31 replies to this topic

[Golrar]

Same HDD? Did you format and reinstall windows with the new mobo and CPU?

[darqsyde]

Jesus DIED for me, on 03 January 2015 - 10:06 PM, said:

I would say it's GPU itself that is bad.

I would second this, it's the only thing you haven't/can't swapped out. By process of elimination....

[evilC]

I would bet dollars to donuts it is something to do with the supply of power.

If your PSU is on it's way out / unable to cope with the load, then you will see very intermittent and strange issues.

It can depend as much on what else is going on in the house power-wise as much as what is going on with the computer - if your PSU is having trouble coping, power fluctuations on the power lines will not be helping it.

Remember, PSUs "droop" over time - they do not suddenly stop working, they just progressively get worse.

The PSU is also one of the easiest bits to swap out and the most compatible across machines.
Find a mate with a PC and borrow his PSU...

Edited by evilC, 08 January 2015 - 04:38 AM.

[MechWarrior4172571]

evilC, on 08 January 2015 - 04:36 AM, said:

I would bet dollars to donuts it is something to do with the supply of power.

If your PSU is on it's way out / unable to cope with the load, then you will see very intermittent and strange issues.

It can depend as much on what else is going on in the house power-wise as much as what is going on with the computer - if your PSU is having trouble coping, power fluctuations on the power lines will not be helping it.

Remember, PSUs "droop" over time - they do not suddenly stop working, they just progressively get worse.

The PSU is also one of the easiest bits to swap out and the most compatible across machines.
Find a mate with a PC and borrow his PSU...

He tried 2 different power supplies (at least one of them, I think is new) and that didn't help, so the problem, if it is power input related, is somewhere else. I'd say that the GPU is throttling, for which could be number of reasons--1. Power input (motherboard insufficient input power (not all power connectors plugged in?) or GPU input (not all power connectors plugged in? not plugged in all the way? VGA's power connector's 'connector' is lose, soldering joint lose?; 2. thermal bummer where thermal paste dried up and possibly heatsink was moved after it dried up = GPU overheats due to that and bam, it goes into throttle; 3. Electrical short which causes brains of the GPU to panic and it throttles--this could be on the motherboard or the GPU but I think he changed the mobo already so that leaves GPU; 4. An electrical short in some USB peripheral or something external like hard drive causes overload on the system and GPU throttles as the result.

Edited by Jesus DIED for me, 08 January 2015 - 06:30 AM.

[MechWarrior4172571]

NeoAres, on 02 January 2015 - 08:27 AM, said:

Well, I think I found the problem, and this one actually makes some sort of sense (though everything else leading up to it makes none). I moved my video card to the lower PCIx16 slot and the problems went away. I moved it back up to the upper x16 slot and the problems returned. I moved it back down to the lower slot and the problems went away again. I ran GPU-Z with the video card in each, including once when the GPU was running in the upper slot and performance was high, and here's what I discovered, regarding the Bus Interface:

1. Upper slot-high performance: PCI-E 3.0x16 @ x16 1.1 idle, 2.0 under load (which makes sense as the mobo's slots are 2.0 unfortunately)
2. Lower slot-high performance: PCI-E 3.0x16 @ x4 1.1 idle, 2.0 under load, which means that despite the lower x# rating, performance was still higher than the following...
3. Upper slot-low performance: PCI-E 3.0x16 @ x16 1.1 idle, 1.1 under load. Regardless of the game or the render test, the bus interface refuses to go up to 2.0 and I think this is the symptom that is cause the game's choppiness. So now that I have a symptom, I need to discover its source and, equally importantly, why that source is intermittent.

I wanted to add about this problem--I just experienced similar issue. My previous MWO session was kinda choppy, so I only guess that GPU throttled or motherboard throttled due to RAM not having enough Voltage. When I exited that game the MWO exit screen hanged (not the computer.) It was the end of the day, so I shut the computer down. Next day I powered it up and went into BIOS and by accident looked into GPU slot rating--it was set at x4 speed. I know I've seen it before set at x16 so I instantly suspected something throttled due to previous experience. Then I thought of your case, so don't get too frustrated with your system--it sounds like a simple case of throttling (not slot related at all.) I suspect your GPU is overheating due to old thermal paste that dried up (not the case with mine because my GPU is new, so I suspect my RAM overclocking without enough Voltage supplied, which forced the motherboard to throttle GPU bus lanes.) Well, to make the long story short, I restarted computer after my BIOS Voltage tweak and my GPU is shown as running at x16 bus lane speed again. Try to up voltages to your components in BIOS and definitely take a look at that GPU--take the heatsink apart and clean it with acetone and then with isopropyl alcohol, and apply new thermal paste on it, and reassemble it, and give it a try.

