From 620076b76aceae6bee1ec3862b1b8bdc3d479721 Mon Sep 17 00:00:00 2001 From: aipaes <82140963+aipaes@users.noreply.github.com> Date: Fri, 6 Mar 2026 11:42:32 +0800 Subject: [PATCH] [bugs] fix install FIA sh (#6989) ### What this PR does / why we need it? Update the replacement shell script for the FIA operator FD feature in CANN 8.5.1 - vLLM version: v0.16.0 - vLLM main: https://github.com/vllm-project/vllm/commit/15d76f74e2fdb12a95ea00f0ca283acf6219a2b7 --------- Signed-off-by: zjks98 Signed-off-by: aipaes <82140963+aipaes@users.noreply.github.com> Co-authored-by: zjks98 --- tools/install_flash_infer_attention_score_ops_a2.sh | 10 ++++++---- tools/install_flash_infer_attention_score_ops_a3.sh | 10 ++++++---- 2 files changed, 12 insertions(+), 8 deletions(-) diff --git a/tools/install_flash_infer_attention_score_ops_a2.sh b/tools/install_flash_infer_attention_score_ops_a2.sh index 96cb8659..2f030e0b 100644 --- a/tools/install_flash_infer_attention_score_ops_a2.sh +++ b/tools/install_flash_infer_attention_score_ops_a2.sh @@ -22,16 +22,18 @@ set -euo pipefail trap 'echo "Error on line $LINENO: command \`$BASH_COMMAND\` failed with exit code $?" >&2' ERR cd /vllm-workspace +mkdir -p fused_infer_attention_score_a2_$(uname -i) # download fused_infer_attention_score related source files -wget https://vllm-ascend.obs.cn-north-4.myhuaweicloud.com/vllm-ascend/cann-8.5.0/fused_infer_attention_score_a2_$(uname -i).tar.gz -tar -zxvf ./fused_infer_attention_score_a2_$(uname -i).tar.gz +wget https://vllm-ascend.obs.cn-north-4.myhuaweicloud.com/vllm-ascend/cann-8.5.1/fused_infer_attention_score_a2_$(uname -i).tar.gz +tar -zxvf ./fused_infer_attention_score_a2_$(uname -i).tar.gz -C ./fused_infer_attention_score_a2_$(uname -i) # replace fused_infer_attention_score operation files -cd $ASCEND_TOOLKIT_HOME/opp/built-in/op_impl/ai_core/tbe/kernel/ascend910b +cd $ASCEND_TOOLKIT_HOME/opp/built-in/op_impl/ai_core/tbe/kernel/ascend910b/ops_transformer rm -rf fused_infer_attention_score cp -r /vllm-workspace/fused_infer_attention_score_a2_$(uname -i)/fused_infer_attention_score . # replace related so cd $ASCEND_TOOLKIT_HOME/opp/built-in/op_impl/ai_core/tbe/op_tiling/lib/linux/$(uname -i) -rm libopmaster_ct.so libopmaster_rt2.0.so liboptiling.so +rm libopmaster_*.so +rm liboptiling.so cp /vllm-workspace/fused_infer_attention_score_a2_$(uname -i)/*.so . diff --git a/tools/install_flash_infer_attention_score_ops_a3.sh b/tools/install_flash_infer_attention_score_ops_a3.sh index 95bedf5a..e9a0be2c 100644 --- a/tools/install_flash_infer_attention_score_ops_a3.sh +++ b/tools/install_flash_infer_attention_score_ops_a3.sh @@ -21,16 +21,18 @@ set -euo pipefail trap 'echo "Error on line $LINENO: command \`$BASH_COMMAND\` failed with exit code $?" >&2' ERR cd /vllm-workspace +mkdir -p fused_infer_attention_score_a3_$(uname -i) # download fused_infer_attention_score related source files -wget https://vllm-ascend.obs.cn-north-4.myhuaweicloud.com/vllm-ascend/cann-8.5.0/fused_infer_attention_score_a3_$(uname -i).tar.gz -tar -zxvf ./fused_infer_attention_score_a3_$(uname -i).tar.gz +wget https://vllm-ascend.obs.cn-north-4.myhuaweicloud.com/vllm-ascend/cann-8.5.1/fused_infer_attention_score_a3_$(uname -i).tar.gz +tar -zxvf ./fused_infer_attention_score_a3_$(uname -i).tar.gz -C ./fused_infer_attention_score_a3_$(uname -i) # replace fused_infer_attention_score operation files -cd $ASCEND_TOOLKIT_HOME/opp/built-in/op_impl/ai_core/tbe/kernel/ascend910_93 +cd $ASCEND_TOOLKIT_HOME/opp/built-in/op_impl/ai_core/tbe/kernel/ascend910_93/ops_transformer rm -rf fused_infer_attention_score cp -r /vllm-workspace/fused_infer_attention_score_a3_$(uname -i)/fused_infer_attention_score . # replace related so cd $ASCEND_TOOLKIT_HOME/opp/built-in/op_impl/ai_core/tbe/op_tiling/lib/linux/$(uname -i) -rm libopmaster_ct.so libopmaster_rt2.0.so liboptiling.so +rm libopmaster_*.so +rm liboptiling.so cp /vllm-workspace/fused_infer_attention_score_a3_$(uname -i)/*.so .