 fanasina
|
1eb30ff794
|
Makefile: change to absolute path, and debug create lib.so if need
|
2025-10-28 19:13:28 +01:00 |
|
|
fanasina
|
60edd5e709
|
tensor_cl: add script test memory cl
|
2025-08-05 21:37:41 +02:00 |
|
|
fanasina
|
65d4d8f67c
|
y_socket: add poll stdin, debug node init
|
2025-07-28 23:03:36 +02:00 |
|
|
fanasina
|
6d5e060d6c
|
trying to understand leak valgrind with opencl
|
2025-07-09 12:05:14 +02:00 |
|
|
fanasina
|
924ff3a7dd
|
add functions calculate parallel updates neurons, and debug some leak functions
|
2024-02-28 09:57:21 +01:00 |
|
|
fanasina
|
ede9ce4d59
|
remove some logs for debug in cl_tensor.c
|
2024-02-12 23:24:06 +01:00 |
|
|
fanasina
|
c07ed1b160
|
debug tensor prod contract 2d cl and pthread
|
2024-02-12 12:08:16 +01:00 |
|
|
fanasina
|
60e4b16308
|
another debug with valgrind in fmock and tensor
|
2024-02-12 01:34:08 +01:00 |
|
|
fanasina
|
e1171d720b
|
debug all leak mem with valgrind in ftest and tensor rep
|
2024-02-08 23:57:10 +01:00 |
|
|
fanasina
|
5fc1681e19
|
add test, I notice that if we increase parallel tests (-p 4 form me) it was less precision in gpu (cl) calculation: some tests failed
|
2024-02-02 23:30:41 +01:00 |
|
|
fanasina
|
7dd749f85e
|
add endian option in cl tensor functions
|
2024-02-02 23:13:22 +01:00 |
|
|
fanasina
|
672cc3c2fa
|
create cl test
|
2024-02-01 17:14:36 +01:00 |
|