Skip to content

[cherry-pick] fix bert bug using trt6 when compile with CUDA_ARCH_NAME=All #24574

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

Merged
merged 1 commit into from
May 15, 2020
Merged

[cherry-pick] fix bert bug using trt6 when compile with CUDA_ARCH_NAME=All #24574

Show file tree
Hide file tree
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
6 changes: 3 additions & 3 deletions paddle/fluid/inference/tensorrt/convert/emb_eltwise_layernorm.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,9 +91,9 @@ class EmbEltwiseLayerNormOpConverter : public OpConverter {
input_embs, bias, scale, emb_sizes, bias_size, scale_size, hidden,
eps);
#else
PADDLE_THROW(
platform::errors::Fatal("use EmbEltwiseLayernormPluginDynamic "
"FP16, but GPU doesn't have FP16."));
plugin = new plugin::EmbEltwiseLayernormPluginDynamic<float>(
input_embs, bias, scale, emb_sizes, bias_size, scale_size, hidden,
eps);
#endif
} else {
plugin = new plugin::EmbEltwiseLayernormPluginDynamic<float>(
Expand Down
1 change: 0 additions & 1 deletion paddle/fluid/inference/tensorrt/op_teller.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,7 +29,6 @@ struct SimpleOpTypeSetTeller : public Teller {
teller_set.insert("fused_embedding_eltwise_layernorm");
teller_set.insert("multihead_matmul");
teller_set.insert("skip_layernorm");
teller_set.insert("slice");
#endif
}

Expand Down
2 changes: 1 addition & 1 deletion paddle/fluid/inference/tests/api/trt_dynamic_shape_ernie_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -120,7 +120,7 @@ void trt_ernie(bool with_fp16, std::vector<float> result) {
if (with_fp16) {
precision = AnalysisConfig::Precision::kHalf;
}
config.EnableTensorRtEngine(1 << 30, 1, 1, precision, false, true);
config.EnableTensorRtEngine(1 << 30, 1, 5, precision, false, true);
config.SetTRTDynamicShapeInfo(min_input_shape, max_input_shape,
opt_input_shape);
std::vector<float> out_data;
Expand Down