MSi r7950 Twin Frozr iii 3gb/boost, possible PSU issue?

[monogwai]

back in December I upgraded to the r7950 (7970 pcb, 6+2 pin and 6 pin). Unfortunately, I hadn't noticed that one of the games I was hoping to play called for a 7970 minimum. So, I flashed the unlocked BIOS to that of an r9 280, which took care of that problem quite fine. No issues with it whatsoever.

Flash forward to this past Sunday afternoon. I had gotten a new HDD-- I was upgrading the 7200 rpm drive I run Windows off of to a 10k rpm drive. So, thinking nothing of it, I just added the new HDD in with the other 3 I run, thinking I'll do a quick Windows install, port over the system settings and save games and such off the old drive, and then decide which HDD I didn't need so much anymore. Unfortunately, I never made it that far. What started happening was, I had at one point needed to open the old Windows install, but the old Windows wouldn't boot. I'd get the start up screen, and then just black... no Welcome screen, but the monitor never went into standby. After much handwringing and head scratching, it turns out that I just needed to boot into safe mode and uninstall the video driver. I had been using the setting in the new Crimson drivers that "extends" your display resolution by doing some kind of function on the card to make more resolutions available on the LCD (i forget the name of the feature, but i'm sure you guys know it), but for some reason, my monitor wasn't digging it anymore. So, having taken care of that, I get into Windows. Any time I tried to open a program like Firefox, I'm greeted by this fullscreen blue-grey color... not a "blue screen" or a lock-up, per se, but I still can only get away from it via the Reset button. So I immediately try switching back to the original BIOS of the video card-- Same thing. I underclock the video card-- Same thing. I unplug 1 HDD, then another, then another-- Same thing. Now, even running with just 1 HDD and nothing else but fans, i get the same thing. I wondered if maybe the HDMI was going bad, but I get the same result with the other output. So I ran GPU-Z and logged the temperatures to a text file, and started causing the blue-grey screen to come up, and I'm showing no problems temperature-wise in the GPU or the VRAM. Both are idling well under 40c. (I even got a blue-grey screen trying to open the freaking text file...) SO... all things considered, I'm guessing that just having added the 4th HDD to the mix just stressed my PSU to the point that now it's permanently crippled? It's only a 500W Antec Earthwatts... I was foolish and didn't consider the output of the thing when I just slammed that new HDD in... Anyhow, I just wanted some alternate opinions. Maybe something's wrong with the new Crimson drivers (I have the new 2016 update) or maybe there's something i'm missing altogether.


edit: I'd also like to add that I do no OC whatsoever. I've seen some posts where people are changing the ULPS settings, but since I don't overclock or have multi-GPUs, i don't know if this would help at all?

 

[natr0n]

omg what did i read

unplug all your drives leave only boot/windows drive.

use this tool http://www.wagnardmobile.com/forums/viewtopic.php?f=5&t=164

it will reboot you into safe mode. click uninstall drivers reboot option....reinstall drivers after it finishes loading into windows

 

[monogwai]

omg what did i read

unplug all your drives leave only boot/windows drive.

use this tool http://www.wagnardmobile.com/forums/viewtopic.php?f=5&t=164

it will reboot you into safe mode. click uninstall drivers reboot option....reinstall drivers after it finishes loading into windows

okay... so you're saying my drivers are the problem? Otherwise I'm unsure as to what you might have read, because I already took care of my booting issue. I already have DDU, so I'll wait patiently until you explain exactly what you mean.

edit: and in the interest of speed, i'll go ahead and ask: If my drivers are the issue, what drivers are you suggesting I replace them with? pre-Hotfix Crimson drivers, or what?

 

[natr0n]

okay... so you're saying my drivers are the problem? Otherwise I'm unsure as to what you might have read, because I already took care of my booting issue. I already have DDU, so I'll wait patiently until you explain exactly what you mean.

edit: and in the interest of speed, i'll go ahead and ask: If my drivers are the issue, what drivers are you suggesting I replace them with? pre-Hotfix Crimson drivers, or what?

Your gpu needs at least 30 amps on psu. Your psu is rated for 22. Hence random screen freezing. So basically yes you were right its a psu issue as youre overloading it with many drives as well and a 10k drive on top of that.

