llama.cpp/ggml/src/ggml-cpu
Daniel Bevenius 3913f8730e
ggml : fix padding in timestep embedding kernels (#15932)
* ggml : remove adding extra dim timestep embedding

This commit updates the ggml_timestep_embedding function to no longer
add an extra dimension when the specified dimension is odd.

The motivation for this change is that this introduces an unnecessary
dimension when the dimension is odd, which caused an issue in the
kernels which were not expecting this extra dimension and it resulted in
uninitialized memory for the second to last dimension.

* ggml-cuda : fix padding in timestep embedding kernel

This commit removes the zeroing out of the last dimension now that we
are not adding the extra padding dimension.

* ggml-metal : fix padding in timestep embedding kernel

This commit fixes the zero padding for odd dimensions in
the timestep embedding kernel

* ggml-opencl : fix padding in timestep embedding kernel

This commit fixes the zero padding for odd dimensions in
the timestep embedding kernel.

* ggml-sycl : fix padding in timestep embedding kernel

This commit fixes the zero padding for odd dimensions in
the timestep embedding kernel.

* ggml-vulkan : fix padding in timestep embedding kernel

This commit fixes the zero padding for odd dimensions in
the timestep embedding kernel.

* ggml-cpu : fix padding in timestep embedding function

This commit removes the zeroing out of the last dimension now that we
are not adding the extra padding dimension.
2025-09-16 15:25:57 +02:00
..
amx ggml-cpu: enable IBM NNPA Vector Intrinsics (#14317) 2025-06-25 23:49:04 +02:00
arch ggml-cpu: clean up s390x SIMD (#15855) 2025-09-08 02:18:28 +08:00
cmake ggml : build backends as libraries (#10256) 2024-11-14 18:04:35 +01:00
kleidiai kleidiai: fix GGML_ASSERT(*cur_backend_id != -1) failed (#15614) 2025-09-11 12:45:40 +02:00
llamafile llamafile: PowerPC Sgemm Optimization (#15558) 2025-08-26 23:35:25 +08:00
CMakeLists.txt ggml-cpu : add check for ARM MATMUL_INT8/i8mm support (#15922) 2025-09-11 14:39:12 +01:00
arch-fallback.h ggml-cpu: Support Q5_0 and Q5_1 on s390x (#15486) 2025-08-22 16:11:04 +08:00
binary-ops.cpp cpu: de-duplicate some of the operators and refactor (ggml/1144) 2025-03-30 08:33:31 +03:00
binary-ops.h cpu: de-duplicate some of the operators and refactor (ggml/1144) 2025-03-30 08:33:31 +03:00
common.h ggml-cpu: enable IBM NNPA Vector Intrinsics (#14317) 2025-06-25 23:49:04 +02:00
ggml-cpu-impl.h ggml-cpu: clean up s390x SIMD (#15855) 2025-09-08 02:18:28 +08:00
ggml-cpu.c ggml: allow casting between f32 and i32 (#15783) 2025-09-08 12:33:01 +02:00
ggml-cpu.cpp vulkan: sort graph to allow more parallel execution (#15850) 2025-09-09 02:10:07 +08:00
hbm.cpp ggml-cpu : split arch-specific implementations (#13892) 2025-06-09 16:47:13 +02:00
hbm.h ggml-cpu : split arch-specific implementations (#13892) 2025-06-09 16:47:13 +02:00
ops.cpp ggml : fix padding in timestep embedding kernels (#15932) 2025-09-16 15:25:57 +02:00
ops.h ggml: add ops for WAN video model (cuda && cpu) (#15669) 2025-09-04 10:38:49 +02:00
quants.c llama : add gpt-oss (#15091) 2025-08-05 22:10:36 +03:00
quants.h llama : add gpt-oss (#15091) 2025-08-05 22:10:36 +03:00
repack.cpp ggml : repack block_iq4_nlx8 (#14904) 2025-08-13 11:09:39 +03:00
repack.h ggml : repack block_iq4_nlx8 (#14904) 2025-08-13 11:09:39 +03:00
simd-mappings.h ggml-cpu: drop support for nnpa intrinsics (#15821) 2025-09-06 11:27:28 +08:00
traits.cpp ggml : fix fallback to CPU for ununsupported ops (#15118) 2025-08-06 14:37:35 +02:00
traits.h ggml : fix fallback to CPU for ununsupported ops (#15118) 2025-08-06 14:37:35 +02:00
unary-ops.cpp cpu: de-duplicate some of the operators and refactor (ggml/1144) 2025-03-30 08:33:31 +03:00
unary-ops.h cpu: de-duplicate some of the operators and refactor (ggml/1144) 2025-03-30 08:33:31 +03:00
vec.cpp ggml-cpu : optimize RVV kernels (#15720) 2025-09-03 16:16:21 +08:00
vec.h ggml-cpu : optimize RVV kernels (#15720) 2025-09-03 16:16:21 +08:00