← all publishers

dev-ansh-r

@dev-ansh-r source repo

Report this profile Claim this profile

• 8 skills
• 0 followers
• 5 hours ago last updated

1. ▌
   Qualcomm Sdk Preflight · dev-ansh-r

   Verify a host can run the QAIRT/QNN toolchain at all (it is x86_64 Linux only - not Windows or macOS) and that the SDK install is complete and usable before converting a model, and locate the documentation that matches the installed version. Checks SDK dependency scripts, HTP backend libraries, Python bindings, ONNX/ONNXRuntime versions and cross-toolchain presence, then finds the version-correct docs (SDK-local HTML first, vendor site second). Use before a first conversion on a new machine, after an SDK upgrade, or when a converter fails with import or library errors.
   0 installs
2. ▌
   Qualcomm Setup · dev-ansh-r bundle

   First-run setup for the Qualcomm skills - get started, set up, configure or onboard a new machine for QAIRT/QNN/AIMET work. Probes what is already installed, asks only what cannot be discovered, and writes the .qualcomm-env config every other Qualcomm skill reads. Covers all layouts - SDK on this machine, a remote Linux build server or cloud VM, a separate AIMET host, board over SSH or ADB or none yet. Run this before any other Qualcomm skill, or when one reports that setup has not been run.
   0 installs
3. ▌
   Qnn Model Export · dev-ansh-r bundle

   Convert an ONNX model to a QNN model library for Qualcomm HTP using the classic flow - static-shape preparation, calibration input_list generation, qnn-onnx-converter with W8A16 quantization, then qnn-model-lib-generator to produce an aarch64 .so, validated with qnn-net-run. Accepts AIMET .encodings via --quantization_overrides alongside --input_list. Use for Qualcomm/Hexagon HTP model conversion, whether a model will run on the NPU, FP16 or float_fallback errors, or any qnn-onnx-converter, model-lib-generator, input_list, bias_bw or W8A16 question. The DLC alternative is qnn-context-binary.
   0 installs
4. ▌
   Qualcomm Sdk Docs · dev-ansh-r bundle

   Extract version-correct reference documentation from an installed QAIRT/QNN SDK - tool flag lists from --help, operator support tables from the SDK HTML docs, and SoC/Hexagon HTP architecture identifiers - into a local greppable cache, then check an ONNX model's operators against it. Use to answer whether a flag exists in your SDK version, whether a model will run on the NPU, which operators a backend supports, what HTP architecture a Qualcomm part needs, or to find Qualcomm documentation for any QCS, QCM, SA, SM or Snapdragon part without a developer account.
   0 installs
5. ▌
   Aimet Quantization · dev-ansh-r

   Quantize an ONNX model for Qualcomm HTP with AIMET - QuantizationSimModel (quantsim) calibration, AdaRound adaptive weight rounding, cross-layer equalization, bias correction and automatic mixed precision - exporting a .encodings file consumed by qnn-onnx-converter or qairt-converter via --quantization_overrides. Use for post-training quantization (PTQ), W8A16/W8A8 accuracy recovery, per-layer quantization sensitivity analysis, or when a converted model runs but its accuracy dropped. Runs on the AIMET host, not the board.
   0 installs
6. ▌
   Qnn Context Binary · dev-ansh-r bundle

   Build a pre-compiled QNN HTP context binary for Qualcomm QCS6490 using the QAIRT/DLC flow - qairt-converter with AIMET quantization_overrides, qairt-quantizer, then qnn-context-binary-generator - and deploy and validate it on the board. Use for production deployment, fastest model load time, applying AIMET .encodings, or any qairt-converter, qairt-quantizer, DLC or context binary question. The classic alternative is qnn-model-export, which reaches a context binary via a cross-compiled .so built on the board instead.
   0 installs
7. ▌
   Qualcomm Cross Compile · dev-ansh-r

   Cross-compile a C/C++ or ROS2 application for a Qualcomm QCS6490 aarch64 board using the Yocto/QIRP eSDK - environment setup order, CMake and colcon invocation, linking the QNN runtime, and the silent-failure modes that exit 0 while producing nothing. Use when the target board has an immutable or OSTree image with no on-device compiler, when a build succeeds but produces no binary, or for toolchain, sysroot, eSDK or aarch64 linking questions.
   0 installs
8. ▌
   Qualcomm Env Discovery · dev-ansh-r bundle

   Discover and verify a Qualcomm development environment for any Snapdragon or QCS/QCM/SA part - QCS6490, QCS8550, QCS8250, RB5, SA8295, Snapdragon X and others. Locates the target board, build host and AIMET host, confirms reachability, and reads back the installed QAIRT/QNN SDK version, eSDK toolchain, SoC identity, Hexagon HTP architecture and board image. Writes a .qualcomm-env config the other Qualcomm skills read. Use at the start of any QNN/QAIRT/AIMET/HTP/NPU task, when a converter or board command fails for unclear reasons, or when asked which SDK version or HTP architecture is installed.
   0 installs