Folks, I'd like to call for a commit freeze on trunk for the 1.5rc2 release. The only allowed changes at this point are bug fixes for priority bugs. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji
Ok, sorry I lied. I just committed some IBM contributed fixes to trunk. But freeze from here on forward :-). -- Pavan On 09/25/2012 11:20 AM, Pavan Balaji wrote:
Folks,
I'd like to call for a commit freeze on trunk for the 1.5rc2 release. The only allowed changes at this point are bug fixes for priority bugs.
-- Pavan
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
Ok. You should include my commit from earlier today. It fixes an oversight needed to round out the new RMA ops. ~Jim. On 9/25/12 2:06 PM, Pavan Balaji wrote:
Ok, sorry I lied. I just committed some IBM contributed fixes to trunk.
But freeze from here on forward :-).
-- Pavan
On 09/25/2012 11:20 AM, Pavan Balaji wrote:
Folks,
I'd like to call for a commit freeze on trunk for the 1.5rc2 release. The only allowed changes at this point are bug fixes for priority bugs.
-- Pavan
We are not forking out a new branch yet, so everything that's committed into trunk will be a part of 1.5. I'm just waiting for Dave's ticket to get fixed. -- Pavan On 09/25/2012 04:57 PM, Jim Dinan wrote:
Ok. You should include my commit from earlier today. It fixes an oversight needed to round out the new RMA ops.
~Jim.
On 9/25/12 2:06 PM, Pavan Balaji wrote:
Ok, sorry I lied. I just committed some IBM contributed fixes to trunk.
But freeze from here on forward :-).
-- Pavan
On 09/25/2012 11:20 AM, Pavan Balaji wrote:
Folks,
I'd like to call for a commit freeze on trunk for the 1.5rc2 release. The only allowed changes at this point are bug fixes for priority bugs.
-- Pavan
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
The simple union fix that I discussed with Darius, that we were fairly sure would work… didn't. :( I'm still hacking on it, but it's not clear what the lowest risk yet still effective fix will be. -Dave On Sep 25, 2012, at 11:58 PM GMT+02:00, Pavan Balaji wrote:
We are not forking out a new branch yet, so everything that's committed into trunk will be a part of 1.5. I'm just waiting for Dave's ticket to get fixed.
-- Pavan
On 09/25/2012 04:57 PM, Jim Dinan wrote:
Ok. You should include my commit from earlier today. It fixes an oversight needed to round out the new RMA ops.
~Jim.
On 9/25/12 2:06 PM, Pavan Balaji wrote:
Ok, sorry I lied. I just committed some IBM contributed fixes to trunk.
But freeze from here on forward :-).
-- Pavan
On 09/25/2012 11:20 AM, Pavan Balaji wrote:
Folks,
I'd like to call for a commit freeze on trunk for the 1.5rc2 release. The only allowed changes at this point are bug fixes for priority bugs.
-- Pavan
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
We should also do a performance sanity check before the release (shared memory latency/bandwidth). Rajeev
Darius, Can you try out the performance to make sure everything is alright? The nightly tests seem to be OK now. Some of the special tests are failing. 1. The build is failing with rlog. I'll look into this. 2. Some tests with ASYNC_PROGRESS are failing. - The Comm_idup tests are expected to fail, as we don't support idup for more than THREAD_SERIALIZED yet (though the error message says greater than THREAD_SINGLE is not supported; this needs to be fixed). - The manyrma2 test is failing. This is expected. - The errinstatwa test is failing. This is not expected. Can someone look into this? -- Pavan On 09/26/2012 02:47 AM, Rajeev Thakur wrote:
We should also do a performance sanity check before the release (shared memory latency/bandwidth).
