33 replies to this topic

[red stapler]

Badconduct, on 30 June 2013 - 03:13 PM, said:

Again, I personally, think 60 is too hot. But check your North Bridge temp. It could be in the 80's or 90's.

I've decided to edit my post. You sound like you're more familiar with an AMD platform, perhaps overclocking Phenom IIs?

Ivy bridge CPUs like his i5 are rated to run at 105C, he's well within range of that. They also don't have a northbridge, the memory controller and pci express lanes etc. are all on-die with the CPU. The only thing the motherboard chipset does is the hard drive controller and USB ports.

Edited by red stapler, 30 June 2013 - 05:17 PM.

[Qbecks]

Thank you for all your assistance.

That games are really a bit old, but my only games that don't allow me to use max settings (Witcher 2) or showing slower performance in complex situation (Metro 2033 - I don't have hard numbers thou). I play MWO on max settings and I don't have FPS issues, it only crashes from time to time, but then in very unpleasant manner.

The sound comes from speakers, it's like short loop with last second of sound I heard. But I'll look for some new sound drivers thou (I don't have dedicated SC, using onboard solution).

Power button works normal, I have to keep it pressed for few seconds to turn off, reset button is working as intended if I don't have "MWO Crash".

Thank you for pointing it out, that no devs are reading here, I hoped that they read it and could try to let it be with admin rights, at least for playing game alone (I don't mind patching with admin rights, even if it's not necessary).

I've checked system log after that crash, but all 3 entries I've found was made after reboot.(I've looked in "Summary of Administrative Events" in Windows Log Viewer):

1) Critical:The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.
2) Errors: "The previous system shutdown at 17:53:08 on ‎2013-‎06-‎30 was unexpected: and my 'standard' booting error: "Session "Microsoft Security Client OOBE" stopped due to the following error: 0xC000000D - no need to worry if I googled it right.So no help here, imho.

I disagree that 550W could be too low. Radeon HD 7850 needs at peak about 250W, my CPU about 100W but for sure not more then 150W. So it lefts 150W for rest of system - more then enough. I know that wattage means nothing if PSU can't give enough power on +12V line, but my have 38A - it's enough too.

I don't overclock. CPU has turbo modus which kicks in only if it has enough cooling. GPU is slightly overclocked by factory (to 975MHz) but I don't OC it further (I have OC options turned off in driver).

GPU temperature wax maximal 60 C while playing.

I'll check windowed modus, letting changing parts out till I'll have similar problems with other applications.
I believe that it some driver issues that would not occur if game will not force one to start it with admin rights, but I'll check further and will give my results if I got something.

Thank you all for valuable input,
regards,
Erich

P.S.Sorry if I sound rude, English is my third language and I usually read / hear it only. No offends intended.

Edited by Qbecks, 30 June 2013 - 06:55 PM.

[Dragoon20005]

i do want to point out that PSU usually don't have the ability to max out it rated wattage as the rest of the power is wasted as heat

your model is a bronze series which means it only outputs at 82% load so it is providing only 450W constant, the best PSU out there is platinum series which only outputs UP to 92%

this game is not optimized so there is a chance your CPU and GPU are drawing a lot of power from the PSU causing the trip as the PSU cant take that much power load

believe me we do have user with similar problems and the culprit is with underrated PSU giving the headaches

[Qbecks]

thanks for hint, but here are specs of my PSU:

http://www.enermaxus...enp550awt_b.php

450W on +12v line, with my hardware it really should not be a problem: CPU can't take more then 150W and GPU more then 250W, no matter how unoptimized the game is. It's still 50W below PSU output on +12V

[Dragoon20005]

best not to assume

always test!!!


or else the damage may be more than just your PSU, if your PSU blows, your GPU,CPU and mobo will go too

[Thorqemada]

You could do a look at your RAM Timings and Voltage - try slower Timing, increased Voltage.

BUS Timings can also be to close to the limit.

The PSU is a quality one that "should" work though a good PSU is only as good as the power source its coneccted too.
The power rail your GPU is connected to may be weak or overloaded.

Some 3rd Party Driver or Software could interfere with MWO - try to run your PC as lean as possible.

Edited by Thorqemada, 30 June 2013 - 07:43 PM.

