dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
2157
share rss forum feed


berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast

EVGA GTX 470 with Corsair HX520W PSU, is ok?

I recently picked up a used GTX 470 to replace a GeForce 9500 GT in my Asus P5Q Deluxe. The two PCI-E 6-pin power plugs are connected. Specs for PSU here and for the GTX 470 here.

QX6700 CPU replaced the E8400 a bit before switching graphics adapters and a little problem started with the GTX 470 upgrade, afaict. Occasionally, on cold boot, in the morning or end of day, mostly, X fails to start the screen or, other times, login screen will crash to reboot. After the first crash, there doesn't seem to be any problem.

I've run memtestG80, EVGA OC Scanner, looking for errors but not seeing any.

There are 5 hard drives running off this board. I see
Minimum of a 550 Watt power supply. (Minimum recommended power supply with +12 Volt current rating of 38 Amps.) in those GTX 470 specs and
Output +3.3@24A,+5V@24A,+12V1@18A,+12V2@18A,+12V3@18A,-12V@0.8A,+5VSB@3A for the Corsair.

Does it look like this PSU might be underpowered for this machine? Thanks.

System Information
Operating System Mandriva Linux 2011.0 3.2.18-nrj-desktop-1mib x86_64
Processor Intel Core 2 Quad Q6700 @ 2.67 GHz
1 Processor, 4 Cores, 4 Threads
Processor ID GenuineIntel Family 6 Model 15 Stepping 7



pog4
Premium
join:2004-06-03
Kihei, HI

playing at »extreme.outervision.com/psucalcu···lite.jsp using some assumptions, it recommended 560W for your system. 520W might cut it but your PSU has 3 anemic rails at 18A each. not sure how you've wired everything up but maybe you've overloaded one or two rails?

You might consider grabbing a decent 650W PSU with a single 12V rail.
--
My Site



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast

Yes, I looked at the 38A 12V requirement for the card, saw 3x18A 12V rails, and wondered whether the PCI-E connectors are on separate rails, how much does the card draw from the slot. Anyway, that is seeming a marginal condition. Thanks for the advice.


asdfdfdfdfdf
Premium
join:2012-05-09
kudos:3
Reviews:
·Verizon Wireless..
reply to berserken

I really doubt that power is a problem. You should have enough 12v power with that supply. If power was a problem it would probably exhibit in a more random way. Have you tried pushing the graphics card with games or something?

You are sure that the system was stable after you replaced your cpu and before the graphics upgrade? You gave it enough time to determine properly?



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast

It's only an occasional, short-lived problem that one reboot has solved for the rest of the uptime, so far, so I haven't been tracking it too closely. I don't do games but ran the memtest stressor and one other game demo sort of thing that's supposed to stress all the features, can't find it right now. I've got a multi-boot with XP, arch, mageia, mandriva main OS, and XP in VirtualBox that keeps me bouncing around with different types of programs and who did what where gets lost, sometimes.

Oh, this is it: Unigine Heaven DX11 Benchmark 3.0 Linux That's the benchmark with game animations that ran fine, don't see any fail in the log.
Thanks for the advice.


asdfdfdfdfdf
Premium
join:2012-05-09
kudos:3
Reviews:
·Verizon Wireless..

If it was a power issue then I would expect stressing the system would tend to make it more unstable.
How long did you use the system after you replaced the processor but before you added the graphics card? You never had any such issue during this period before adding the card?



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast

I'm looking at not quite a month, 4/13-5/11 between one email about the qx6700 and one email about the gtx470, so, about a month, a little less. I'm pretty sure the apparent video crashing issue only happened after installing the gtx470. Since about the time I started this thread, it hasn't been an issue. Like I said, it's relatively rare, maybe as often as a couple of cold boots in a row, but not for a few days, now. I'll hold off throwing parts at it for a little while. Thanks.



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast
reply to berserken

Keeping the thread warm. Just keeping track, there have been a couple more early-session crashes just this morning and yesterday. One was while starting OpenOffice to view an odt document, which entails loading a pretty big program.

My expertise with computer hardware is limited but, however unrelated to reality, my thinking includes the following observations.

1. Crashes always occur after cold boot and don't recur after reboot.
2. The crash triggers a reiserfsck and rebuild of dirty filesystem on next boot, at least the last two times I was watching.
3. Card requirements/recommendations include Minimum of a 550 Watt power supply. (Minimum recommended power supply with +12 Volt current rating of 38 Amps.)
4. PSU specs 520W: +12V1@18A,+12V2@18A,+12V3@18A
5. If pci-e connectors are fed from different rails, that seems to be 36A rated supply, vs. 38A card requirement.
6. I've seen other posters, other forums, speak as if splitting the 12V power can lead to an insufficient supply condition.
7. It's not hard to deduce that the power supply is marginal and fails, occasionally, when cold.
8. I've opened the case, pushed on the pci-e connectors and the card into the slot but there is no apparent looseness.
9. In case it happens again, I'm looking at something like OCZ ZS Series 650W 80PLUS

tyvm


asdfdfdfdfdf
Premium
join:2012-05-09
kudos:3
Reviews:
·Verizon Wireless..

