• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

LinX Test shuts down pc after 10 seconds

Associate
Joined
7 Jun 2010
Posts
482
Location
Tamworth
Hi all

I tried to run a linX test twice this morning to see how my pc is running the GFlops.

Anyway ive had my pc over a year now and had it all setup by OcUK guys and rjkoneill all info is in my sig.

Problem i had is after 10 seconds the comp just turns off now the only thing i have changed is installing windows onto a new M4 64GB SSD everything else is as how it was from first setup.

All my bios settings i had that were stable and correct that i was given by rjkoneill were written down so i just had to put it back in after changing bios.

So for the life of me i cannot understand why my pc shuts down.

Any help appreciated.

blitzyuk
 
Maybe after a year of use theres been a little bit of degradation and it needs more voltage to be stable in extreme stress, this kind of thing is not hugely uncommon.

Throw OcUK a note in the CS forum, they're the experts :')
 
Good shout actually, if its year old I didn't even think of dust buildup or temps lol :3

That's a good place to start indeed
 
Its cooled by the H50 in push pull. I also de-dust my pc usually every 2 months and get the vacuum on the radiator to suck out and bits that dont budge with the use of compressed air. I did a test the other day started off with:
problem size 1000 memory 10mb run 10 times all ok.
problem size 2000 memory 34mb run 10 times all ok.
problem size 3000 memory 72mb run 10 times all ok.
problem size 4000 memory 127mb run 10 times all ok.

Then did:

problem size 5000 memory 196mb run 10 times and that failed to even get past 2 seconds on first run just shut down pc.

So do you think i need to increase the voltage?
 
What temps do you get when running those tests? yes you have a H50 but the pump could have stopped working.

The reason I asked for temps is because a computer switching itself off sounds like a overheat protection mechanism kicking in, normal crashes usually result in BSOD's, freezes or errors in Linx.

Also might help people if you list your full specs.
 
Last edited:
Hard shut downs are usually either the PSU is knackered (and the power draw under load kicks in the OVP).

Or overheating.

Personally, with how fast it cuts out I'd say it's more likely to be the PSU.
 
Here is my full spec as requested.

- Case: Antec 300 Gaming Case - Black
- Power Supply: OCZ 600W SteathXStream II
- CPU: AMD Phenom II X6 Six Core 1090T Black Edition 3.20GHz oc'd to 4.GHz
- Motherboard: Asus M4A88TD-V EVO/USB3 AMD 880G (Socket AM3)
- Cooler: Corsair H50 Watercooling
- RAM: Corsair XMS3 4GB (2x2GB) DDR3 PC3-12800C9 1600MHz
- Hard Drive:
- Crucial RealSSD M4 64GB SATA 6Gb/s
- WD Caviar Black 640GB 3Gb/s 32MB cache
- 2X Samsung 500GB in spanned volume 931GB total.
- Graphics Card: Powercolor ATI Radeon HD 5670 1024MB GDDR5
: Sapphire ATI Radeon HD 5670 1024MB GDDR5 (CrossfireX)
- Sound:
- Creative X-fi Extreme Gamer 7.1
- Optical Drive:
- LG DVD+/-RW SATA Drive
- LG Supermulti Lightscribe DVD+/-RW ATA Drive
- Microsoft Windows 7 Professional x64
- NZXT Sentry LXE Fan Controller

My temps when running the tests go to anywhere between 50 - 60 degrees C
Current temp is 45 degrees C
 
As you've already noticed, very small problem sizes in linpack (certainly less than 1 GB memory usage) do very little to stress the CPU. As such the temps are also lower than they would be if you ran linx correctly.

The GFLOPS measured in Linpack are strongly dependent on the problem size (mem use) with a theoretical peak of 24 times the clock speed (for your 6 core phenom anyway, as it can perform four operations on each of its six cores, 6*4=24). At 4 GHz this is then 4*24=96 GFLOPS.

As I said already, for low memory usage you won't see anywhere near 96 GFLOPS, which is why it's not a stress test. Only by increasing the memory usage to a couple (or more) GB will you see the GFLOPS tending to 96, and probably maxing out around 80.

Good luck with finding the cause of the shutdowns :)
 
Right i may have sorted it i ran a test 5 times at 1846 MB ram problem size 15500 but went to settings and it was in 64bit mode so changed it to 32bit mode and passed all 5. I also ran prime 95 for 30 minutes to kick out some heat to see if it was a cooling issue and that worked fine (i know you have to do it at least 8 hours).

Its a bit warm in my office with this weather so temp goes upto 60 degrees C but going to try run linx 10 times with same setting.

post back with results if any.
 
Successfully ran 10 times without shutdown:


Seems it was a problem running in 64bit mode.
My GFlops started at 50 and went down to 41 but i remember it being like this when i had it it started at 61 and went down to 40 something.

not sure why even rjkoneill didnt know why.
 
As I said before, the GFLOPS column is a measure of how intensive the test is. 50 is too low, let alone 37, which is probably why it passed. By changing to 32 bit you're giving it an easier time, so it doesn't cack itself :)

You're not out of the woods yet - it should be able to happily do 80 GFLOPS, until it is you've got a problem.
 
Successfully ran 10 times without shutdown:


Seems it was a problem running in 64bit mode.
My GFlops started at 50 and went down to 41 but i remember it being like this when i had it it started at 61 and went down to 40 something.

not sure why even rjkoneill didnt know why.


GFlops dropping like that means your CPU is throttling.

Either the actual CPU is getting too hot, or the PWM circuitry on your mobo is overheating and reducing the power draw of the CPU to cool itself down buy throttling the CPU.

You will probably find if you watch CPU-Z/CoreTemp while running the test the CPU speed will drop to 1600MHz or so now and again while it throttles.

I used to get that on an old mATX board when running a 1090T and the board couldn't handle the power draw of the 6 core.

1.5V on an X6 @ 4GHz is a LOT of power draw.

Also not being able to run in 64bit mode indicates that the CPU overclock isn't actually stable, as Phenoms are known to clock higher (and be stable at higher clocks) in 32bit mode. Which is probably why your CPU will pass in 32bit but not 64bit.


Try putting the CPU back to stock and running the tests again, I bet it'll pass them all no problem and the GFlops won't drop over time.
 
Last edited:
Right then.

Put cpu and anything that was overclocked back to stock i.e. loaded setup defaults in bios.

CPU is now at 3.2Ghz with core voltage set as 1.46 V re-run LinX in 64bit mode with problem size set to 19070 and all ram set as 2790 highest GFlops was 60.9945 it stayed at 3.2Ghz throughout and vcore stayed the same.

Now do i leave it at stock or return back to overclock that OcUK guys had got it set at?
 
So it could be either the PSU or temps. HeX may be right and it may simply be running too hot, which you can easily check by seeing if it's throttling.
Or it's the PSU. The PSU is NOT certified for xfire which obvs isn't great, and the load power consumption of the machine is pretty hefty - I reckon approx:

190W on the CPU (max)
123W each at idle for the GPUs = 246W
Maybe 60W on the mobo

Must be a good 550W, which is getting close to the peak.

It looks also like the PSU can provide 18A on each of four +12V rails. Might be a good idea to try and see if you can spread the mobo and graphics across the rails more equally.
 
Last edited:
cpu showing 141.60 W max as this is what the cpu boasts as only using 141.60 W. I don't know what your on about with the last bit after the 18A on each of four +12V rails.

I will at a later date buy another higher wattage psu but a modular one so much easier to manage cables.

Im leaving cpu at 3.2Ghz for the time being although never had any bsods with it at 4Ghz.
 
Back
Top Bottom