Commit 80ec4234 authored by Jonathan Jenkins's avatar Jonathan Jenkins

rm dead files

parent fb0d408a
......@@ -20,7 +20,7 @@ pkgconfig_DATA = maint/codes.pc
$(pkgconfig_DATA): config.status
EXTRA_DIST += \
prepare.sh COPYRIGHT TODO.txt configure.ac uc-codes.cfg reformat.sh \
prepare.sh COPYRIGHT configure.ac uc-codes.cfg reformat.sh \
misc/README misc/ptrn_loggp-2.4.6.patch CONTRIBUTORS.md
AM_CPPFLAGS = -I$(top_srcdir)/src ${ROSS_CFLAGS}
......
1- Torus network model:
- Check how to specify the buffer size in the config file (buffer size is in flits)--- Option 1 specify the buffer size in packets and convert it into flits?
- Check how to specify the number of torus nodes from the config file? Currently they are being calculated from the file... How to integrate this with the mapping thing?
Option: Specify the torus network and the length of each torus dimension.
2- Tests
- clean up modelnet-test.c and code based on it. The way it calculates lpids is
extremely fragile (a bunch of constants + some assumptions that lead to very
confusing code)
......@@ -12,7 +12,6 @@ EXTRA_DIST += src/iokernellang/codesparser.y.in \
src/util/README.codes-mapping.txt \
src/util/README.lp-io.txt \
src/workload/README.codes-workload \
src/workload/TODO.codes-workload \
src/network-workloads/README_synthetic.txt \
src/network-workloads/README_traces.txt \
src/network-workloads/workloads.conf \
......@@ -30,8 +29,7 @@ EXTRA_DIST += src/iokernellang/codesparser.y.in \
src/networks/model-net/doc/README.loggp.txt \
src/networks/model-net/doc/README.simplenet.txt \
src/networks/model-net/doc/README.simplep2p.txt \
src/networks/model-net/doc/README.torus.txt \
src/networks/model-net/MN_TODO
src/networks/model-net/doc/README.torus.txt
#src/network-workload/README
#src/network-workload/offsets
......
1- Torus model integration, possible additions:
- Allow some compute nodes to be idle (not mapped to a server) i.e. we can specify any number of server and torus nodes that we want in the config file.
The torus configuration parameters should set an upper bound on the number of servers/compute nodes.
- Change the buffer overflow message? May be allow a slowed down packet injection rate if the buffer overflows and report the average injection rate at the
end of the simulation.
- Generalize the notation used to identify entities in the application (they
are referred to by ranks right now). See email from Dries 9/18/2013.
- Implement some "real" workload generator method examples.
- Plug workload API into storage simulations.
- Implement a trace replay tool as an MPI program that uses the same API so
that we can replay workloads on real systems in addition to simulation.
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment