Commits

Commits on Jun 12, 2024

Fix 'get_real_value' on placeholder nodes (#127698)

BowenBao
authored and
facebook-github-bot
committedJun 12, 2024
Fix 'get_attr' call in dynamo 'run_node' (#127696)

BowenBao
authored and
facebook-github-bot
committedJun 12, 2024

Commits on May 9, 2024

Benchmark onnx export w/ ort fusions (#125700)

BowenBao
authored and
facebook-github-bot
committedMay 9, 2024

Commits on Feb 1, 2024

Deepcopy model to another device before export to avoid OOM (#118710)

BowenBao
authored and
facebook-github-bot
committedFeb 1, 2024

Commits on Jan 2, 2024

Add copy before export for perf bench to avoid mutating base model (#115945)

BowenBao
authored and
facebook-github-bot
committedJan 2, 2024

Commits on Dec 18, 2023

Add proper iobinding synchronize for ONNX cuda bench (#115773)

BowenBao
authored and
facebook-github-bot
committedDec 18, 2023
Dump sarif diagnostics for failed onnx exports in benchmark (#115673)

BowenBao
authored and
facebook-github-bot
committedDec 18, 2023
Set correct cuda.current_device for multi-device onnx performance bench (#115670)

BowenBao
authored and
facebook-github-bot
committedDec 18, 2023

Commits on Dec 8, 2023

Fix model name retrieval and remove unused argument (#115108)

BowenBao
authored and
facebook-github-bot
committedDec 8, 2023

Commits on Dec 4, 2023

Remove double export and session init in perf test (#114907)

BowenBao
authored and
facebook-github-bot
committedDec 4, 2023

Commits on Dec 1, 2023

Add warmup for onnx cuda runs (#114821)

BowenBao
authored and
facebook-github-bot
committedDec 1, 2023
Relax tolerance for cuda accuracy check (#114767)

BowenBao
authored and
facebook-github-bot
committedDec 1, 2023

Commits on Nov 23, 2023

Benchmark to save sample inputs to disk before running (#114163)

BowenBao
authored and
facebook-github-bot
committedNov 23, 2023

Commits on Nov 18, 2023

Fix bench w/ iobinding; Remove cpu fallback (#113703)

BowenBao
authored and
facebook-github-bot
committedNov 18, 2023

Commits on Nov 15, 2023

Fix stable_diffusion_xl fp16 (#2032 )

BowenBao
authored and
facebook-github-bot
committedNov 15, 2023

Commits on Oct 25, 2023

Delete deepcopied model after use in benchmark to reduce memory consumption (#111868)

BowenBao
authored and
facebook-github-bot
committedOct 25, 2023
Apply same 'pick_grad' on generating fp64 reference outputs (#111593)

BowenBao
authored and
facebook-github-bot
committedOct 25, 2023
Enable onnx inlining in benchmark for >2GB models (#111867)

BowenBao
authored and
facebook-github-bot
committedOct 25, 2023

Commits on Oct 20, 2023

Benchmark to store test data along exported model (#111095)

BowenBao
authored and
facebook-github-bot
committedOct 20, 2023

Commits on Oct 19, 2023

Add dynamo_onnx_aot_inline to bench (#110183)

BowenBao
authored and
facebook-github-bot
committedOct 19, 2023

Commits on Oct 18, 2023

Fix 'stable_diffusion_xl' (#1994 )

BowenBao
authored and
facebook-github-bot
committedOct 18, 2023

Commits on Jul 19, 2023

Remove unnecessary model attribute assignment on 'freqs_cis' (#1766 )

BowenBao
authored and
facebook-github-bot
committedJul 19, 2023