 

[monogwai]

Your gpu needs at least 30 amps on psu. Your psu is rated for 22. Hence random screen freezing. So basically yes you were right its a psu issue as youre overloading it with many drives as well and a 10k drive on top of that.

ok, good deal. that's what i thought.

 

[Filip Georgievski]

I would not point at the PSU.
Since you pointed out that trying only with 1 drive still gives you the problem.
Had a similar problem with a card as well.
Ended up switching for more power hungry card, and no issues what so ever.
Check my specs, i was going for a Dual X R9 280x but ended up getting a more powerfull Vapor R9 280X and no issues what so ever.
If your Antec is real 500w, it should hold this card. But my advice is for you to reset bios, install fresh Windows on it, use 15.7 catalyst drivers and do a stress test with only 1 HDD.

If your PSU cant supply the card, it wont boot in the first place since when booting, your system is checking for errors itself, and the beep you get (1 beep) says it is all good.
Now if you get 2 beeps or more continues beeps, means something is wrong with your system.
Weird thing in mine and your problem was/is that screen crashes when doing simple stuff. I did get a bunch of cards and tested all, and in my problem, only the 1 i got made this kind of problems. I suggest you do the same.

 

[EarthDog]

So, I flashed the unlocked BIOS to that of an r9 280, which took care of that problem quite fine.

So you flashed it to the same card... and things changed? The r9 280 IS a 7950....

Your gpu needs at least 30 amps on psu. Your psu is rated for 22.

The GPU itself doesn't need 30A. That is a SYSTEM rating.

Second, his PSU can output 408W or 34A on the 12V rail(s) assuming its in good shape. So his PSU should be fine, its a quality PSU and not some proven hunk of junk like Diablotek, Hantol, PowMax, etc...

Assuming the PSU is doing what it is supposed to still as far as putting out its rated power, then that isn't the culprit. Contrary to its label (like many PSUs) this is also a SINGLE rail PSU so its not a problem with overloading a rail either.

I would look into drivers or flash that back to what it 'used to be' since your flash did nothing but change the name and clockspeeds anyway.

 

[monogwai]

So you flashed it to the same card... and things changed? The r9 280 IS a 7950....

...hence why it's taken the new BIOS so well, I guess? Shoot man, I don't make the rules. All I know is: 7950 = no go, r9 280 = go on ahead, sir. Don't ask me, ask Square Enix.

I would look into drivers or flash that back to what it 'used to be' since your flash did nothing but change the name and clockspeeds anyway.

Flashback not necessary, I can simply flip the bios switch. Which I did try...

I would not point at the PSU.
Since you pointed out that trying only with 1 drive still gives you the problem.
Had a similar problem with a card as well.
Ended up switching for more power hungry card, and no issues what so ever.
Check my specs, i was going for a Dual X R9 280x but ended up getting a more powerfull Vapor R9 280X and no issues what so ever.
If your Antec is real 500w, it should hold this card. But my advice is for you to reset bios, install fresh Windows on it, use 15.7 catalyst drivers and do a stress test with only 1 HDD.

If your PSU cant supply the card, it wont boot in the first place since when booting, your system is checking for errors itself, and the beep you get (1 beep) says it is all good.
Now if you get 2 beeps or more continues beeps, means something is wrong with your system.
Weird thing in mine and your problem was/is that screen crashes when doing simple stuff. I did get a bunch of cards and tested all, and in my problem, only the 1 i got made this kind of problems. I suggest you do the same.

Well, the story keeps getting more and more interesting as the day goes on. So, last night I got my hands on a spare PSU and one of those ATX daisy chain doohickeys, so now I'm running the video card on its very own 400W PSU. Hooked the HDD's back up (to the original PSU, of course) and got on with business. Immediately things seemed better-- the advanced resolution with the video drivers was working, I was able to open Firefox and Thunderbird, and everything was carrying on as it should. That is, until I read the email informing me of the replies to my message. The second I clicked on the Log In link here at the forums, I was greeted by the lovely Monochromatic Screen 'O Doom (tm), though this time in a drab grey, not the lovely blue-grey from before. Remembering what little I'd read in the replies, I went ahead and ran DDU and installed Catalyst 15.7, and still I get the same issue with trying to log in to the forum. So I've since rebooted (again), and however inexplicable, I am able to successfully log in to the forums with IE (guess it's not totally useless after all!), which brings us to right now.

