Version 2022.07.00

CHANGELOG.md

@@ -1,3 +1,16 @@
+2022.07.00
+  - Improved performance of QR factorization on GPUs by moving panel to GPU:
+    5.5x faster on tall-skinny problem
+  - Added Cholesky QR `cholqr`; added as option in least squares solver, `gels`
+  - Added GPU implementation of `gemmA`, used when n is small (e.g., n <= nb)
+  - Added row and column scaling, `scale_row_col`
+  - Added print of individual tile
+  - Removed use of life counter in `gemm`, `herk`
+  - Removed setting MKL threads, which is no longer needed
+  - Removed `SLATE_NO_{HIP, CUDA}` macros in favor of
+    `BLAS_HAVE_{CUBLAS, ROCBLAS}` macros from BLAS++
+  - Introduced `tile` namespace
+
 2022.06.00
   - Fixed algorithm selection (issue #41)
   - Fixed set for triangular, trapezoid, symmetric, Hermitian matrices (tzset)

CMakeLists.txt

@@ -17,7 +17,7 @@ cmake_minimum_required( VERSION 3.18 )
 
 project(
     slate
-    VERSION 2022.06.00
+    VERSION 2022.07.00
     LANGUAGES CXX Fortran
 )
 

docs/doxygen/doxyfile.conf

@@ -40,7 +40,7 @@ PROJECT_NAME           = "SLATE"
 # could be handy for archiving the generated documentation or if some version
 # control system is used.
 
-PROJECT_NUMBER         = "2022.06.00"
+PROJECT_NUMBER         = "2022.07.00"
 
 # Using the PROJECT_BRIEF tag one can provide an optional one line description
 # for a project that appears at the top of each page and should give viewer a

include/slate/slate.hh

@@ -27,8 +27,8 @@
 namespace slate {
 
 // Version is updated by make_release.py; DO NOT EDIT.
-// Version 2022.06.00
-#define SLATE_VERSION 20220600
+// Version 2022.07.00
+#define SLATE_VERSION 20220700
 
 int version();
 const char* id();