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

Jump to bottom

Update test-submission-checker.yml | Check v4.1 results and not v4.0 … #1867

Closed
wants to merge 6 commits into from
Closed

Update test-submission-checker.yml | Check v4.1 results and not v4.0 … #1867

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
1a65774
Update test-submission-checker.yml | Check v4.1 results and not v4.0 …
arjunsuresh Oct 4, 2024
5a8367d
Update submission_checker.py | fix to skip accuracy files check for SDXL
arjunsuresh Oct 8, 2024
b7df93b
[Automated Commit] Format Codebase
arjunsuresh Oct 8, 2024
92d2a56
Merge branch 'master' into arjunsuresh-patch-1
arjunsuresh Oct 8, 2024
93fb9f8
[Automated Commit] Format Codebase
arjunsuresh Oct 8, 2024
0426afe
Update test-submission-checker.yml
arjunsuresh Oct 8, 2024
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
4 changes: 2 additions & 2 deletions .github/workflows/test-submission-checker.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,7 +29,7 @@ jobs:
- name: Install dependencies
run: |
python3 -m pip install cm4mlops
git clone https://github.com/mlcommons/inference_results_v4.0 --depth 1
git clone https://github.com/mlcommons/inference_results_v4.1 --depth 1
- name: Test MLPerf inference submission checker
run: |
cm run script --tags=run,mlperf,inference,submission,checker --adr.inference-src.tags=_branch.${{ github.event.pull_request.head.ref }},_repo.${{ github.event.pull_request.head.repo.html_url }} --adr.inference-src.version=custom --input=`pwd`/inference_results_v4.0 --src_version=v4.0 --quiet
cm run script --tags=run,mlperf,inference,submission,checker --adr.inference-src.tags=_branch.${{ github.event.pull_request.head.ref }},_repo.${{ github.event.pull_request.head.repo.html_url }} --adr.inference-src.version=custom --input=`pwd`/inference_results_v4.1 --src_version=v4.1 --extra_args=" --skip-extra-accuracy-files-check" --quiet
62 changes: 36 additions & 26 deletions loadgen/test_settings_internal.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,7 +53,7 @@ TestSettingsInternal::TestSettingsInternal(
server_ttft_latency(requested.server_ttft_latency),
server_tpot_latency(requested.server_tpot_latency),
infer_token_latencies(requested.infer_token_latencies),
token_latency_scaling_factor(requested.token_latency_scaling_factor){
token_latency_scaling_factor(requested.token_latency_scaling_factor) {
// Target QPS, target latency, and max_async_queries.
switch (requested.scenario) {
case TestScenario::SingleStream:
Expand Down Expand Up @@ -158,14 +158,15 @@ TestSettingsInternal::TestSettingsInternal(
// performance_sample_count == 0 makes it to be equal to loaded_samples.size()
if (sample_concatenate_permutation &&
requested.scenario == TestScenario::SingleStream) {
// set slack larger for 3D-UNet KiTS19 distribution, i.e. 50% latency << 90% latency
// set slack larger for 3D-UNet KiTS19 distribution, i.e. 50% latency << 90%
// latency
constexpr double kSlack = 2.0;
uint64_t expected_queries = kSlack * DurationToSeconds(target_duration) * target_qps;
min_query_count = min_query_count > expected_queries
? min_query_count
: expected_queries;
min_query_count +=
qsl_performance_sample_count - (min_query_count % qsl_performance_sample_count);
uint64_t expected_queries =
kSlack * DurationToSeconds(target_duration) * target_qps;
min_query_count =
min_query_count > expected_queries ? min_query_count : expected_queries;
min_query_count += qsl_performance_sample_count -
(min_query_count % qsl_performance_sample_count);
}

