OSPRI

Settings | Report Duplicate

1

I Use This!

Inactive

Commits : Listings

Analyzed about 7 hours ago. based on code collected about 7 hours ago.

Commit Message	Contributor	Files Modified	Lines Added	Lines Removed	Code Location	Date
Aug 31, 2024 — Aug 31, 2025 Showing page 74 of 84 Search / Filter on:
adding un/likely_if when clearly useful removing some extra whitespace and formatting code with Eclipse	Jeff Hammond as jhammond	More...				about 15 years ago
added scalar lock/trylock/unlock calls	Jeff Hammond as jhammond	More...				about 15 years ago
The main thread has to call handoff_progress if it is holding the lock and waiting for completions. Enabling handoff be default (ofcourse only when CHT is enabled), does not have an impact on the regular bandwidth. Have to write a benchmark to show the benifits of the overlap and evaluate any memory contention between the main thread and cht.	potluri	More...				about 15 years ago
some API additions, small formatting changes, MPID notes updated	Jeff Hammond as jhammond	More...				about 15 years ago
adding dcmfd_cht file, missed in the last checkin	potluri	More...				about 15 years ago
adding code for handoff - currently handoff packing to cht. it is disabled by default. I will enable it after some testing	potluri	More...				about 15 years ago
adding more benchmarks	potluri	More...				about 15 years ago
adding some benchmarks and renaming some, again	potluri	More...				about 15 years ago
We need a handle_ptr to be exchanged for non-blocking packed GetS, to track completion once response packets are received	potluri	More...				about 15 years ago
1) Fixing GetS usage of buffer pool 2) Packed PutS/GetS/PutAccS need not wait for local completion to happen. as data is copied over.	potluri	More...				about 15 years ago
1) we cannot force advance from inside callback, so we allocate memory when the pools are empty 2) fixing an error in PutAccS while using buffer pools 3) there is a pending issue with GetS	potluri	More...				about 15 years ago
more formatting only	Jeff Hammond as jhammond	More...				about 15 years ago
Formatting only, because I think Sreeram likes the way Eclipse does it automagically.	Jeff Hammond as jhammond	More...				about 15 years ago
formatting only	Jeff Hammond as jhammond	More...				about 15 years ago
using buffer pool for putaccs and gets operations	potluri	More...				about 15 years ago
1) Adding buffer pools. PutS uses buffer pools and shows notable improvment in latency benchmark. Have to modify GetS and PutAccS operations to use the pools. 2) Modifying counter functions to match new prototypes and moving them to a new file. 3) The RMW code is commented out. This had to be fixed to match the new prototypes.	potluri	More...				about 15 years ago
renaming the benchmarks to make them more clear	potluri	More...				about 15 years ago
creating a separate Get benchmark for ARMCI over A1	potluri	More...				about 15 years ago
1. API changes, especially to atomics, moved RMW next to PutAcc 2. fixed include in MPID 3. added MPID developer notes that are incomplete	Jeff Hammond as jhammond	More...				about 15 years ago
flushing active handles and freeing active handle list structure in finalize	potluri	More...				about 15 years ago
changing the request pool design again. having request list inside handle was not a good idea. This was leading to request getting blocked until wait is called on the handle or the active count goes down to zero.	potluri	More...				about 15 years ago
minor stuff	Jeff Hammond as jhammond	More...				about 15 years ago
spelling and formatting there is still a TODO related to immediate flush. i'm not sure what we should do about that idea. i'm beginning to think it was a bad idea in the first place.	Jeff Hammond as jhammond	More...				about 15 years ago
added non-blocking flush, barrier and sync to the API	Jeff Hammond as jhammond	More...				about 15 years ago
fixing logic in non-recursive strided ops	potluri	More...				about 15 years ago
packing for few non-contguous chunks does not give performance, though this is a border case. adding thresholds to avoid this. fixing packet structure headers	potluri	More...				about 15 years ago
minor fixes using posix_memalign	potluri	More...				about 15 years ago
fixing use of Exchange_segments and using A1 calls instead of MPI in armci benchmarks	potluri	More...				about 15 years ago
adding ARMCI_Init_args and fixing a dumb error	potluri	More...				about 15 years ago
adding header	potluri	More...				about 15 years ago

←
1
2
…
70
71
72
73
74
75
76
77
78
…
83
84
→