Qualcomm® AI HubAI Hub

Profile Job Results

Jobs
jp4vd91vp
Results Ready
Name
whisper_base_float_HfWhisperDecoder
Target Device
  • QCS8550 (Proxy)
  • Android 12
  • Qualcomm® QCS8550
Creator
ai-hub-support@qti.qualcomm.com
Target Model
Input Specs
input_ids: int32[1, 1]
position_ids: int32[1]
k_cache_self_0_in: float32[8, 1, 64, 199]
v_cache_self_0_in: float32[8, 1, 199, 64]
attention_mask: float32[1, 1, 1, 200]
k_cache_cross_0: float32[8, 1, 64, 1500]
v_cache_cross_0: float32[8, 1, 1500, 64]
k_cache_self_1_in: float32[8, 1, 64, 199]
v_cache_self_1_in: float32[8, 1, 199, 64]
k_cache_cross_1: float32[8, 1, 64, 1500]
v_cache_cross_1: float32[8, 1, 1500, 64]
k_cache_self_2_in: float32[8, 1, 64, 199]
v_cache_self_2_in: float32[8, 1, 199, 64]
k_cache_cross_2: float32[8, 1, 64, 1500]
v_cache_cross_2: float32[8, 1, 1500, 64]
k_cache_self_3_in: float32[8, 1, 64, 199]
v_cache_self_3_in: float32[8, 1, 199, 64]
k_cache_cross_3: float32[8, 1, 64, 1500]
v_cache_cross_3: float32[8, 1, 1500, 64]
k_cache_self_4_in: float32[8, 1, 64, 199]
v_cache_self_4_in: float32[8, 1, 199, 64]
k_cache_cross_4: float32[8, 1, 64, 1500]
v_cache_cross_4: float32[8, 1, 1500, 64]
k_cache_self_5_in: float32[8, 1, 64, 199]
v_cache_self_5_in: float32[8, 1, 199, 64]
k_cache_cross_5: float32[8, 1, 64, 1500]
v_cache_cross_5: float32[8, 1, 1500, 64]
Completion Time
10/25/2025, 2:18:48 PM
Options
--qairt_version latest
Versions
  • QAIRT: v2.39.0.250925215840_163802
  • QNN Backend API: 5.39.0
  • QNN Core API: 2.29.0
  • Android: 14 (UP1A.231005.007)
  • AI Hub: aihub-2025.10.21.0
Estimated Inference Time
5.98 ms
Estimated Peak Memory Usage
40 ‑ 43 MB
Compute Units
NPU
820
StageTimeMemory
First App Load
236 ms6‑7 MB
Subsequent App Load
211 ms1‑2 MB
Inference
5.98 ms40‑43 MB
QNNValue
context_options.htp_options.performance_modeBURST
default_graph_options.htp_options.optimizations[0].typeFINALIZE_OPTIMIZATION_FLAG
default_graph_options.htp_options.optimizations[0].value3.0
default_graph_options.htp_options.precisionFLOAT16
default_graph_options.htp_options.vtcm_size0

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

Run on device