So, now I'm not sure where to look next. I didn't look at the ratings on the new PSU, but could it be possible that a 400 watt PSU could have trouble running just a single r7950 at stock speeds? Could the card be damaged, somehow? I'd imagine if I had fried some component, I'd have certainly smelled that. I have no problems taking the fan/heatsink assembly off the card and taking a peek around, if anyone has any ideas what I should be looking for... (before anyone asks, I am well stocked in TIM).

Having said all that, let me ask you Filip-- you mention you tested several cards, and only the 1 was giving you problems like mine is... so, end result was, you replaced the card with another? No ideas on what caused the problem to start in the first place, or how to take care of it? This isn't a new card, but it's new to me, and I'd hate to have to replace it so soon. Especially when it was working flawlessly until I just added a component to my system. Seems a silly reason to go haywire. <shrug>


edit: just took a glance at the label on the secondary PSU, and it's got 348W and 29 A on the 12v rails. My multimeter has no batteries though, so I can't verify that.

 

[EarthDog]

You can't verify that with a MM in the first place, it doesn't read capacities, just voltage/impedance, etc. So long as it isn't a hunk of garbage PSU, it should be fine for 250W load at max.

At this point, I would test the card in another PC if you have not already. I would also test by removing the HDD, again, if you haven't already.

 

[monogwai]

You can't verify that with a MM in the first place, it doesn't read capacities, just voltage/impedance, etc. So long as it isn't a hunk of garbage PSU, it should be fine for 250W load at max.

At this point, I would test the card in another PC if you have not already. I would also test by removing the HDD, again, if you haven't already.

Well, I was thinking I'd effectively removed the HDD(s) out of the equation by isolating the video card on its own PSU. But, for the sake of arguments, I unplugged all but the boot drive. No change. Well, that's a bit disingenuous. It in fact changed for the worse-- I'm now responding to this message on a different PC, as not even IE would allow me to respond as it had done previously. I decided to do a remove/reseat, and while I had the card out, I took off the fan/HS and looked around. Everything looks fine, no damaged capacitors or mosfet here. Anyhow, to address what you've said, it's def. not a "hunk of garbage PSU", it was taken from a working system.

I'd like to test the card in another PC, but I don't have another PC with a PSU capable of supporting this card, else I'd have just swapped PSU's with that machine at the beginning of this problem... no friends nearby who'd have an appropriate system, either. I mean, I used to run a GTX 295 in this machine, which required 313 watts. 113 more than this card requires. The new HDD only uses 6.4 watts (4.3 @ idle). So, i'm using 106.6 less watts than before, and that causes me trouble? That's just not making sense.

 

[cdawall]

I run my overclocked 7950 on a single rail 300w FSP. It isn't the PSU. Sounds like a software issue not a hardware one.

 

[EarthDog]

it's def. not a "hunk of garbage PSU", it was taken from a working system.

Working and hunk of garbage are not synonyms nor interchangeable. PSUs can work, and still be hunks of garbage as far as their quality is concerned. There are concerns about if it can output its labeled power and clean (in spec ripple, transient response, etc), and at what temperature etc. So just because its working, doesn't mean its any better or worse.

 

[monogwai]

I run my overclocked 7950 on a single rail 300w FSP. It isn't the PSU. Sounds like a software issue not a hardware one.

Well, there's a lot of details you've left out, i.e. how many other devices you have drawing off that PSU. As far as it being a software issue, I'm certainly open to the possibility, but nothing has changed, software-wise. And i've rolled back drivers anyway, ones that I'd already confirmed as "working" prior to this mess starting. The only thing that has recently changed was the addition of the 10k rpm HDD. Which I subsequently removed from the system, and still couldn't restore proper working order. I've now removed the Antec PSU from the list of possible culprits, and I'm still having issues. Realistically, the only thing I can really consider is that the GPU is just slowly giving up.

Working and hunk of garbage are not synonyms nor interchangeable. PSUs can work, and still be hunks of garbage as far as their quality is concerned. There are concerns about if it can output its labeled power and clean (in spec ripple, transient response, etc), and at what temperature etc. So just because its working, doesn't mean its any better or worse.

well, I don't even know what "spec ripple" and "transient response" are... so I'd have no way of gauging the unit for these measurements. But I'm all ears, if you know a program or something that'll elucidate these specifics for me...

