Skip to content
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

[GPU] fix property overwritten issue #28209

Merged
merged 3 commits into from
Jan 3, 2025
Merged

[GPU] fix property overwritten issue #28209

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e637a9f
[GPU] fix property overwritten issue
riverlijunjie Dec 26, 2024
b732978
Update
riverlijunjie Jan 2, 2025
be9f488
Merge branch 'master' into river/fix_property_overwritten_issue
riverlijunjie Jan 2, 2025
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
3 changes: 3 additions & 0 deletions src/plugins/intel_gpu/include/intel_gpu/runtime/execution_config.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -150,6 +150,7 @@ class ExecutionConfig {
void apply_performance_hints(const cldnn::device_info& info);
void apply_priority_hints(const cldnn::device_info& info);
void apply_debug_options(const cldnn::device_info& info);
void update_specific_default_properties(const cldnn::device_info& info);

template <typename T, PropertyMutability mutability>
void apply_rt_info_property(const ov::Property<T, mutability>& property, const ov::RTMap& rt_info) {
Expand All @@ -167,6 +168,8 @@ class ExecutionConfig {

std::map<std::string, PropertyVisibility> supported_properties;
std::map<std::string, BaseValidator::Ptr> property_validators;

bool specific_default_properties_is_set = false;
};

} // namespace intel_gpu
Expand Down
30 changes: 20 additions & 10 deletions src/plugins/intel_gpu/src/runtime/execution_config.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -229,7 +229,27 @@ void ExecutionConfig::apply_hints(const cldnn::device_info& info) {
apply_debug_options(info);
}

void ExecutionConfig::update_specific_default_properties(const cldnn::device_info& info) {
// These default properties should be set once.
if (specific_default_properties_is_set)
return;
specific_default_properties_is_set = true;

// Enable KV-cache compression by default for non-systolic platforms
if (get_property(ov::hint::kv_cache_precision) == ov::element::undefined && !info.supports_immad) {
set_property(ov::hint::kv_cache_precision(ov::element::i8));
}

// Enable dynamic quantization by default for non-systolic platforms
if (get_property(ov::hint::dynamic_quantization_group_size) == 0 && !info.supports_immad) {
set_property(ov::hint::dynamic_quantization_group_size(32));
}
}

void ExecutionConfig::apply_user_properties(const cldnn::device_info& info) {
// Update specific default properties, call once before internal_properties updated.
update_specific_default_properties(info);

// Copy internal properties before applying hints to ensure that
// a property set by hint won't be overriden by a value in user config.
// E.g num_streams=AUTO && hint=THROUGHPUT
Expand All @@ -249,16 +269,6 @@ void ExecutionConfig::apply_user_properties(const cldnn::device_info& info) {
set_property(ov::intel_gpu::queue_type(QueueTypes::in_order));
}

// Enable KV-cache compression by default for non-systolic platforms
if (!is_set_by_user(ov::hint::kv_cache_precision) && !info.supports_immad) {
set_property(ov::hint::kv_cache_precision(ov::element::i8));
}

// Enable dynamic quantization by default for non-systolic platforms
if (!is_set_by_user(ov::hint::dynamic_quantization_group_size) && !info.supports_immad) {
set_property(ov::hint::dynamic_quantization_group_size(32));
}

user_properties.clear();
}

Expand Down
Loading