cuda : Use byte strides for roll to allow non-contiguous ROLL operations (#29547)

This commit is contained in:
Bertay Eren authored and GitHub committed 2026-10-08 09:46:38 +03:00
1 parent fda1866613
commit d888016041
2 files changed
+7 -3

No files matched your search

+1 -1
View File
@@ -5658,7 +5658,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
return max_bias == 0.0f;
}
case GGML_OP_ROLL:
if(op->src[0]->type == GGML_TYPE_F32 && ggml_is_contiguous(op->src[0])) {
if(op->src[0]->type == GGML_TYPE_F32) {
return true;
}
return false;
+6 -2
View File
@@ -17,6 +17,10 @@ static __global__ void roll_f32_cuda(const float * __restrict__ src,
const int64_t ne01,
const int64_t ne02,
const int64_t ne03,
const int64_t nb00,
const int64_t nb01,
const int64_t nb02,
const int64_t nb03,
const int s0,
const int s1,
const int s2,
@@ -39,7 +43,7 @@ static __global__ void roll_f32_cuda(const float * __restrict__ src,
const int64_t d3 = wrap_index(i3 - s3, ne03);
dst[i3 * (ne00 * ne01 * ne02) + i2 * (ne01 * ne00) + i1 * ne00 + i0] =
src[d3 * (ne00 * ne01 * ne02) + d2 * (ne01 * ne00) + d1 * ne00 + d0];
src[(d3 * nb03 + d2 * nb02 + d1 * nb01 + d0 * nb00) / sizeof(float)];
}
void ggml_cuda_op_roll(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
@@ -63,5 +67,5 @@ void ggml_cuda_op_roll(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
int64_t num_blocks = (sz + CUDA_ROLL_BLOCK_SIZE - 1) / CUDA_ROLL_BLOCK_SIZE;
roll_f32_cuda<<<num_blocks, CUDA_ROLL_BLOCK_SIZE, 0, stream>>>(
src0_d, dst_d, ne00, ne01, ne02, ne03, s0, s1, s2, s3);
src0_d, dst_d, ne00, ne01, ne02, ne03, nb00, nb01, nb02, nb03, s0, s1, s2, s3);
}