Skip to content

Enhance performance for GpuToCpu for intel integrated GPU #311

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
hmaarrfk wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Enhance performance for GpuToCpu for intel integrated GPU #311

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 14 additions & 0 deletions src/vulkan/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -817,6 +817,20 @@ impl Allocator {
let mut memory_type_index_opt =
self.find_memorytype_index(&desc.requirements, mem_loc_preferred_bits);

// A GpuToCpu allocation exists to be read by the CPU, so being cached
// matters more than being coherent. Some adapters -- Intel's ANV on
// parts whose GPU does not share the CPU's last level cache -- expose
// cached memory and coherent memory but never both, and reading the
// uncached (write-combined) type is about two orders of magnitude
// slower. Callers must flush and invalidate non-coherent ranges
// regardless, so prefer cached before giving that up.
if memory_type_index_opt.is_none() && desc.location == MemoryLocation::GpuToCpu {
memory_type_index_opt = self.find_memorytype_index(
&desc.requirements,
vk::MemoryPropertyFlags::HOST_VISIBLE | vk::MemoryPropertyFlags::HOST_CACHED,
);
}

if memory_type_index_opt.is_none() {
let mem_loc_required_bits = match desc.location {
MemoryLocation::GpuOnly => vk::MemoryPropertyFlags::DEVICE_LOCAL,
Expand Down
Loading