intel / pti-gpu

Profiling Tools Interfaces for GPU (PTI for GPU) is a set of Getting Started Documentation and Tools Library to start performance analysis on Intel(R) Processor Graphics easily
MIT License
202 stars 57 forks source link

Assertion `instance->device_sync <= queued' failed #41

Closed alyashev closed 1 year ago

alyashev commented 1 year ago

run: ~/src/pti-gpu/tools/onetrace/build/onetrace --chrome-call-logging --chrome-device-timeline python -u ds_scripts/bloom/bloom_ds_inference.py --name bigscience/bloom-560m --dtype=float16 --benchmark --ki --greedy message: python: /home/alyashev/src/pti-gpu/tools/onetrace/../cl_tracer/cl_kernel_collector.h:417: static void ClKernelCollector::ComputeHostTimestamps(const ClKernelInstance*, cl_ulong, cl_ulong, uint64_t&, uint64_t&, uint64_t&, uint64_t&): Assertion `instance->device_sync <= queued' failed.

vladimir-tsymbal commented 1 year ago

With the latest driver onetrace is not asserting anymore.