Anyone have a spare 850-1,000w power supply I can borrow?

[rekabis]

I am looking to debug some very strange behaviour in my “daily driver” that has sent me up the wall over the last two months.

The machine has the following specs:

• SuperMicro X5DAE
• 2 × Gallatin-class Xeon 3.5GHz 32-bit processors
• 12GB ECC REG DDR-400 PC-3200 RAM
• ATI Radeon HD 4650 AGP video card
• 128GB Seagate IDE 7200RPM hard drive (primary boot)
• 4-drive hotswap backplane, containing:
  
  • 4 × 1.5TB Seagate 7200RPM 3½″ drives in a RAID-5 array (3 drives in array, 1 hotspare)
  • 2 × 500GB Seagate 7200RPM 2½″ drives in a RAID-0 array
  • All driven by a 3Ware 9550-12SX PCI-E RAID card
• HP Gigabit PCI-E network card
• PCI card with both USB and Firewire ports (both external as well as internal pin headers)
• USB hub installed in a 3¼″ drive bay, powered by the power supply
• IDE DVD-writer
• IDE Zip drive
• Sony FD-50 combo 3½″ and 5¼″ diskette drive in a single 5¼″ bay
• Thanks to a combination of a mobo that can only handle 32-bit processors (thanks to the Intel E7505 chipset) and the 12GB of RAM installed, I am forced to use Windows Server 2003 R2 Enterprise Edition.

…and this is all driven by a single, year-old 650w SSI-class power supply. Yeah, I’m thinking that it might be just a little bit underpowered for the job at hand.

Please note: all hardware has been checked multiple times and certified as being error free. The Windows installation is a fresh one, with all service packs, patches and updates. Antivirus is installed and fully functional (Symantec Endpoint Protection, pretty well the business-class industry standard).

At various times, Windows grinds to a complete halt. There is no processor activity, no drive activity, but everything just starts freezing a bit at a time, eventually requiring a complete hard reset. Other times, drives drop out of the array due to power loss/reset issues. And when I connect internal hard drives via USB (I have one of those IDE->USB dongles), if I connect them to the power supply for power issues, I can see the data transfer slow to a crawl, and data read errors spike through the roof. Yet when I connect them to another computer, the drive functions perfectly.

So I am imploring if anyone has a 850-1,000w power supply I can borrow for a week, to see if the problems vanish. I have plenty of power supplies, including new 650w ones, just nothing more powerful. A modular power supply would be nice, as I won’t have to unplug my cables from my devices just to swap the power supply, but I’ll take anything that has or can accept the dual four-pin SSI plugs (meant for server motherboards). I am hoping to confirm or deny the power supply issue before spending big bucks on a brand new high-efficiency unit.

[Bleach]

I have nothing like that. However you have more PSU's, why not daisy chain two? Try that out.

With that spec 650W is the bare basic, you could use. putting all your drive on a different PSU will take 80W off, that might tell you that a new more powerful PSU is need.

Good luck, sever class PSU are pricey.

[rekabis]

Good luck, sever class PSU are pricey.

Tell me about it. I bought four of those Mushkin 650w power supplies a little over a year and a half ago, when they were on clearance and heavily discounted, and I still paid close to $100 apiece for them. The main reason why I went for them was the dual 4-pin plugs (one of which is your traditional P4 plug) that allow an SSI board to properly function. Most motherboards have only the P4 plug, and most power supplies have only the single P4 plug. I needed a power supply that had the full pair to plug into the 8-pin SSI plug on the motherboard.

Four out of my five machines make use of full-SSI (12×13 inch) motherboards that require both server-class cases (to allow them to fit while allowing room for items in the 3½″ and 5¼″ bays) and server-class power supplies. The last? It’s a hackintosh. Basic mid-range hardware there.

[rekabis]

I have nothing like that. However you have more PSU's, why not daisy chain two? Try that out.

Not a bad idea, but I would still prefer to have a single power supply to fully rule out the current one installed. That way, I know for sure that any issues are with that particular power supply.

[Bleach]

Not a bad idea, but I would still prefer to have a single power supply to fully rule out the current one installed. That way, I know for sure that any issues are with that particular power supply.

I agree, however it's a tall order to find that PSU on this forum. You could try future shop, online they have a good corsair PSU for $200, and a very easy return, if should not work or work very shortly. It's got an EPS.

http://www.futureshop.ca/en-CA/product/corsair-corsair-1050w-pc-power-supply-cmpsu-1050hx-cmpsu-1050hx/10190694.aspx?path=66036c2d7b74aeea690bcc46c164f4e9en02

http://www.corsair.com/professional-series-hx1050-80plus-silver-certified-modular-power-supply.html

[rekabis]

Whelp, my IDE0 channel suddenly stopped responding. The system is able to see drives, but any drive attached to IDE0 is unable to act as a boot drive (the OS doesn’t boot). Methinks that power supply issues are not the only issues I have.