[Qbecks]

maybe I stated it wrong:

Worst (highest) wattage info I've found on HD 7850 was 250W. (but professional made measures shows about 150W);
http://www.techpower...HD_7870/24.html

Worst (highest) wattage info on i5-3470 was 150W. (once again, professional made measures shows about 100W, I've just assumed worst case scenario).

It makes 400W together (worst case), and PSU specification says: 456W avilable on +12V

So it's not a assumption, I just used known data about all 3 components and as long as Enermax is not lying, it just can't be too much.

[Kaboodle]

I've gotten this before, screen hard freezes, and sound shrieks w/e is currently last played. I narrowed it down to a ram error after LOTS of testing. It's probably a small bad sector in your ram, that you rarely ever actually hit. (Very possible nowadays since most games still run in 32bit mode which caps out at 2gigs of ram, with most systems having 8+.) Easiest way to test is to remove a single stick if you have 2, and try and recreate the event. If it does it, swap sticks and it shouldn't. If you have dual channel or only one stick, you can try one of the MANY software based ram testers, but they can often take hours to run a truly thorough check.

[Catamount]

For the record, PSUs do output all of their stated wattage (if they don't suck); it's just that they draw more. An 82% efficient PSU doesn't draw 550W from the wall and only output 451W; it outputs 550W, but draws 670 from the wall to do it.

With that said, OP, that 12v rail really isn't that great for stability. It's decent enough that I don't think it's your problem, but I will note that it could be better. I'd just keep an eye on it as time goes on, because MWO is clearly drawing a lot of power out of your PSU.


Also, Tarl Cabot is correct about fullscreen mode. You want to run in a windowed mode, because it's a lot less buggy. Full Window gives you the same experience, only it's more stable.

Edited by Catamount, 01 July 2013 - 06:17 AM.

[Alex Warden]

Loc Nar, on 30 June 2013 - 09:26 AM, said:

Definitely some kind of problem with your machine. My puter is not much nigher spec'd (3570k/aftermarket cooler/7850OC) and I pull 50-60fps on max settings with no issues.

Almost sounds like your cpu cooler might not be properly seated...

to me it also sounds like an heat issue... i have almost the same which OP describes in APB (sound crackles, FPS drops etc.) if i don´t force my GPU cooler to 85% permanently... and i had a temperature shutdown in MWO once...

since i manually force my cooler to do it´s job, everything is fine... (Athlon II X4 3.0ghz, GTX 260,8 GB DDR3... 40 FPS medium - high settings)

Edited by Alex Warden, 30 June 2013 - 09:44 PM.

[egreSS]

It's either GPU or GPU power consumption. To prove it you can try this.

Video Memory stress Test is a program similar to MemTest86+ but for GPU/Video RAM. You can test the memory using DirectX, CUDA or OpenGL.

There is a windows version, and a bootable .iso or floppy version available as well. Try this and post results.
http://mikelab.kiev....programs/vmt_en

Being ATI and MWO, select DirectX.

Edited by egreSS, 30 June 2013 - 10:13 PM.

[Badconduct]

egreSS, on 30 June 2013 - 10:10 PM, said:

It's either GPU or GPU power consumption. To prove it you can try this.

Video Memory stress Test is a program similar to MemTest86+ but for GPU/Video RAM. You can test the memory using DirectX, CUDA or OpenGL.

There is a windows version, and a bootable .iso or floppy version available as well. Try this and post results.
http://mikelab.kiev....programs/vmt_en

Being ATI and MWO, select DirectX.

You can run Memtest CL, but you would have to run all of them at once to really recreate the problem.


OP, how many fans/HD's are you running on your system?

[Badconduct]

For error #2)
Try this
http://answers.micro...6f-7aee8b31f49a

But for error #1), a sudden loss of power.... I'd listen to the above.

[Catamount]

If the OP's GPU is a problem, I would think it would show symptoms on Metro 2033. Besides, testing VRAM only shows unstable VRAM. It's not going to stress the GPU itself enough to reveal problems on that end.

I think the OP should start with power consumption, however. Running Prime95+Furmark would pretty thoroughly eliminate power as a possibility, although I have to warn that if a system is running too hot and/or is getting bad power, Furmark can be downright dangerous, so it's not something I'd run on a system I wasn't confident in.

Edited by Catamount, 01 July 2013 - 06:23 AM.