>I indeed have a bit of practical experience to share on O2s and RAM.
>We have Bruker spectrometers with O2's and when we received them, I rapidly
>planned a memory upgrade in order to speed work on these boxes. Considering
>the O2's peculiar internal architecture, it seemed - and it still does seems
>so today - that 64Mb is rather unefficient. So I upgraded my O2s to more
>than 256Mb. Then problems started, the spectrometers kept crashing for no
>apparent reasons and not totally reproducibly.
Sounds familiar. I found out that SGI didn't sell any 64 Mbyte simms.
They could not explain why, but apparently it is for good reasons. The
3rd party 64 Mbyte simms I have on my desk are from Camintonn.
>Finally, it turned out that there is a *bug* in Irix 6.3 that makes the O2
>crash in the situation where the following three conditions are
>simultaneously fulfilled:
>
>(1) there are two Ethernet cards active
>(2) there are more than 256Mb of RAM
>(3) the RAMs are from different origins (vendor)
Here the two last conditions are fullfilled.
>The patches installed for allowing Irix 6.3 to handle more than 256Mb do not
>seem to matter, read they all fail. SGI does not seem to care either. Their
>first reaction was: this extra RAM is not SGIs', so none of our business.
Agrees with our experience.
>A small point to note that the sensitivity to different RAMs is a known
>weakness from some UNIX boxes, but I found it pretty ironic that SGI is
>buying its RAMs from the same vendor I got my extra ones, tests them
>somewhat, throws away a few (many ?) and sells the resulting ones for ~5
>times the original price. And does not even support their (co)existence
>properly.
The SGI price is a factor 5 more expensive which they were somewhat
embarrassed to confirm. I will dump the SGI simms and go for the 3rd
party. Great many thanks for your comments.
73, Peter
@ @
..
==================================================ooOO==========OOoo======
| Peter Lundberg Ph: (+46 13) 22 27 90 |
| MR-Unit Fax: (+46 13) 22 47 49; 22 27 92 |
| Dept of Radiation Physics Email: PeterL@mr.us.lio.se |
| University of Linkoping |
| S-581 85 Linkoping Efax: PeterLundberg@f013224749.fax.sunet.se |
| Sweden www: http://www.liu.se/hu/imv/radiofysik |
==========================================================================
() "To a poet nothing is useless." (SJ) ooOO OOoo