Rajeev
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On 09/26/2012 08:33 PM, Pavan Balaji wrote:
1. The build is failing with rlog. I'll look into this.
rlog doesn't quite seem to have been ported to the new build system. I'm not entirely sure, but shouldn't the configure.ac be converted to subconfigure.m4 for this to work correctly? Also, several RMA tests to be failing in the special tests as well. I didn't notice that the first time around. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji
On 09/26/2012 09:49 PM, Pavan Balaji wrote:
On 09/26/2012 08:33 PM, Pavan Balaji wrote:
1. The build is failing with rlog. I'll look into this.
rlog doesn't quite seem to have been ported to the new build system. I'm not entirely sure, but shouldn't the configure.ac be converted to subconfigure.m4 for this to work correctly?
Never mind. There's already a ticket for this. https://trac.mcs.anl.gov/projects/mpich2/ticket/1604 -- Pavan Balaji http://www.mcs.anl.gov/~balaji
On 9/26/12 8:33 PM, Pavan Balaji wrote:
2. Some tests with ASYNC_PROGRESS are failing.
We have a ticket for this: http://trac.mcs.anl.gov/projects/mpich2/ticket/1676 I looked into some of the tests and they weren't regressions from 1.4. I can dig deeper if we want a fix in 1.5. ~Jim.
The ticket only seems to talk about one test. There were more tests than that failing, IIRC. Also, some other RMA tests were failing for other configurations as well. -- Pavan Balaji http://www.mcs.anl.gov/~balaji On Sep 27, 2012, at 1:11 AM, Jim Dinan <[email protected]> wrote:
On 9/26/12 8:33 PM, Pavan Balaji wrote:
2. Some tests with ASYNC_PROGRESS are failing.
We have a ticket for this:
http://trac.mcs.anl.gov/projects/mpich2/ticket/1676
I looked into some of the tests and they weren't regressions from 1.4. I can dig deeper if we want a fix in 1.5.
~Jim.
Rajeev, The special tests doesn't seem to have completed yet. Is this normal? -- Pavan Balaji http://www.mcs.anl.gov/~balaji On Sep 27, 2012, at 6:33 AM, Pavan Balaji <[email protected]> wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
Also, some other RMA tests were failing for other configurations as well.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 27, 2012, at 1:11 AM, Jim Dinan <[email protected]> wrote:
On 9/26/12 8:33 PM, Pavan Balaji wrote:
2. Some tests with ASYNC_PROGRESS are failing.
We have a ticket for this:
http://trac.mcs.anl.gov/projects/mpich2/ticket/1676
I looked into some of the tests and they weren't regressions from 1.4. I can dig deeper if we want a fix in 1.5.
~Jim.
They start at 9am. I have started them by hand today. On Sep 27, 2012, at 6:36 AM, Pavan Balaji wrote:
Rajeev,
The special tests doesn't seem to have completed yet. Is this normal?
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 27, 2012, at 6:33 AM, Pavan Balaji <[email protected]> wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
Also, some other RMA tests were failing for other configurations as well.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 27, 2012, at 1:11 AM, Jim Dinan <[email protected]> wrote:
On 9/26/12 8:33 PM, Pavan Balaji wrote:
2. Some tests with ASYNC_PROGRESS are failing.
We have a ticket for this:
http://trac.mcs.anl.gov/projects/mpich2/ticket/1676
I looked into some of the tests and they weren't regressions from 1.4. I can dig deeper if we want a fix in 1.5.
~Jim.
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently. ~Jim.
I usually do, but before I got to it, the tests disappeared. -- Pavan Balaji http://www.mcs.anl.gov/~balaji On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
5 days of history is maintained directly, and another 10 days is compressed. You can edit the date field of the URL to see the previous day's run. Rajeev On Sep 28, 2012, at 9:56 AM, Pavan Balaji wrote:
I usually do, but before I got to it, the tests disappeared.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
The tests are all dead. Something is terribly wrong here. On 09/28/2012 10:05 AM, Rajeev Thakur wrote:
The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
5 days of history is maintained directly, and another 10 days is compressed. You can edit the date field of the URL to see the previous day's run.
