Skip to content

Commit 4170e60

Browse files
committed
Incorporate Shreyas comments
--- Signed-off-by: Brad Chamberlain <[email protected]>
1 parent 3069d5c commit 4170e60

File tree

1 file changed

+5
-5
lines changed

1 file changed

+5
-5
lines changed

CHANGES.md

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -9,7 +9,7 @@ released December 12, 2024
99
Highlights (see subsequent sections for further details)
1010
--------------------------------------------------------
1111
* added a new package module that supports calling to Python from Chapel
12-
* improved the Chapel linter, language server, and Python bindings in many ways
12+
* made improvements to the Chapel linter, language server, and Python bindings
1313
* added new builds for AmazonLinux 2023, Fedora 41, and Spack using ROCm
1414
* significantly reduced overhead for many tasking idioms due to Qthreads 1.21
1515
* optimized key communication idioms when using `CHPL_COMM=ofi`
@@ -59,7 +59,7 @@ Changes / Feature Improvements in Package Modules
5959
New Standard Layout and Distribution Features
6060
---------------------------------------------
6161
* added new utility methods for sparse layouts and distributions
62-
- `.getCoordinates()` returns indices of nonzeroes for COO sparse arrays
62+
- `.getCoordinates()` returns indices of non-zeroes for COO sparse arrays
6363
- `.localSubarrays()` yields each locale's subarray for a block-sparse array
6464
- `.getLocalSubarray()` returns the local subarray for a block-sparse array
6565
- added `ref`-returning overloads of existing `.getLocalSubarray()` methods
@@ -88,7 +88,7 @@ Performance Optimizations / Improvements
8888
* optimized inter-locale transfers of sparse CSR/CSC arrays for `--fast` runs
8989
* optimized `CHPL_COMM=ofi` by adding support for non-blocking PUT and GET ops
9090
* optimized `CHPL_COMM=ofi` active messages via use of non-blocking PUTs
91-
* improved performance on Arm-based Macs by using a 128-bit cache line
91+
* improved performance on ARM-based Macs by using a 128-bit cache line
9292

9393
GPU Computing
9494
-------------
@@ -880,7 +880,7 @@ Standard Layouts and Distributions
880880
* added a new initializer to 'LayoutCS', supporting positional arguments
881881
* added new utility routines for sparse arrays stored using 'LayoutCS':
882882
- `rows`/`cols()` queries to return the dense range of indices per dimension
883-
- `[cols|rows]AndVals()` iterators that yield nonzeroes' indices and values
883+
- `[cols|rows]AndVals()` iterators that yield non-zeroes' indices and values
884884
* enabled `.targetLocales()` queries on sparse `blockDist` domains/arrays
885885
* added new utility routines for sparse `blockDist`-distributed domains/arrays:
886886
- `setLocalSubdomain()` to assign the current locale's subdomain of nonzeroes
@@ -7681,7 +7681,7 @@ Bug Fixes
76817681
* fixed problems when an `owned` argument is used in a later default expression
76827682
* `isRecordType` now returns `false` for `owned` and `shared` classes
76837683
* fixed a bug in which parsing files multiple times caused duplicate modules
7684-
* fixed a bug in tracking number of nonzeroes for distributed sparse domains
7684+
* fixed a bug in tracking number of non-zeroes for distributed sparse domains
76857685
* fixed a bug in index addition to distributed CSC domains
76867686
* fixed a bug in which missing line numbers in the IR caused an internal error
76877687
* fixed a null-termination issue with `string.join`

0 commit comments

Comments
 (0)