Added on 01/11/2015--Apparently my first slot is running now at x4 bus speed even though it should be running at x16 speed and I did see it working at x16 speed before. I moved my graphics card to x8 slot and it worked at x8 speed. Then I moved it back to slot 1 and it run at x4 speed again. Then I moved my video card to slot 3 which is rated at x16 bus speed and the video card run at x16 speed. I think that my first slot in the motherboard is gone bad.. but.. I think that it may also be something that is sticking out too far on the back in that area shorting out against the metal chassis plane beneath it. Slot 1 worked fine at x16 speed just recently so I am baffled. I will have to take the motherboard out and look if anything is sticking out too far (including those funny long things on the end of each PCIe slot on the back of the motherboard--they could be touching the case). I might have to place metal washers (one or two) on the top of the stand-offs to lift the motherboard higher or I might have to place some insulating material in between the mobo and the PC case back-plane or I might have to snip the tall protruding metal pins on the back of the motherboard. It's trouble-shooting time. Download Speccy-free info software by Piriform and it will tell you in Windows environment at what bus speed your Video card is running. If your video card is running at x4 speed on a x16 bus speed lane then you have a problem. I suspect a short on the back but I will trouble-shoot this issue further. I am glad it's not my new video card though. I'll update this thread if I get my Slot 1 to work at it's proper rated x16 speed with the video card that I have.

Added, after troubleshooting: Well, I tried to clean the slot's connectors and I lifted the motherboard a bit by adding washers to the stand-offs.. no go. Looks like my Slot 1 PCIe is bad--only works at x4 bus speed for some reason. I just moved the VGA to slot 3 which is rated at x16 bus speed and actually does work at that speed, at x16. Just got this motherboard from somebody, second hand. Looks like not fully functional, but I knew what I was getting into.. just nice to find a problem though. I hope you find yours. Take care.

Edited by Jesus DIED for me, 11 January 2015 - 06:31 PM.

[NeoAres]

Glad to hear you fixed your troubles. I am suspecting the GPU at this point, though I'm still going to try reinstalling windows/nuking the HDD first to make sure. I went searching online for similar symptoms to mine (I also found several entries of similar symptoms to yours and most of them involved seating issues so I think your analysis of a bad slot is bang-on) and found several instances but zero fixes.

Regarding the Trinity APU, when I first started playing MWO I didn't have the GPU and was using the Trinity for graphics. The result was terrible, even on low settings, and prompted me to buy the 260X in the first place. This was also the big problem troubleshooting-wise. The problem with the GPU is one that has existed from the very beginning, but I didn't notice it at the time because it was still a huge upgrade from the Trinity's graphics. I thought it was working normally and that the bottleneck was in the power, mobo, or cpu (hence the recent upgrades of all). The first time I ran the computer in its upgraded state is the first time I ever noticed the high-powered performance I have occasionally been getting. And then it disappeared on the next boot-up. The defining moment of my troubleshooting was when I switched back to the old mobo and cpu and still got the high performance, albeit once again for just one boot.

Anyhow, current events:
I RMAed my new mobo (before I made the aforementioned discovery) so I'm currently waiting for the replacement to arrive. The last couple weeks I've been doing a passive troubleshooting by checking GPU-Z every time I boot up, and again at random points during my sessions, to see if the problem ever randomly corrects itself without manual intervention. It has not done so even once, so the issue cannot be something random in nature like a short or a heating issue. Whatever makes the card decide to run at 1.1 instead of 2.0, it happens on bootup and it happens every bootup unless my tinkering somehow negates it temporarily.

My next step is to reinstall windows (without nuking). I'm withholding the nuke until my replacement mobo comes back, so this won't completely rule out software as the culprit, but it's not exactly a high-risk fix attempt either so why not.

If that fails and the nuke fails, then and only then will I shell out again for a new GPU because there will no longer be doubt in my mind that the GPU is faulty and has been from the start and it took me a year to figure that out.

Edited by NeoAres, 12 January 2015 - 08:37 AM.

[MechWarrior4172571]

NeoAres, on 12 January 2015 - 08:36 AM, said:

Glad to hear you fixed your troubles. I am suspecting the GPU at this point, though I'm still going to try reinstalling windows/nuking the HDD first to make sure. I went searching online for similar symptoms to mine (I also found several entries of similar symptoms to yours and most of them involved seating issues so I think your analysis of a bad slot is bang-on) and found several instances but zero fixes.

Regarding the Trinity APU, when I first started playing MWO I didn't have the GPU and was using the Trinity for graphics. The result was terrible, even on low settings, and prompted me to buy the 260X in the first place. This was also the big problem troubleshooting-wise. The problem with the GPU is one that has existed from the very beginning, but I didn't notice it at the time because it was still a huge upgrade from the Trinity's graphics. I thought it was working normally and that the bottleneck was in the power, mobo, or cpu (hence the recent upgrades of all). The first time I ran the computer in its upgraded state is the first time I ever noticed the high-powered performance I have occasionally been getting. And then it disappeared on the next boot-up. The defining moment of my troubleshooting was when I switched back to the old mobo and cpu and still got the high performance, albeit once again for just one boot.