"2. The crash triggers a reiserfsck and rebuild of dirty filesystem on next boot, at least the last two times I was watching."

Because the drive wasn't unmounted properly when the system crashed.

38 amps is not what the card requires. It is a suggested total for the machine the card is in but is only a padded estimate to lower the probability of problems.

Your supply specs label here, you have 40 amps or 480w on your 12v. I think you are worrying unnecessarily about this.

»www.newegg.com/Product/Product.a···17139001

"7. It's not hard to deduce that the power supply is marginal and fails, occasionally, when cold."

I'm not sure why you come to this conclusion and I don't think it is warranted. It could be that there is something about the power up timing of particular devices on boot up but if you had a power problem the system should either be randomly unstable or become more unstable when you are stressing the cpu and graphics card. Neither seems to be the case here. If the power supply was being stressed on boot up from powering up devices and drives then I would expect the crash to happen during boot and not a long time after.

You have tested memory. My next suspect would be the fact that you have 5 drives. Are your hard drives in a raid array or just 5 separate drives? If the latter then I would disconnect the 4 drives that aren't necessary for booting and see if the system becomes stable.

Other opinions about this?


asdfdfdfdfdf
Premium
join:2012-05-09
kudos:3
Reviews:
·Verizon Wireless..
reply to berserken

"1. Crashes always occur after cold boot and don't recur after reboot."

To really test this I would start rebooting immediately after cold boot completes and see if you are really correct about this and the problem never crops up if you do this.



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast
reply to asdfdfdfdfdf

said by asdfdfdfdfdf:

"2. The crash triggers a reiserfsck and rebuild of dirty filesystem on next boot, at least the last two times I was watching."

Because the drive wasn't unmounted properly when the system crashed.

Right. I know about reiserfsck, a bit, and offered this as a metric of the nature of the crash, i.e. X can crash w/o leaving an unclean filesystem.

38 amps is not what the card requires. It is a suggested total for the machine the card is in but is only a padded estimate to lower the probability of problems.

Your supply specs label here, you have 40 amps or 480w on your 12v. I think you are worrying unnecessarily about this.

»www.newegg.com/Product/Product.a···17139001

"7. It's not hard to deduce that the power supply is marginal and fails, occasionally, when cold."

I'm not sure why you come to this conclusion and I don't think it is warranted. It could be that there is something about the power up timing of particular devices on boot up but if you had a power problem the system should either be randomly unstable or become more unstable when you are stressing the cpu and graphics card. Neither seems to be the case here. If the power supply was being stressed on boot up from powering up devices and drives then I would expect the crash to happen during boot and not a long time after.

I was not trying to imply the deduction was, necessarily, an accurate one. I haven't seen a crash except soon after a cold boot and don't think I said anything about "a long time after."

I have been reading a lot of related resources that refer to the limitation of multiple rails including, possibly, over-taxing the per-rail overcurrent rating. Some might be based on inaccurate deductions but others, such as the following, seem well-reasoned:

»www.youtube.com/watch?v=4EtNTjrX8Ug


Such sorts of reasoning, taken together with the results of my system entered into the eXtreme Power Supply Calculator Lite,




are what make me say something like, "It's not hard to deduce that the power supply is marginal and fails, occasionally, when cold."

You have tested memory. My next suspect would be the fact that you have 5 drives. Are your hard drives in a raid array or just 5 separate drives? If the latter then I would disconnect the 4 drives that aren't necessary for booting and see if the system becomes stable.

Other opinions about this?

With about 11 days between occurrences, I'm not too motivated to do multiple reboots/unplug drives.

asdfdfdfdfdf
Premium
join:2012-05-09
kudos:3
Reviews:
·Verizon Wireless..

"With about 11 days between occurrences, I'm not too motivated to do multiple reboots/unplug drives."

I understand, intermittent problems can be tedious to troubleshoot. Unfortunately you either have to do these things or live with the problem.

"don't think I said anything about "a long time after."

What I meant was that the system has booted completely and is being used. If it was a power draw on powering up components issue then it should show up well before this during boot.

I don't know what to tell you. If you really feel sure that the power supply is the problem and want to replace it then don't let me talk you out of it. You should do what you feel comfortable doing.



berserken

join:2011-03-27
Oakland, CA
kudos:1

Still mulling. Thanks for your advice.



