getting time out error

General
1

-mapping validation: 1
Reverts on split failed: 0
18:34
Reverts on cluster mapping: 0
Reverts on inter-cluster connectivity: 3
Reverts on pre-mapping validation: 1
Reverts on split failed: 0
[info] Iterations: 12
Reverts on cluster mapping: 0
Reverts on inter-cluster connectivity: 3
Reverts on pre-mapping validation: 1
Reverts on split failed: 0
[error] Mapping Failed (allocation time: 19m 30s)
[info] Running Auto-Merger
[info] Auto-Merger is done
[info] Running Auto-Merger
[info] Auto-Merger is done
[info] Running Auto-Merger
[info] Auto-Merger is done
[info] Using Single-context flow
[info] Resources optimization guidelines: Strategy → GREEDY Objective → MAX_FPS
[info] Resources optimization params: max_control_utilization=117.5%, max_compute_utilization=100%, max_compute_16bit_utilization=100%, max_memory_utilization (weights)=90%, max_input_aligner_utilization=100%, max_apu_utilization=100%
[info] Using Single-context flow
[info] Resources optimization guidelines: Strategy → GREEDY Objective → MAX_FPS
[info] Resources optimization params: max_control_utilization=117.5%, max_compute_utilization=100%, max_compute_16bit_utilization=100%, max_memory_utilization (weights)=90%, max_input_aligner_utilization=100%, max_apu_utilization=100%
i keep getting this time out even after setting the normalization1 = normalization([0.0, 0.0, 0.0], [255.0, 255.0, 255.0])

change_output_activation(conv42, sigmoid)
change_output_activation(conv53, sigmoid)
change_output_activation(conv63, sigmoid)

quantization_param([conv42, conv53, conv63], force_range_out=[0.0, 1.0])

model_optimization_flavor(optimization_level=1, compression_level=0)

performance_param(compiler_optimization_level=max)

nms_postprocess(“../../../postprocess_config/yolov8n_nms_config.json”, meta_arch=yolov8, engine=cpu)

allocator_param(timeout=100h) . i keep getting time out error even after setting to 100h. what should i do about this?

2

Hi @shaahidh_shaji
If you need help compiling YOLO models, I would like to recommend a tool we developed known as the cloud compiler that compiles your checkpoint to hailo without any local setup. See Early Access to DeGirum Cloud Compiler. Please see it can help.

3

Hey @shaahidh_shaji ,

The timeout and mapping failures you’re seeing aren’t actually solved by setting allocator_param(timeout=100h). Even with that extended timeout, you’re hitting deeper architectural issues that no amount of waiting time will fix.

What’s Really Happening:

Your allocator_param(timeout=100h) just tells the compiler to keep trying for up to 100 hours before giving up. But the “Mapping Failed” errors with those reverts indicate structural problems that won’t resolve with more time:

  • 3 inter-cluster connectivity reverts: The compiler can’t efficiently route data between clusters, usually because of bandwidth limits, poor layer partitioning, or overloaded cluster resources
  • 1 pre-mapping validation revert: There are incompatible layer configurations, problematic quantization settings, or unresolvable operator fusions

The repeated Auto-Merger attempts show the compiler is trying to optimize layer merging but failing, likely due to your forced quantization ranges restricting merge flexibility.

Here’s what to try:

Fix the cluster issues:

  • Switch from single_context_flow() to multi_context_flow() for more flexible cluster assignment

Debug the problem layers:

  • Use dump_debug_graph(...) to see exactly which layers are causing the inter-cluster reverts

Rethink your quantization:

  • That quantization_param([conv42, conv53, conv63], force_range_out=[0.0, 1.0]) might be hurting more than helping
  • Try removing force_range_out and let the compiler choose optimal ranges

Tone down resource limits:

  • Your max_control_utilization=117.5% is over 100%, which can cause allocations that get rejected later
  • Try downgrading the optimization level

Hope this help !