Attached are the new results with the improved TCP BTL. It really look a
lot better.
Thanks,
george.
On Tue, 29 Nov 2005, Tim S. Woodall wrote:
Due to the structure of the pml/btl interface, the btl code does an
extra recv call. The cost of this varies, on odin it appears to be
closer
On Nov 29, 2005, at 7:08 PM, George Bosilca wrote:
I found something strange in iof and/or orte. If we run an
application and it does not finish correctly (one of the processes
segfault) or it leaks some orted on the nodes all subsequent MPI runs
will not have any output. The output just disappe
You probably already know this, but this problem is still present in
the trunk. It appears to be fixed in the 1.0.x tree.
Greg
No, I was not aware of this -- I migrated all the changes from the
trunk to the v1.0 branch (not the other way around).
What kind of systems are you running into this on?
On Nov 30, 2005, at 10:37 AM, Greg Watson wrote:
You probably already know this, but this problem is still present in
the
Fedora Core 4 on x86. I installed the overnight snapshot from trunk
first and immediately got the error. Then I tried 1.0.x and it
worked. Want debugging info?
Greg
On 30/11/2005, at 10:14 AM, Jeff Squyres wrote:
No, I was not aware of this -- I migrated all the changes from the
trunk to t
On Nov 30, 2005, at 2:12 PM, Greg Watson wrote:
Fedora Core 4 on x86. I installed the overnight snapshot from trunk
first and immediately got the error. Then I tried 1.0.x and it
worked. Want debugging info?
Blah!
Yes, please send any debugging info that you have -- SVN r number and a
backtr
