Python wrapper for Nvidia CUDA parallel computation API with object cleanup, automatic error checking, and convenient abstractions.
{
"context": "This evaluation assesses how well the engineer uses PyCUDA's module linking capabilities to compile CUDA source code to PTX intermediate format, link multiple PTX modules together, and execute functions from the linked modules.",
"type": "weighted_checklist",
"checklist": [
{
"name": "PTX compilation",
"description": "Uses pycuda.compiler.compile() function with target='ptx' to compile CUDA source code into PTX intermediate representation",
"max_score": 30
},
{
"name": "Linker creation",
"description": "Creates a pycuda.driver.Linker object to perform module linking operations",
"max_score": 20
},
{
"name": "PTX addition",
"description": "Uses Linker.add_data() or Linker.add_ptx() method to add PTX modules to the linker",
"max_score": 20
},
{
"name": "Module linking",
"description": "Calls Linker.link_module() to link the added PTX modules into a single executable module",
"max_score": 15
},
{
"name": "Function retrieval",
"description": "Uses the linked module's get_function() method to retrieve kernel functions by name",
"max_score": 15
}
]
}tessl i tessl/pypi-pycuda@2025.1.0docs
evals
scenario-1
scenario-2
scenario-3
scenario-4
scenario-5
scenario-6
scenario-7
scenario-8
scenario-9
scenario-10