Qualcomm® AI HubAI Hub

Profile Job Results

Jobs
jgonyl94p
Results Ready
Name
whisper_base_HfWhisperDecoder
Target Device
  • Samsung Galaxy S25
  • Android 15
  • Snapdragon® 8 Elite for Galaxy | SM8750-AC
Creator
ai-hub-support@qti.qualcomm.com
Target Model
Input Specs
input_ids: int32[1, 1]
attention_mask: float16[1, 1, 1, 200]
k_cache_self_0_in: float16[8, 1, 64, 199]
v_cache_self_0_in: float16[8, 1, 199, 64]
k_cache_self_1_in: float16[8, 1, 64, 199]
v_cache_self_1_in: float16[8, 1, 199, 64]
k_cache_self_2_in: float16[8, 1, 64, 199]
v_cache_self_2_in: float16[8, 1, 199, 64]
k_cache_self_3_in: float16[8, 1, 64, 199]
v_cache_self_3_in: float16[8, 1, 199, 64]
k_cache_self_4_in: float16[8, 1, 64, 199]
v_cache_self_4_in: float16[8, 1, 199, 64]
k_cache_self_5_in: float16[8, 1, 64, 199]
v_cache_self_5_in: float16[8, 1, 199, 64]
k_cache_cross_0: float16[8, 1, 64, 1500]
v_cache_cross_0: float16[8, 1, 1500, 64]
k_cache_cross_1: float16[8, 1, 64, 1500]
v_cache_cross_1: float16[8, 1, 1500, 64]
k_cache_cross_2: float16[8, 1, 64, 1500]
v_cache_cross_2: float16[8, 1, 1500, 64]
k_cache_cross_3: float16[8, 1, 64, 1500]
v_cache_cross_3: float16[8, 1, 1500, 64]
k_cache_cross_4: float16[8, 1, 64, 1500]
v_cache_cross_4: float16[8, 1, 1500, 64]
k_cache_cross_5: float16[8, 1, 64, 1500]
v_cache_cross_5: float16[8, 1, 1500, 64]
position_ids: int32[1]
Completion Time
9/27/2025, 2:10:12 PM
Options
--qairt_version 2.37
Versions
  • ONNX Runtime: 1.22.2
  • QAIRT: v2.37.1.250807093845_124904
  • Android: 15 (AP3A.240905.015.A2)
  • AI Hub: aihub-2025.09.23.1
Estimated Inference Time
2.88 ms
Estimated Peak Memory Usage
0 ‑ 13 MB
Compute Units
NPU
820
StageTimeMemory
First App Load
134 ms150‑158 MB
Subsequent App Load
130 ms144‑156 MB
Inference
2.88 ms0‑13 MB
ONNX RuntimeValue
execution_modeSEQUENTIAL
intra_op_num_threads0
inter_op_num_threads0
enable_memory_patternfalse
enable_cpu_memory_arenafalse
graph_optimization_levelENABLE_ALL
QNN Execution ProviderValue
htp_performance_mode"burst"
htp_graph_finalization_optimization_mode"3"
enable_htp_fp16_precision"1"
capture_network_visualizationsfalse

Sign up to run this model on a hosted Qualcomm® device!

Run on device