min_sample_count = min_query_count * samples_per_query;
Expand Down Expand Up @@ -335,13 +336,16 @@ void LogRequestedTestSettings(const TestSettings &s) {
MLPERF_LOG(detail, "requested_performance_sample_count_override",
s.performance_sample_count_override);
MLPERF_LOG(detail, "requested_sample_concatenate_permutation",
s.sample_concatenate_permutation);
s.sample_concatenate_permutation);
// Token latencies specific values
if (s.use_token_latencies){
MLPERF_LOG(detail, "requested_use_token_latencies", s.use_token_latencies);
if (s.scenario != TestScenario::Offline){
MLPERF_LOG(detail, "requested_server_ttft_latency", s.server_ttft_latency);
MLPERF_LOG(detail, "requested_server_tpot_latency", s.server_tpot_latency);
if (s.use_token_latencies) {
MLPERF_LOG(detail, "requested_use_token_latencies",
s.use_token_latencies);
if (s.scenario != TestScenario::Offline) {
MLPERF_LOG(detail, "requested_server_ttft_latency",
s.server_ttft_latency);
MLPERF_LOG(detail, "requested_server_tpot_latency",
s.server_tpot_latency);
}
}
#else
Expand Down Expand Up @@ -488,7 +492,7 @@ void TestSettingsInternal::LogAllSettings() const {
void TestSettingsInternal::LogSummary(AsyncSummary &summary) const {
summary("samples_per_query : ", samples_per_query);
summary("target_qps : ", target_qps);
if (!use_token_latencies){
if (!use_token_latencies) {
summary("target_latency (ns): ", target_latency.count());
} else {
summary("ttft_latency (ns): ", server_ttft_latency);
Expand Down Expand Up @@ -686,30 +690,36 @@ int TestSettings::FromConfig(const std::string &path, const std::string &model,
&performance_issue_same_index, nullptr);
lookupkv(model, scenario, "performance_sample_count_override",
&performance_sample_count_override, nullptr);
if (lookupkv(model, scenario, "sample_concatenate_permutation", &val, nullptr))
if (lookupkv(model, scenario, "sample_concatenate_permutation", &val,
nullptr))
sample_concatenate_permutation = (val == 1) ? true : false;
if (lookupkv(model, scenario, "test05", &val, nullptr))
test05 = (val == 1) ? true : false;
lookupkv(model, scenario, "test05_qsl_rng_seed", &test05_qsl_rng_seed, nullptr);
lookupkv(model, scenario, "test05_sample_index_rng_seed", &test05_sample_index_rng_seed,
lookupkv(model, scenario, "test05_qsl_rng_seed", &test05_qsl_rng_seed,
nullptr);
lookupkv(model, scenario, "test05_schedule_rng_seed", &test05_schedule_rng_seed, nullptr);
lookupkv(model, scenario, "test05_sample_index_rng_seed",
&test05_sample_index_rng_seed, nullptr);
lookupkv(model, scenario, "test05_schedule_rng_seed",
&test05_schedule_rng_seed, nullptr);

// keys to measure token metrics
if (lookupkv(model, scenario, "use_token_latencies", &val, nullptr)){
if (lookupkv(model, scenario, "use_token_latencies", &val, nullptr)) {
use_token_latencies = (val == 1) ? true : false;
}
if (use_token_latencies){
lookupkv(model, "Server", "ttft_latency", &server_ttft_latency, nullptr, 1000 * 1000);
lookupkv(model, "Server", "tpot_latency", &server_tpot_latency, nullptr, 1000 * 1000);
if (use_token_latencies) {
lookupkv(model, "Server", "ttft_latency", &server_ttft_latency, nullptr,
1000 * 1000);
lookupkv(model, "Server", "tpot_latency", &server_tpot_latency, nullptr,
1000 * 1000);
}

// keys to infer token metrics
if (lookupkv(model, scenario, "infer_token_latencies", &val, nullptr)){
if (lookupkv(model, scenario, "infer_token_latencies", &val, nullptr)) {
infer_token_latencies = (val == 1) ? true : false;
}
if (infer_token_latencies){
lookupkv(model, scenario, "token_latency_scaling_factor", &token_latency_scaling_factor, nullptr, 1);
if (infer_token_latencies) {
lookupkv(model, scenario, "token_latency_scaling_factor",
&token_latency_scaling_factor, nullptr, 1);
}
// keys that apply to SingleStream
lookupkv(model, "SingleStream", "target_latency_percentile", nullptr,
Expand Down
Loading
Loading