Rajeev
On Sep 28, 2012, at 9:56 AM, Pavan Balaji wrote:
I usually do, but before I got to it, the tests disappeared.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
Not the nightly. The special tests have a failure message because two got started simultaneously, and the second one failed. On Sep 28, 2012, at 10:20 AM, Pavan Balaji wrote:
The tests are all dead. Something is terribly wrong here.
On 09/28/2012 10:05 AM, Rajeev Thakur wrote:
The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
5 days of history is maintained directly, and another 10 days is compressed. You can edit the date field of the URL to see the previous day's run.
Rajeev
On Sep 28, 2012, at 9:56 AM, Pavan Balaji wrote:
I usually do, but before I got to it, the tests disappeared.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
Yes, I meant all the special tests, since that's what we were discussing here. On 09/28/2012 10:29 AM, Rajeev Thakur wrote:
Not the nightly. The special tests have a failure message because two got started simultaneously, and the second one failed.
On Sep 28, 2012, at 10:20 AM, Pavan Balaji wrote:
The tests are all dead. Something is terribly wrong here.
On 09/28/2012 10:05 AM, Rajeev Thakur wrote:
The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
5 days of history is maintained directly, and another 10 days is compressed. You can edit the date field of the URL to see the previous day's run.
Rajeev
On Sep 28, 2012, at 9:56 AM, Pavan Balaji wrote:
I usually do, but before I got to it, the tests disappeared.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
OK. The history I was referring to is in the nightly tests, not special tests. On Sep 28, 2012, at 10:37 AM, Pavan Balaji wrote:
Yes, I meant all the special tests, since that's what we were discussing here.
On 09/28/2012 10:29 AM, Rajeev Thakur wrote:
Not the nightly. The special tests have a failure message because two got started simultaneously, and the second one failed.
On Sep 28, 2012, at 10:20 AM, Pavan Balaji wrote:
The tests are all dead. Something is terribly wrong here.
On 09/28/2012 10:05 AM, Rajeev Thakur wrote:
The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
5 days of history is maintained directly, and another 10 days is compressed. You can edit the date field of the URL to see the previous day's run.
Rajeev
On Sep 28, 2012, at 9:56 AM, Pavan Balaji wrote:
I usually do, but before I got to it, the tests disappeared.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
Ah, OK. Sorry, I just gave wrong information to Jim then. On 09/28/2012 10:38 AM, Rajeev Thakur wrote:
OK. The history I was referring to is in the nightly tests, not special tests.
On Sep 28, 2012, at 10:37 AM, Pavan Balaji wrote:
Yes, I meant all the special tests, since that's what we were discussing here.
On 09/28/2012 10:29 AM, Rajeev Thakur wrote:
Not the nightly. The special tests have a failure message because two got started simultaneously, and the second one failed.
On Sep 28, 2012, at 10:20 AM, Pavan Balaji wrote:
The tests are all dead. Something is terribly wrong here.
On 09/28/2012 10:05 AM, Rajeev Thakur wrote:
The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
5 days of history is maintained directly, and another 10 days is compressed. You can edit the date field of the URL to see the previous day's run.
Rajeev
On Sep 28, 2012, at 9:56 AM, Pavan Balaji wrote:
I usually do, but before I got to it, the tests disappeared.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
It's ok, I'm used to it. ;) On 9/28/12 10:39 AM, Pavan Balaji wrote:
Ah, OK. Sorry, I just gave wrong information to Jim then.
On 09/28/2012 10:38 AM, Rajeev Thakur wrote:
OK. The history I was referring to is in the nightly tests, not special tests.
On Sep 28, 2012, at 10:37 AM, Pavan Balaji wrote:
Yes, I meant all the special tests, since that's what we were discussing here.
On 09/28/2012 10:29 AM, Rajeev Thakur wrote:
Not the nightly. The special tests have a failure message because two got started simultaneously, and the second one failed.
On Sep 28, 2012, at 10:20 AM, Pavan Balaji wrote:
The tests are all dead. Something is terribly wrong here.
On 09/28/2012 10:05 AM, Rajeev Thakur wrote:
The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
5 days of history is maintained directly, and another 10 days is compressed. You can edit the date field of the URL to see the previous day's run.
