On Thu, 15 Jul 2004, Stephen M. Jones wrote:

> Michael writes:
>
> > Hmmm.  I just started running some tests between my CS20s (one running
> > 1.6.1 and the other running 1.6ZC - both SMP), and I am seeing 10-11MB/sec
> > transfers.  However, I'm not seeing any fxp timeouts - which would
> > significantly affect the throughput.

  I did just get one transfer that was only 860KB/sec, but the following 9
tests ran at 11MB/sec.  I'm not sure what was going on for that one
transfer.

> Are you using iothreads or anything but the defaults on your mounts?  What

  I'm using the defaults.

> sort of etherswitch are you using or are you using a cross over cable between

  I'm using a crossover cable between the CS20s.

> the two?  How often do you experience hangs with 1.6x using SMP?  I've been
> sort of directed to work with 2.0x from now on, but it seems that there
> might be some fixes in 1.6x that aren't getting or haven't been getting
> merged in 2.0x

  I've not seen any hangs with 1.6.1 or 1.6ZC (but I'm running with the
patch from the PR I submitted on the SMP hangs).  I doubt very much there
are any "fixes" in 1.6x not being merged into 2.0, since anything going
into 1.6.x branch is supposed to have come from the HEAD of the tree,
and the 2.0 branch was exactly the 1.6Z* at the time of the 2.0 branch.
There may be changes in -current that haven't made it into the 2.0
branch (I found one a few days ago and had it pulled up to the 2.0
branch).

> implemented on the primary ethernet interface.  Does pressing the switch
> on the front end of the CS20 reset it or do you need to actually use the
> back switch?  I might have to look into wiring up the front end reset
> switches on mine just to save the power cycle ;-)  I'm running out of
> spare power supplies .. thats for sure.

  The front switch is the HALT switch, and halts the system back to SRM.
I've considered attempting to wire mine into something I can control
remotely - although I'm somewaht closer to my machines.  [It's only a 10
minute walk from home :-)].  Once the machine has been halted, I can
usually CONTINUE and get back into DDB to poke around in the machine,
although sometimes I can't dump or halt the machine from DDB.  I think
I've fixed one of the reasons I've been unable to halt the machine from
DDB (the primary CPU is paused if the other CPU is running DDB, and
a halt from DDB just halted the second CPU, leaving the primary CPU
spinning.  That problem should now be fixed.

  I've been learning the use RCM [or RMC] on some other machines - that's
very nice to have!

--
Michael L. Hitch			mhitch@montana.edu
Computer Consultant
Information Technology Center
Montana State University	Bozeman, MT	USA