 

[EarthDog]

The underlying point: A working PSU is not (always)(but can be)a QUALITY PSU.

You can't test for that stuff without expensive equipment.

What brand and model is this new PSU?

 

[monogwai]

What brand and model is this new PSU?

The news on that isn't good, I'm afraid. It's an APEX SolyTech (who?) SL-400TF
http://www.soly-tech.com/ch_product_data.php?pid=48&class_type=1&mainid=&subid=&page=1

I meditated a moment on your words concerning the quality issues that this secondary PSU could have, and decided to try it running all the HDDs instead. So now the Antec is powering the mobo, gfx, dvd-rom, and 2 small case fans. We'll see if this has any positive effect on the issue in the coming hours. First time I tried the system in that configuration, I got all sorts of patterns on the display... I immediately cut power and jiggled all the connections, made sure they were good and tight, and powered back up. So far, so good... we'll see.

 

[cdawall]

Wait so you replaced your antec with a solytech? lol awesome.

 

[monogwai]

Wait so you replaced your antec with a solytech? lol awesome.

no.... I added the solytech to the mix in effort to try and see if the Antec was under supplying power to the card. Doesn't look to have been that way.

 

[cdawall]

I wouldn't use a solytech to power a fish tank pump.

 

[monogwai]

Okay, so I wanted to take an easy test drive. I opened a dx9 game and watched the intro for a while. All was good. So, I decided to open up aida64's preferences screen, as I had yet to set up the OSD for my new win install. Merely moving the options up and down in the list gave me the grey screen. HOW CAN I DO FULLSCREEN GRAPHICS IN A GAME, BUT CAN'T FIDDLE WITH SOME FREAKING SETTINGS IN SOME STUPID PROGRAM?!?!?!?!?!?!?! Sure, the graphics were only a mild use of the gfx card's abilities, but it was much more than is necessary to reorder some headings in a freaking GUI. I swear to jeebus i'm going to bash my head into the nearest brick wall with great force if something doesn't work this out-- and quickly too. (the work-out, not the bashing).

 

[monogwai]

I wouldn't use a solytech to power a fish tank pump.

well, when one is between a rock and a hard place, and is limited in options, one does what he/she has to in order to figure out what's up. I'm not planning on using this stupid thing for any long period of time. Just trying to get its help to try and determine if i've got a power problem, or a hardware problem. or even a software problem, as someone had previously mentioned, though I don't have any idea how that could be the case...

Anyhow, I think i'll go dig up my old GTX 295, as it could at least (possibly) answer the question about the Antec's maybe being jacked up. If the 295 will still run without a hiccup, then clearly the video card is to blame here. right?

 

[cdawall]

It's crashing under 2D clocks if the powersupply is that bad you would see way more issues. Do a clean install of the OS and try again.

Here is that cards power consumption under 2D clocks

Even average under load wouldn't be an issue.

This is a software issue. Correct it by doing a clean install.

 

[monogwai]

It's crashing under 2D clocks if the powersupply is that bad you would see way more issues. Do a clean install of the OS and try again.

Here is that cards power consumption under 2D clocks

[handy graph]

Even average under load wouldn't be an issue.

[another handy graph]

This is a software issue. Correct it by doing a clean install.

WELLLLLL.... customarily I'd take your word for it (seeing as I came here to be advised), but this time I don't think that's necessary. My reasoning is three-fold: A) card ran w/o a problem on my Win 7 install on my 400 GB 7.2k rpm drive; B) the install on the 300 GB 10k rpm drive is a fresh install; C) I already swapped out my r7950 for the GTX 295 (as I had mentioned in my previous post) and it works 100%. Already gone and did all the things that made the r7950 give me that lovely Monochromatic Screen 'O Doom (tm).

That being said, I certainly do appreciate your researching and posting the handy bar graphs.

I just wonder what could have happened. I know for certain the gfx card wasn't even bothered during the HDD install. My case has a little metal sleeve holding the HDDs that pulls out, you just screw the drive into place, and slide it back in. Didn't even have to move the case-- just unplug, unscrew the 2 thumb screws holding the side panel, and get busy. Guess this card's gonna be going back on the market. :*( I think I will attempt to do due diligence and test it in another machine before I part with it, though... just in case.

Meanwhile, anyone interested in a MSi r7950 Twin Frozr iii 3gb/OC Boost that's having some kinda problem with its 2D clocks?