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast

Update: After about an hour of down time, boot up crashed to reboot from the login screen this a.m.

Yesterday, after considering the TDP of the 470 gtx, maybe 220W, compared to something like 50W for the 9500 gt it replaced, and that I can't find a significant benefit to running the 470 when I don't play games, I bought a gt 440 (56-81W).


asdfdfdfdfdf
Premium
join:2012-05-09
kudos:3

You mean it is behaving the same way with your new 440gt?



berserken

join:2011-03-27
Oakland, CA
kudos:1

No, @ bought that item on ebay, don't have it, yet.



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast
reply to berserken

The GT 440 didn't give me any such problems for about 2 weeks using it but it started to bug me that the GTX 470 was just sitting on the shelf.

So, I got a Cooler Master Silent Pro M600 and have been using that with the GTX 470 for 2.5 weeks, no problems that I can tell.


asdfdfdfdfdf
Premium
join:2012-05-09
kudos:3
reply to berserken

Thanks for the update.
I'm sorry that I was so completely wrong.



berserken

join:2011-03-27
Oakland, CA
kudos:1

Not at all. It *seems* as if it is a power issue, here. Conveying all the observations that go into a hunch at this end does not ever get the whole picture across to the other end. Thanks for giving your time.



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast

Square one

I am, now, more inclined to think it is some engineering defect with the gtx 470 rather than a power issue. There was no such crash since my last post until I added some hardware,

•a Deskstar 7K2000, since I'm a hoarder, suspected the 600W Cooler Master was, still, a bit marginal for all the parts in the box, wanted to combine the contents of two sata drives to this one,

•an OCZ Vertex Plus, for fun,

•a Syba SD-SA2PEX-2IR PCIe SATA II 2x Ports RAID Controller Card, since two burners and seven hd's were using all but one of the ports on this P5Q Deluxe and come to find out the open port on the Marvell Raid controller wouldn't work for me as non-raid. It seems to be limited to a single disk or the built-in Drive Xpert raid software.

I have merged two drives into the Hitachi, lessening the SATA HD power load by one, approximately, and boot off the SSD. Since this change, I am getting the black screen or some corrupted colored squares soon after some cold boots, pretty much like before.

Digging deeper at google, I see quite a few reporters having this problem with this card. Some have been solved by downgrading the nvidia proprietary driver, some by removing a Realtek audio driver, some by RMA but others where RMA doesn't work. Some report reverting to a slightly older 460 makes the problem disappear. IIRC. So, I tend to believe there might be various hardware conflict problems with this particular card. At this point.



norwegian
Premium
join:2005-02-15
Outback
reply to berserken

Re: EVGA GTX 470 with Corsair HX520W PSU, is ok?

Sorry if I've missed this already - have you reviewed event viewer logs or relevant memory dump, crash report files relevant to the O/S'.

Also is it happening after a certain O/S is rebooted? I know the crashes are minimal to may checking harder for this though. If it is graphics related then maybe it is a driver conflict to 1 specific O/S you have installed? Even though it seems relative to the 470, there may be just a new driver conflict that is young and the 470 is highlighting it not causing it.

I take it all Nvidia drivers have been fully uninstalled, cleaned and reloaded from a fresh installer in the battery of tests?
--
The only thing necessary for the triumph of evil is for good men to do nothing - Edmund Burke



berserken

join:2011-03-27
Oakland, CA
kudos:1
Reviews:
·Comcast

It is random and can be quite occasional. Hasn't happened since I last posted and it was some months between crashes before that, until I started jumbling around the hardware.

I have a multi-boot box, using Mandriva Linux x86_64 99% of the time with nvidia proprietary driver 295.40 and 32-bit XP-Pro SP3 up-to-date used very infrequently. Now that you mention it, this might have occurred on a boot to XP but I don't recall. I don't cold-boot to XP, only reboots from Linux, and, AFAICR, this only happens on cold boot.

Google gtx 470 random reboot and there seem to be enough reports to implicate something about the card, itself. IDK but re-installing drivers or driver cleaners does not seem to be so much the way to solve problems in Linux. I can imagine nVidia writing the same flaw cross-platform to Windows and Linux drivers but I really don't know enough to draw conclusions.

ATM, not happening for two months, it's more like an annoyance and curiosity, unless it gets more frequent. There might be other elements already written to this thread. Looking at the X log is a good idea and I'll try to remember to do that if/when it happens again. Maybe the card has settled down after the recent musical chairs of parts. Thanks.



norwegian
Premium
join:2005-02-15
Outback

1 recommendation

No worries, just thought it a possibility. But if the system is lasting for a few months, you would think a driver conflict would show more regular than that.