]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
kompute : improve backend to pass test_backend_ops (#10542)
authorSergio López <redacted>
Thu, 28 Nov 2024 11:51:38 +0000 (12:51 +0100)
committerGitHub <redacted>
Thu, 28 Nov 2024 11:51:38 +0000 (12:51 +0100)
commit2025fa67e94358deda4740a74fe9803916cb2f60
tree8ec1be18c5eb6f87544029d9153e9d6ea0e54f31
parentc6bc73951ed52466392b1abda98c28ecbe522c7f
kompute : improve backend to pass test_backend_ops (#10542)

* kompute: op_unary: reject unsupported parameters

Signed-off-by: Sergio Lopez <redacted>
* kompute: softmax: implement ALiBi support

Signed-off-by: Sergio Lopez <redacted>
* kompute: rope: implement neox and phi3 support

Signed-off-by: Sergio Lopez <redacted>
* kompute: op_mul_mat_q4_k permutted support

Signed-off-by: Sergio Lopez <redacted>
* kompute: op_mul_mat_[q4_0|q4_1|q8_0] permutted support

Signed-off-by: Sergio Lopez <redacted>
* kompute: op_mul_mat_f16 permutted support

Signed-off-by: Sergio Lopez <redacted>
* kompute: op_mul_mat_q6_k permutted support

Signed-off-by: Sergio Lopez <redacted>
---------

Signed-off-by: Sergio Lopez <redacted>
16 files changed:
ggml/src/ggml-kompute/CMakeLists.txt
ggml/src/ggml-kompute/ggml-kompute.cpp
ggml/src/ggml-kompute/kompute-shaders/common.comp
ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_f16.comp
ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_k.comp
ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q6_k.comp
ggml/src/ggml-kompute/kompute-shaders/op_mul_mv_q_n.comp
ggml/src/ggml-kompute/kompute-shaders/op_mul_mv_q_n_pre.comp
ggml/src/ggml-kompute/kompute-shaders/op_rope_f16.comp [deleted file]
ggml/src/ggml-kompute/kompute-shaders/op_rope_f32.comp [deleted file]
ggml/src/ggml-kompute/kompute-shaders/op_rope_neox_f16.comp [new file with mode: 0644]
ggml/src/ggml-kompute/kompute-shaders/op_rope_neox_f32.comp [new file with mode: 0644]
ggml/src/ggml-kompute/kompute-shaders/op_rope_norm_f16.comp [new file with mode: 0644]
ggml/src/ggml-kompute/kompute-shaders/op_rope_norm_f32.comp [new file with mode: 0644]
ggml/src/ggml-kompute/kompute-shaders/op_softmax.comp
ggml/src/ggml-kompute/kompute-shaders/rope_common.comp