Rajeev
On Sep 28, 2012, at 9:56 AM, Pavan Balaji wrote:
I usually do, but before I got to it, the tests disappeared.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
Pavan and I located the ticket for the other test that is failing, lockopts: https://trac.mcs.anl.gov/projects/mpich2/ticket/1683 ~Jim. On 9/28/12 10:42 AM, Jim Dinan wrote:
It's ok, I'm used to it. ;)
On 9/28/12 10:39 AM, Pavan Balaji wrote:
Ah, OK. Sorry, I just gave wrong information to Jim then.
On 09/28/2012 10:38 AM, Rajeev Thakur wrote:
OK. The history I was referring to is in the nightly tests, not special tests.
On Sep 28, 2012, at 10:37 AM, Pavan Balaji wrote:
Yes, I meant all the special tests, since that's what we were discussing here.
On 09/28/2012 10:29 AM, Rajeev Thakur wrote:
Not the nightly. The special tests have a failure message because two got started simultaneously, and the second one failed.
On Sep 28, 2012, at 10:20 AM, Pavan Balaji wrote:
The tests are all dead. Something is terribly wrong here.
On 09/28/2012 10:05 AM, Rajeev Thakur wrote:
The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
5 days of history is maintained directly, and another 10 days is compressed. You can edit the date field of the URL to see the previous day's run.
Rajeev
On Sep 28, 2012, at 9:56 AM, Pavan Balaji wrote:
I usually do, but before I got to it, the tests disappeared.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Sep 28, 2012, at 8:14 AM, Jim Dinan <[email protected]> wrote:
On 09/27/2012 06:33 AM, Pavan Balaji wrote:
The ticket only seems to talk about one test. There were more tests than that failing, IIRC.
I do remember another test where we saw a failure, but unfortunately that info was lost when the test suite was re-run. My guess is that there is a concurrency issue somewhere in the RMA code, but that it's not a regression from 1.4. We made a conscious decision to focus on MPI-3 implementation right now and fix this when there's time (or fix it as a natural part of MPI-3 development).
Also, some other RMA tests were failing for other configurations as well.
If you see a failure, please capture as much info as you can in a ticket. The problem is that every time the tests run, we lose info from the last run and the RMA async progress failures tend to happen pretty infrequently.
~Jim.
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
I ran the perf/ tests on bb70 and nestvec, nestvec2 and indexperf failed: nestvec: MPI_Pack time = 2.288818e-04, manual pack time = 1.192093e-05 nestvec2: MPI_Pack time using struct with vector = 2.240896e-03, manual pack time = 1.938343e-04 nestvec2: MPI_Pack using a vector = 2.880096e-04, using a struct with vector = 2.240896e-03 indexperf: MPI_Pack (block index) time = 2.098083e-05, manual pack time = 7.867813e-06 indexperf: MPI_Pack (struct of block index)) time = 2.193451e-05, manual pack time = 7.867813e-06 I got 172 ns 0-byte latency and 6.5 GB/s bandwidth So I think we're good. -d On Sep 26, 2012, at 8:33 PM, Pavan Balaji wrote:
Darius, Can you try out the performance to make sure everything is alright?
The nightly tests seem to be OK now.
Some of the special tests are failing.
1. The build is failing with rlog. I'll look into this.
2. Some tests with ASYNC_PROGRESS are failing.
- The Comm_idup tests are expected to fail, as we don't support idup for more than THREAD_SERIALIZED yet (though the error message says greater than THREAD_SINGLE is not supported; this needs to be fixed).
- The manyrma2 test is failing. This is expected.
- The errinstatwa test is failing. This is not expected. Can someone look into this?
-- Pavan
On 09/26/2012 02:47 AM, Rajeev Thakur wrote:
We should also do a performance sanity check before the release (shared memory latency/bandwidth).
Rajeev
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
participants (5)
-
Darius Buntinas -
Dave Goodell -
Jim Dinan -
Pavan Balaji -
Rajeev Thakur