

Check.

Also double check the chipset heatsinks that they are firmly in place.

Check, there are screws that tightens them firmly to the motherboard.

It may help to remove them and replace the thermal paste.

Motherboard is just out of the box, and I don't want to void warranty.

Is the 8-pin power supply connection fully inserted and connected to the CPU power connector to the top left of the cpu socket?

Check.

It might help to manually set the memory to a lower speed setting on AUTO in the bios while troubleshooting. This will ensure the problem is not caused by improperly set memory timings.

I changed the auto settings for the factory timings and freq of the memory. Should I underclock?

s/defectuous/defective/

Sorry I'm french, my english must be weird sometimes ^_^

Assuming your PSU is not a dud, 550 should be fine.

All right, this comforts me as it is a Corsair 80+ Bronze.

If you are running an old version of CPUID, that could account for the strange readings. I have also seen that AMD systems in general aren't the best at reading volts/temps. I never had a problem with mine, but I haven't had an 'current' AMD system in several years.

Running last version (V1.25), must be "AMD thing".

I tell you what happens when I get the replacement graphic card, there's still some hope :biggrin:

 

[chusteczka]

My apologies, it was late last night, I was tired, and I misread pieces of your description. The description of your testing wasn't clear that it was thorough enough to eliminate basic tests as root cause.

Here is what I do to ensure the system is stable as possible.

1. Build the system
2. configure the bios
3. check temps in bios
4. run Memtest through at least two cycles, overnight for 12-24 hours if possible
5. install operating system and motherboard drivers, turn off Windows automatic updates
6. install CPUID and check cpu, memory, temps, voltages
7. run Prime95 for 12-24 hours (I haven't used OCCT in a long time)
8. install Windows updates
9. run Prime95 again for an hour or two
10. install and test applications and/or games.

OCCT is similar to Prime95, one not being better than the other, so far as I am aware. Although OCCT (the last time I used it) assumes high priority while Prime95 assumes a lower priority, which allows for memory context switching and checking/configuring the system while it runs.

You are better at English than I am with French. Your description was perfectly written with exception for "2x 4Go", use 'GB' instead of 'Go'; and I do not believe I have ever heard anyone use the term 'defectuous' before. Without that 'defectuous', I would never have guessed you were a non-native English speaker.

With all of that out of the way and secure, then we are free to look at other possibilities, such as the video card when heavily used with a game.

I once ran into a problem that I thought was caused by a defective video card but was actually my own fault. I had an Automatic Voltage Regulator for a few years and had forgotten it was limited to 500watts. I built a new DDR3 system with a larger video card than I had previously used, installed a 600w PSU, then my system started crashing when using the video card. I sent in the video card and the store told me they tested it fine but on my system it didn't work. I was stumped for a long time until I realized a few months later the AVR was limited to 500w and that with the video card my system required more. Once I upgraded the AVR to 1000watts, everything was fine.

I am glad the store was able to test and determine the video card was faulty. Hopefully your problem is solved.

 

[Jhkh]

@chusteczka Thanks for the method, I'll use it for my future builds.

Finally, there was a mistake with the diagnosis report of the video card (It works xD ). So I payed for the store to rebuild my system (not that expensive), and there was no issue after that when they ran the following tests : :hmm:

Test Furmark (38min), Unigine Heaven 4.0 (5h), BF4 (2h35), Crysis 3 (6h): ok
Test video outputs (2 DVI, HDMI & DP): ok
Test windows performances: ok
Test HD video: ok
Test sleep mode: ok
Test with intel and amd plateform: ok
Software and games:
Test Assassin's Creed IV OK 2h30
Test Megui OK 6h of encoding
Test Furmark OK 30min (T°=68°)
Test Heaven Unigine 4.0 OK 1h30
Test SDK DirectX : Soft Particules OK 30min

There was probably a mistake somewhere... I will never know for sure^^

Anyway, problem solved, thanks a lot for your help everyone! I'm glad that the machine is coming back in working condition after all this mess:biggrin:
Will I ever try to build a system again, that is the question!

 

[chusteczka]

Of course you will try to build another system. All it takes is practice and the willingness to take the system apart piece by piece until the problem is discovered. There are numerous times I have been completely confused about the source of a problem and couldn't figure out how to fix it. That is why I now make sure each step is complete and properly tested before moving to the next step.