forked from wu-kan/HPL-AI
-
Notifications
You must be signed in to change notification settings - Fork 0
/
NEWS
23 lines (19 loc) · 866 Bytes
/
NEWS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
==============================================================
HPL-AI Mixed-Precision Benchmark v2.3a -- March 14, 2021
==============================================================
History
- 03/14/2021 HPL-AI Public release of Version 2.3a
Junkang Huang added the parallel GMRES and matrix construction
part (<https://github.com/schuangs/hpl-ai-with-IR>) for HPL-AI.
Note that the GMRES did not check the backward-error, which may
cause the result to be invalid.
Kan Wu reimplemented HPL-AI with some bug fixes and some minor
optimizations. The C souce code of hpl-2.3 remained the same,
while took advantages of modern C++ features such as templates,
overloading, and GPU backends via blaspp:
-- [A] include/hplai*.hh
-- [A] src/*/HPLAI_*.cc
-- [A] testing/*/HPLAI_*.cc
-- [M] configure.ac
-- [M] src/Makefile.am
-- [M] testing/Makefile.am