 Attila Afra
|
d94e393596
|
SyncMode cleanup
|
2024-11-09 18:50:17 +02:00 |
|
|
Attila Afra
|
a179f0b909
|
fixed memory leak when releasing SYCL/CUDA/HIP devices, minor cleanup
|
2024-02-24 23:39:29 +02:00 |
|
|
Attila Afra
|
aae372821a
|
fixed crash when releasing a buffer after the device, cleanup
|
2024-01-06 22:33:24 +02:00 |
|
|
Attila Afra
|
6431f3e7c6
|
added Heap, Arena and ArenaPlanner, removed MetalGraph, refactoring
|
2023-11-11 23:38:04 +02:00 |
|
|
Attila Afra
|
5bd83ca3e8
|
Merge devel into aafra/metal
# Conflicts:
# api/api.cpp
# core/buffer.cpp
# core/buffer.h
# core/context.cpp
# core/generic_graph.cpp
# core/tensor.cpp
# core/tensor.h
# devices/gpu/gpu_input_process.h
# devices/gpu/gpu_output_process.h
# devices/sycl/sycl_conv_gen.h
|
2023-10-14 02:16:42 +03:00 |
|
|
Attila Afra
|
28cae7701f
|
removed Buffer map/unmap, minor Tensor refactoring
|
2023-10-08 23:11:04 +03:00 |
|
|
Attila Afra
|
d4af2c6674
|
fixed possible crash when releasing GPU devices, buffers or filters
|
2023-10-08 01:11:52 +03:00 |
|
|
Attila Afra
|
019277b717
|
added more API parameter checks
|
2023-10-06 22:39:50 +03:00 |
|
|
Attila Afra
|
b11a55a9e0
|
initial MetalBuffer refactor
|
2023-06-28 02:21:17 +03:00 |
|
|
Attila Afra
|
4714191afb
|
USM cleanup
|
2023-05-29 19:41:37 +03:00 |
|
|
Attila Afra
|
73deab0e7a
|
updated copyright year
|
2023-05-15 20:09:34 +03:00 |
|
|
Attila Afra
|
9f14004a77
|
use managed memory by default for buffers (if supported)
|
2023-05-10 20:09:10 +03:00 |
|
|
Attila Afra
|
db2ad62960
|
fixed bugs in mapping buffers
|
2023-05-09 17:29:11 +03:00 |
|
|
Attila Afra
|
18581c739b
|
fixed buffer bug
|
2023-05-09 02:37:04 +03:00 |
|
|
Attila Afra
|
96188c2086
|
allocate unpinned memory when mapping buffer due to high cost of pinned alloc
|
2023-05-08 01:42:12 +03:00 |
|
|
Attila Afra
|
a563021988
|
added test whether the output is deterministic
|
2023-04-08 17:38:33 +03:00 |
|
|
Attila Afra
|
2a32cb07c5
|
Graph class split into abstract Graph and GenericGraph
|
2023-04-05 03:43:43 +03:00 |
|
|
Attila Afra
|
5704d6aab2
|
added automatic tensor allocation and op unfusion to Graph
|
2023-04-03 17:36:43 +03:00 |
|
|
Attila Afra
|
7c10b8bfe0
|
added physical device, refactored get/set API, fixed crash when destroying uncommitted device
|
2023-03-22 02:29:48 +02:00 |
|
|
Attila Afra
|
b4fea90f3d
|
avoid symbol collision when using OIDN_API_NAMESPACE
|
2023-02-13 15:35:16 +02:00 |
|
|
Attila Afra
|
fbcd71d48e
|
minor Device/Engine refactoring
|
2022-12-17 15:53:29 +02:00 |
|
|
Attila Afra
|
e677f98be0
|
added external buffer API, CUDA implementation
|
2022-12-15 14:51:26 +02:00 |
|
|
Attila Afra
|
ac3c488205
|
USMBuffer fix
|
2022-11-05 21:34:09 +02:00 |
|
|
Attila Afra
|
ef88c7d33b
|
API: added async buffer read/write functions
|
2022-11-04 22:46:37 +02:00 |
|
|
Attila Afra
|
cec74460fb
|
added SYCL multi-tile GPU support, major refactoring
|
2022-10-21 20:54:19 +03:00 |
|
|
Attila Afra
|
5d6d9c6fe1
|
check whether user-specified pointers are accessible to the device/host
|
2022-05-17 17:59:45 +03:00 |
|
|
Attila Afra
|
e06fe90763
|
added initial HIP support, extended the buffer API
|
2022-03-30 02:34:17 +03:00 |
|
|
Attila Afra
|
3c0850f420
|
improved scratch memory management, cleanups, bumped version to 2.0
|
2022-03-12 15:30:05 +02:00 |
|