Skip to content

Commit 1473f1a

Browse files
committed
tp support
Signed-off-by: jthomson04 <[email protected]>
1 parent 95ac241 commit 1473f1a

File tree

1 file changed

+23
-0
lines changed

1 file changed

+23
-0
lines changed

components/backends/trtllm/performance_sweeps/submit_disagg.sh

Lines changed: 23 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -419,6 +419,29 @@ main() {
419419

420420
run_single $ctx_num $ctx_tp_size $ctx_ep_size $ctx_enable_attention_dp $gen_num $gen_tp_size $gen_tp_size $gen_batch_size $gen_max_num_tokens $gen_gpu_memory_fraction $gen_mtp_size $gen_eplb_num_slots "$gen_concurrency_list"
421421
;;
422+
"tp")
423+
if [ $# -ne 14 ]; then
424+
echo "Error: TP mode requires 14 additional parameters (including mtp_mode)"
425+
usage
426+
fi
427+
428+
local ctx_num=$3
429+
local ctx_tp_size=$4
430+
local ctx_ep_size=$5
431+
local ctx_enable_attention_dp=$6
432+
local gen_num=$7
433+
local gen_tp_size=$8
434+
local gen_batch_size=$9
435+
local gen_max_num_tokens=${10}
436+
local gen_gpu_memory_fraction=${11}
437+
local gen_mtp_size=${12}
438+
local gen_eplb_num_slots=${13}
439+
local gen_concurrency_list=${14}
440+
441+
echo "Running TP mode ($mtp_mode) with ctx_num=$ctx_num, gen_num=$gen_num, gen_tp_size=$gen_tp_size, gen_ep_size=1, gen_batch_size=$gen_batch_size, gen_max_num_tokens=$gen_max_num_tokens, gen_gpu_memory_fraction=$gen_gpu_memory_fraction, gen_mtp_size=$gen_mtp_size, gen_eplb_num_slots=$gen_eplb_num_slots, gen_concurrency_list=\"$gen_concurrency_list\""
442+
443+
run_single $ctx_num $ctx_tp_size $ctx_ep_size $ctx_enable_attention_dp $gen_num $gen_tp_size 1 $gen_batch_size $gen_max_num_tokens false $gen_gpu_memory_fraction $gen_mtp_size $gen_eplb_num_slots "$gen_concurrency_list"
444+
;;
422445
*)
423446
echo "Error: Unknown mode '$mode'"
424447
usage

0 commit comments

Comments
 (0)