SGI: hinv

hinv -vm


gfxinfo -vv


uname -aR


scsicontrol -i /dev/scsi/*


/sbin/diskpatch -v


l1cmd serial all


l1cmd flash status


l1cmd env


l1cmd pci


Notes:
- The S2iO 10Gb Ethernet card is running at pretty high temperature
- It appears that opening the left side of the case (the one with VPro & PSU) lowers the temperature, while opening the right side (with CPUs and RAM) rises it and the machine shuts down.

Very nice, congratulations!

devv wrote: Notes:
- The S2iO 10Gb Ethernet card is running at pretty high temperature
- It appears that opening the left side of the case (the one with VPro & PSU) lowers the temperature, while opening the right side (with CPUs and RAM) rises it and the machine shuts down.

I bucked the less-fan-noise trend and added fans to mine .

Hehe, sure, I've been watching that thread. I'll sure try doing that as well; the point I'm kind of stuck at at the moment is locating the pins to which I should connect the fan I'd add above the VPro.
Actually I just found it -- as you said "SGI included an <unused> three-pin fan header on the Tezro version of the V12 that isn't present on the Fuel version of the V12.".
I'll try to follow what you did there and fill in with any other info if I find something new/untried yet or some good fans. Thanks!

Actually, recondas -- where is that 3-pin fan header? I've opened the cover of the VPro and I only see two connectors of which none seem to be it.
There is a 2-pin header between the two heatsinks, and there is a 6-pin header deeper above the larger heatsink, close to the PCI-X attachment.
I don't see anything with 3 pins.

Also, is the header a standard 3-pin fan or I need to change the order of wires as shown in viewtopic.php?f=3&t=9046 ?

I've played a bit with determining the effect of additional fans on the temperature.

In the tests, I've been measuring:
- With and without the left panel (the one on the VPro side) being present
- With and without the VPro cover being present
- With and without additional fan present (used Noctua NF-P14, running at 1200 RPM: http://noctua.at/en/nf-p14-flx/specification )

Overall, it appears that the largest positive influence on the temperature has the removal of the left panel.
Without it installed, it is possible to achieve the lowest temperature of the VPro, as well as the fact that it also reduces the temperature of BEDROCK by about 5 C and of NODE by 2-3 C without me doing anything to them.

So, temperature of "4 ODYSSEY" in various combinations (in order of best first):

- Without panel, without cover, fan on top (perpendicular to case air flow): 32 C (image of this setup it attached)
- Without panel, without cover, fan parallel (in the direction of air flow): 34 C
- With panel, with cover, fan on top: 36-37 C
- With panel, without cover, fan parallel: 36-37 C
- With panel, without cover, fan on top: 36-37 C
- Without panel, with cover, no fan: 39 C
- With panel, with cover, no fan: 42+ C (this is standard setup without modification)

Notes:
- I did not examine the effects if the front panel is removed rather than the left panel.
- I did confirm that removing the right panel (one on the CPUs/RAM side) quickly increases temperature of BEDROCK, and then initiates shutdown. (So as mentioned, removing the left panel is beneficial, while removing the right is not).
- The image of the most effective variant found (reaching down to 32 C) is attached.

Additional mini-chapter:
- I did not find the 3-pin fan connector on the VPro. Instead, I've plugged a dedicated fan control box in between the PSU 12V power and the DVD drive. (I happened to have this fan controller handy: http://sun-trust.net/products/scythe/1068.html ).
- I actually like this approach best and it does not require modifying the fans since they're connected to a compatible controller.
- The only drawback is that if one wants to replace Tezro's DVD drive with a 1x5.25 -> 4.x2.5 SAS/SATA drive enclosure (such as Icy Dock MB994SP-4SB-1 or Logilink UA0208), then there may not be enough depth available to mount both the drives enclosure in the back and the fan control in the front of the 5.25 slot. (In the attached image, it can be seen that there is only little unused depth remaining between the end of the DVD and start of the PSU, although it might be just enough - didn't try yet!).

Dude, where's the (plastic) skin(s)?

I've taken it off, my mini-goal here was to measure temperatures, and without the left panel (skin) I was able to produce the lowest temp.
Will post some more pics of it tomorrow, in a more proper dressing

Nice! Congratulations.

Outstanding machine old chap. Does that peak SGI Workstations?

I wish Still hunting for 4x1Ghz instead of 4x800Mhz. But I have a feeling that'll be postponed for a couple years, given the prices of 4x1Ghz boards

What's 200MHz(x4) among friends

Last time I checked, it was about 1500 EUR

Less than a night with a high class prostitute.. but lasts forever I say!

devv wrote: Actually, recondas -- where is that 3-pin fan header?

The fan header is located between the small and large heat sinks . The part number for that dead-on-arrival / fan-header-included V12 is 030-1884-002, revision D.

Sorry for the confusion - the initial post in that thread (with the hinv and other data) was updated over time as I made changes to the configuration. What you may not notice because of the updates to the hinv is when I acquired that Tezro the original V12 was dead (also doesn't help that the underlying story about the acquisition/resurrection of that Tezro is located in another thread).

So while the original V12 did indeed have a three-pin fan header, I installed and ran a Fuel V12 (without an integrated fan header) for a year and a half. So I powered the supplemental fans in much the same way as you did; with a molex power splitter :

in his Tezro hinv post, recondas wrote: I didn't have any 60mm fans <nor did any of the sources local to me>, but I did have two 38mm fans pre-wired in a two-fan array with a molex-splitter power attachment. For initial testing I zip-tied the 38mm fans to the perforated metal shielding, directly in front and blowing onto the perforations in the node board cooling shroud, and made an air duct so that air was directed into the openings in the node board shroud <rather than the dead air space above the shroud. The molex splitter was connected at the <molex> power feed for the hard drive backplane <located directly above the node board>. The 38mm fans are only rated at 4.5CFM each <I plan to order three 12CFM-each low-DBA 60mm fans as replacements>, but made a noticeable 6C/11F drop in the operating temperature of the bedrock a ASIC.

In the time since that was written I've replaced the 38mm fans by the node board cooling shroud with two 60mm fans ( 1GHz node boards come from the factory with cooling fans located in a similar position ). I still have the 80mm fan mounted on top of the perforated V12 cooling shroud, and I moved the two 38mm fans mentioned above to exhaust air through the perforated bulkhead at the back of the PCI cage. All are still powered via molex splitters.

Well that's interesting, my V12 has a 2-pin header in place of the 3-pin fan header. I will check on the part number etc.
Great, thanks for the tips.