vulkan: add pipeline barriers for memcpy read operations (llama/23770)

* vulkan: add pipeline barriers for memcpy read/write operations

* remove unnecessary host write pipeline barriers
This commit is contained in:
Ruben Ortlam 2026-06-12 16:43:50 +02:00 committed by Georgi Gerganov
parent f049fff95a
commit dc195118ef
1 changed files with 17 additions and 0 deletions

View File

@ -7741,6 +7741,23 @@ static void ggml_vk_buffer_read_2d(vk_buffer& src, size_t offset, void * dst, si
if(src->memory_property_flags & vk::MemoryPropertyFlagBits::eHostVisible && src->device->uma) {
GGML_ASSERT(src->memory_property_flags & vk::MemoryPropertyFlagBits::eHostCoherent);
std::lock_guard<std::recursive_mutex> guard(src->device->mutex);
vk_context subctx = ggml_vk_create_temporary_context(src->device->compute_queue.cmd_pool);
ggml_vk_ctx_begin(src->device, subctx);
subctx->s->buffer->buf.pipelineBarrier(
vk::PipelineStageFlagBits::eComputeShader | vk::PipelineStageFlagBits::eTransfer,
vk::PipelineStageFlagBits::eHost,
{},
{ { vk::AccessFlagBits::eShaderWrite | vk::AccessFlagBits::eTransferWrite,
vk::AccessFlagBits::eHostRead } },
{}, {});
ggml_vk_ctx_end(subctx);
ggml_vk_submit(subctx, src->device->fence);
VK_CHECK(src->device->device.waitForFences({ src->device->fence }, true, UINT64_MAX),
"vk_buffer_read_2d uma waitForFences");
src->device->device.resetFences({ src->device->fence });
ggml_vk_queue_command_pools_cleanup(src->device);
if (width == spitch && width == dpitch) {
memcpy(dst, (const uint8_t *) src->ptr + offset, width * height);
} else {