Anyhow, current events:
I RMAed my new mobo (before I made the aforementioned discovery) so I'm currently waiting for the replacement to arrive. The last couple weeks I've been doing a passive troubleshooting by checking GPU-Z every time I boot up, and again at random points during my sessions, to see if the problem ever randomly corrects itself without manual intervention. It has not done so even once, so the issue cannot be something random in nature like a short or a heating issue. Whatever makes the card decide to run at 1.1 instead of 2.0, it happens on bootup and it happens every bootup unless my tinkering somehow negates it temporarily.

My next step is to reinstall windows (without nuking). I'm withholding the nuke until my replacement mobo comes back, so this won't completely rule out software as the culprit, but it's not exactly a high-risk fix attempt either so why not.

If that fails and the nuke fails, then and only then will I shell out again for a new GPU because there will no longer be doubt in my mind that the GPU is faulty and has been from the start and it took me a year to figure that out.

I used Speccy program to spot at what bus speed my card run, (also possible in my ROG BIOS) not the PCIe version, like you did, but bus speed--just a clarification. I ended up giving up on that slot as being bad. I did try to clean it with a tooth-pick and acetone and isopropyl alcohol afterwards but to no avail. I suspect some micro-chip that powers that slot went bonkers.. or something similar. Possible, that it could still be connection issues but I seriously doubt it. I reseated that GPU a few times before and after cleaning of the inside of the slot--it WAS kind of dusty in there, because my toothpick ended up looking dirty, but I don't think that's the problem though, although in some remote sense it just might still be. I'll probably sell the mobo as working but having 1 defective slot. Still a good motherboard otherwise, just not for anyone wanting top slot to use or SLI/Crossfire.

[MechWarrior4172571]

NeoAres, on 12 January 2015 - 08:36 AM, said:

Whatever makes the card decide to run at 1.1 instead of 2.0,
it happens on bootup and it happens every bootup
unless my tinkering somehow negates it temporarily.

Something that you should be aware of is that Video cards would go into lower performance state on bootup unless forced into higher performance state via 3D application (or manually). I figured that part out just now--my card would stay at PCIe 1.1 at idle at x16 bus lanes but jump to PCIe 2.0 when MWO is running (also at x16 bus lanes). The key is if you have it running at maximum bus lanes that your slot is rated at.

Here is a snap shot of idle state of my VGA..



As you can see,
that at idle it is staying at 1.1 PCIe.

And here is a snap shot of MWO 3D activated state
where it jumps to PCIe 2.0 state...



The key here
is that it is running
at x16 bus lanes in both cases.

[NeoAres]

Nailed it! Finally!

The problem was my DirectX (or at least the fixable portion of what is probably a more complex issue)

I went ahead with the nuking of my HDD yesterday and everything was rosy at first. I had GPU-Z running on every boot from the very first and it was reading all systems go every time. Then an hour ago I updated my DirectX in preparation for reinstalling MWO and the problem immediately resurfaced. I uninstalled the updated DirectX and rebooted (which automatically reinstalled the original version), and the problem went away again. I rebooted a second time just to make sure this wasn't another false fix and the problem stayed gone. Fortunately, I am still able to play MWO with the old DirectX so at least where that game is concerned, the problem is fixed. Hopefully by the time I actually need the latest DirectX in order to play a game, whatever problem that exists (presumably a compatibility problem) will have been corrected.

[MechWarrior4172571]

NeoAres, on 24 January 2015 - 05:15 AM, said:

Nailed it! Finally!

The problem was my DirectX (or at least the fixable portion of what is probably a more complex issue)

I went ahead with the nuking of my HDD yesterday and everything was rosy at first. I had GPU-Z running on every boot from the very first and it was reading all systems go every time. Then an hour ago I updated my DirectX in preparation for reinstalling MWO and the problem immediately resurfaced. I uninstalled the updated DirectX and rebooted (which automatically reinstalled the original version), and the problem went away again. I rebooted a second time just to make sure this wasn't another false fix and the problem stayed gone. Fortunately, I am still able to play MWO with the old DirectX so at least where that game is concerned, the problem is fixed. Hopefully by the time I actually need the latest DirectX in order to play a game, whatever problem that exists (presumably a compatibility problem) will have been corrected.

Cool. Maybe I can get some extra fps by reverting my DirectX as well? hmm.. it's worth checking out.

[NeoAres]

Well, reverting DirectX isn't exactly easy unless you've updated it recently. It can't be uninstalled conventionally so I had to perform a system restore. I doubt that's an option for you. The only other solution I found is to nuke the HDD and start over.

The sad thing is that I can't use MWO's new 64 bit version without the updated DirectX. I'm going to send a message to Microsoft support, hopefully get the ball rolling on a patch to fix this issue.

[NeoAres]

Just for the sake of the record, the DirectX wasn't the problem. It was another false fix. I've since re-updated it and it works fine.

The real problem (and I made sure to wait a few weeks before posting about it this time) was my GPU's VBIOS. Thanks to TechPowerUp I was able to download an updated VBIOS for my card and a flasher utility. I'll never be sure whether the VBIOS actually needed updating or whether it was simply faulty and in need of reinstalling, but at this point I don't care. The system hasn't so much as hiccuped since so I consider this case finally closed.

Now I get to fight a new battle, figuring out how to get more that 30fps. The journey continues!