[DANSPEED]

Can you monitor the mobo voltages and temperatures in the OS using some app? Looking at your mobo manual it has hardware monitoring logic although viewing in BIOS at boot isn't really taxing your system.

Maybe bad mobo caps?

[Loed]

Sounds and acts exactly like a mobo issue, not a psu issue.

A quality 650w PSU is more than enough to power those components(see: quality).
Bumping up to an 850w range will give you slightly more headroom with a lot less 80% usage peaks on the psu.

That said, you definitely have some mobo issues that need rectifying.

[rekabis]

All the caps are fine, I think that this is a chipset issue. As in, the Intel E7505 chipset is slowly giving out; a sort of slow-motion crash. It started out having memory issues with more than one channel being filled at a time causing the BIOS to hang just before it switched over to loading the OS (and *after* all BIOS checks were finished), and then it went to IDE issues, and now finally the entire system is starting to act wonky.

Anyone happen to have a bead on a local SuperMicro X5DAE (or the SCSI variant) in the Okanagan? I know it’s a long shot, but the only people on eBay selling that model for under $USD300 (as in, $USD40-80 before shipping) don’t ship outside of the continental US. And they’re adamant about it.

[rekabis]

A quality 650w PSU is more than enough to power those components

AFAIK, the Mushkin PSU that I have is a quality piece. It weighs a metric assload even without cables attached, and this is typically a pretty accurate indicator of quality -- the heavier the PSU for any given wattage, the higher the quality of its components because they’re engineered to be bigger, thicker, heavier and more resilient to wearing out and degradation.

[DANSPEED]

All the caps are fine, I think that this is a chipset issue. As in, the Intel E7505 chipset is slowly giving out; a sort of slow-motion crash. It started out having memory issues with more than one channel being filled at a time causing the BIOS to hang just before it switched over to loading the OS (and *after* all BIOS checks were finished), and then it went to IDE issues, and now finally the entire system is starting to act wonky.

Have you done the Intel Processors diagnostic tool checks, the SeaTools for DOS, test on your hdd and Memtest86+ DOS a RAM test? To test your system for heating issues try running the CPU stress test and a GPU stress test like rthdribl (I can't find the link) at the same time.

[rekabis]

Have you done the Intel Processors diagnostic tool checks, the SeaTools for DOS, test on your hdd and Memtest86+ DOS a RAM test? To test your system for heating issues try running the CPU stress test and a GPU stress test like rthdribl (I can't find the link) at the same time.

Yes. Processor is fine, drive is in excellent health and the RAM is perfect. I have been checking all of those on a regular basis for the last few weeks. This appears to be a chipset issue, not anything else.

And now my problem is that anything on an IDE channel will no longer load. This includes anything in a CD-ROM drive. I would have to build a bootable USB stick in order to run any more diagnostics.

Edit: Although I do have to say that those Intel processor diagnostics do look interesting. Up until now, I’ve just been using Mersenne prime tests (and Prime95) to confirm processor stability and functionality.

[DANSPEED]

Hmmm .... if those tests pass then it sounds like the E7505 is okay. Looking at the MCH it's controlling almost everything. Your mobo has a few other chips; ICH4, AIC-7902 and I/O BRIDGE. Your BIOS PC health monitoring features looks like it's continuously checking your mobo for potential errors -- I assume you have it enabled. And I assume your using the tested RAM. Your BIOS is up to date?

Do the problems start at Windows startup (cold boot) or do they seem to get worse as the system heats up? Have you tried running a Linux live CD install (like Mint) and see if you can duplicate some of the problems -- although I doubt if Linux can handle your mobo's full potential. Maybe for a test swap your hdd for a fresh Windows install and see if the errors appear. Sometime I like to bypass Windows altogether and boot DOS or Linux for troubleshooting -- mind you I doubt both could handle your dual CPU setup and graphics card.

[DANSPEED]

Just to add. Are you running Super Micro's Supermicro's SuperDoctor® III Server Monitoring Software? It's health log might be useful in troubleshooting. Anythings better than Windows vague Event Viewer!

[Loed]

Linux would handle a dual cpu set-up out of the box better than most windows installations prior to the advent of server 2011.

Definitely try swapping in one of your other PSU's, and go through your mobo and reseat... everything. Mushkins are a quality unit, you are right.
It can be a hassle but I've had reseating solve a multitude of issues in the past, it's worth a try!

Then just do a staged boot up process(install each extraneous piece at a time, boot, test, repeat). It's worth it to save some cash if you can spare the time.