Skip to content

Remove Triton cache manager patch #3523

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

Open
janEbert wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Remove Triton cache manager patch #3523

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
1 change: 0 additions & 1 deletion examples/mamba/run_text_gen_server_8b.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,6 @@ export NCCL_IB_TIMEOUT=19
export NCCL_IB_QPS_PER_CONNECTION=4

export TRITON_CACHE_DIR="./triton-cache/"
export TRITON_CACHE_MANAGER="megatron.core.ssm.triton_cache_manager:ParallelFileCacheManager"

torchrun $DISTRIBUTED_ARGS ../../tools/run_mamba_text_generation_server.py \
--tensor-model-parallel-size 1 \
Expand Down
1 change: 0 additions & 1 deletion examples/mamba/train.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,6 @@ mkdir -p ${DATACACHE_DIR}
mkdir -p ${TENSORBOARD_DIR}

export TRITON_CACHE_DIR="./triton-cache/"
export TRITON_CACHE_MANAGER="megatron.core.ssm.triton_cache_manager:ParallelFileCacheManager"

SEQ_LEN=4096
TRAIN_SAMPLES=73242188 # 300B tokens / 4096
Expand Down
4 changes: 1 addition & 3 deletions examples/multimodal/pretrain_mistral_clip.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,8 +20,6 @@ LOGS_DIR="${OUTPUT}/logs"
TENSORBOARD_DIR="${OUTPUT}/tensorboard"

export TRITON_CACHE_DIR="${WORKSPACE}/triton-cache/"
# The following patch to the Triton cache manager is needed for Triton version <= 3.1
export TRITON_CACHE_MANAGER="megatron.core.ssm.triton_cache_manager:ParallelFileCacheManager"

if [[ -z $LOAD_NAME ]]; then
echo "Please set LOAD_NAME for input model name."
Expand Down Expand Up @@ -129,4 +127,4 @@ OPTIONS=" \
export NVTE_APPLY_QK_LAYER_SCALING=0
export NVTE_ALLOW_NONDETERMINISTIC_ALGO=${NONDETERMINISTIC_ATTN}

torchrun --nproc_per_node 8 examples/multimodal/train.py ${OPTIONS}
torchrun --nproc_per_node 8 examples/multimodal/train.py ${OPTIONS}
2 changes: 0 additions & 2 deletions examples/multimodal/sft_mistral_clip.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,8 +20,6 @@ LOGS_DIR="${OUTPUT}/logs"
TENSORBOARD_DIR="${OUTPUT}/tensorboard"

export TRITON_CACHE_DIR="${WORKSPACE}/triton-cache/"
# The following patch to the Triton cache manager is needed for Triton version <= 3.1
export TRITON_CACHE_MANAGER="megatron.core.ssm.triton_cache_manager:ParallelFileCacheManager"

if [[ -z $LOAD_NAME ]]; then
echo "Please set LOAD_NAME for input model name."
Expand Down
81 changes: 0 additions & 81 deletions megatron/core/ssm/triton_cache_manager.py
View file
Open in desktop

This file was deleted.

Loading