a408a3b2c9
Drop Imath; Add lookup table for f16 -> f32 conversions
3 years ago
d8f356ac9f
Merge branch 'master' of github.com:ggerganov/whisper.cpp
3 years ago
4362818573
Wrap Imath calls to avoid static function warnings
3 years ago
67a1aa398a
Add Imath to system information
3 years ago
7bba2dd63d
Add Imath support for fp16-fp32 conversions
3 years ago
97e0dfc8ba
Add SSE3 to system information
3 years ago
c4d8664603
Add support for SSE3 SIMD
3 years ago
424c410c42
ggml : improve f16 acceleration for POWER9 ppc64le
3 years ago
4e0b2069e7
ggml : barrier refactor + static functions
3 years ago
ac521a566e
ggml : simplify the SIMD code ( #324 )
...
* ggml : simplify the SIMD code
* ggml : generic reduce for all register sizes + comments
3 years ago
7282e2109e
ggml : use vaddvq_f32 for slightly more efficient reduce
3 years ago
466ceebb78
ggml : add f16 acceleration for POWER9 ppc64le
3 years ago
493d94130d
ggml : make consts static ( #317 )
...
These shouldn't be able to be referenced outside the compilation unit.
3 years ago
fa463313ad
minor : small code cleanups ( #302 )
...
* Small code cleanups
- fix indentation
- remove extra semicolons
- remove extra break after returns in case statements
- remove unnecessary call to .data() on string
- use empty() instead of checking size()
- no need to check for nullptr before free
- remove unnecessary initialization of string to ""
* minor : switch case always break
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
3 years ago
e1432dd91a
Check for both __ARM_NEON and __ARM_FEATURE_FMA so that the project can be compiled for armv7a.
...
Android armeabi-v7a's NEON support doesn't support FMA unless configured with `-mfpu=neon-fp-armv8`, which would need runtime checks.
* Also removed ABI filter from Android project.
3 years ago
419b8a6402
Add AVX,AVX2 support for ggml_vec_scale_f32
3 years ago
a7047b2a28
ggml : implement ggml_compute_forward_dup_f16() special cases
3 years ago
0f11759406
ggml : make more compatible with c99 ( #262 )
3 years ago
f66ac6dc4f
ggml : fix indentation
3 years ago
9955fa4ed7
ggml : make compatible with c99 ( #262 )
3 years ago
e70d47baab
Remove C++20 requirement ( #257 )
...
* Remove C++20 requirement
* Roll back C features not supported in VS2017
3 years ago
3b1aacbe6d
talk : talk with AI in the terminal
3 years ago
50a061b313
ggml : add alternative cblas_sgemm call
3 years ago
04a16bbf11
fix compilation on haiku
3 years ago
b6597539f9
ggml : fix typo in previous commit
3 years ago
9a4b7a916e
ggml : use macros to inline FP16 <-> FP32 conversions
3 years ago
f8ec718b76
ggml : add F16C CPU flag check
3 years ago
35b40a93b9
add fp16/fp32 convert intrinsics
3 years ago
061fc81bd6
ggml : remove inline specifier from fp16 <-> fp32 converters
3 years ago
388e9f79ad
ggml : fix the fix
3 years ago
35cd29ce1f
ggml : fix cross-compile Linux -> Window with mingw ( #168 )
3 years ago
804f36aa2c
ggml: change inline ggml_fp16_to_fp32, ggml_fp16_t ggml_fp32_to_fp16
3 years ago
83456076f0
add AVX support
3 years ago
2065572a11
ggml : fix Windows build
3 years ago
0bfe728b84
Fix the Windows pthread_create shim
...
The current implementation doesn't actually set the out parameter,
and it returns 0 on failure instead of on success.
3 years ago
75171c2b79
ggml : multi-thread the ggml_add operator
3 years ago
137321915f
ggml : fix the check for NEON support ( #7 )
...
Was using the wrong preprocessor macro
3 years ago
24cd12f647
Cross compilation ( #121 )
...
* Cross compile windows
* set env properly
* rm log
* fix review
* Add back space
3 years ago
8dac3c6e10
Fixed sched_yield
3 years ago
6417e59aad
Implemenated sched_yield function for Windows
3 years ago
e5044f87d9
ggml : fix barrier
3 years ago
a272f10b2e
ggml : fix thread-safety of ggml_init and ggml_free
3 years ago
fbd513b813
Add OpenBLAS support
...
Supported via CMake - just add:
cmake .. -DWHISPER_SUPPORT_OPENBLAS=ON
On Ubuntu, you have to install the library like this:
apt install libopenblas-dev
Unfortunately, I don't observe any benefit compared to the
original AVX2 + FP16 implementation. Maybe I'm missing something
3 years ago
34bb3ab0cf
ggml : add system info functions
3 years ago
c6710efde2
refactoring : move main + stream in examples + other stuff
3 years ago
db460b78ff
wip : WASM 128-bit SIMD support
3 years ago
e905c6f827
wip : initial WASM port
...
Works but it is very slow because no SIMD is used.
For example, jfk.wav is processed in ~23 seconds using "tiny.en" model
3 years ago
19817711b4
Add reference to FP16 repo
3 years ago
e36aabe00d
Correct implementation of FP16 GELU
...
Can toggle it via the GGML_GELU_FP16 macro
3 years ago
91632eb6ea
Revert GELU change
...
Seems it does not work on x86 for some reason
3 years ago
Abitofevrything