add llvm 11.0, llvmlite and numba

[wolfv]

[wolfv]

Test script:

from ctypes import CFUNCTYPE, c_double

import llvmlite.binding as llvm


# All these initializations are required for code generation!
llvm.initialize()
llvm.initialize_all_targets()
llvm.initialize_all_asmprinters()  # yes, even this one

llvm_ir = """
   ; ModuleID = "examples/ir_fpadd.py"
   target triple = "wasm32"
   target datalayout = ""

   define double @"fpadd"(double %".1", double %".2")
   {
   entry:
     %"res" = fadd double %".1", %".2"
     ret double %"res"
   }
   """

def create_execution_engine():
    """
    Create an ExecutionEngine suitable for JIT code generation on
    the host CPU.  The engine is reusable for an arbitrary number of
    modules.
    """
    # Create a target machine representing the host
    target = llvm.Target.from_triple("wasm32")
    target_machine = target.create_target_machine()
    # And an execution engine with an empty backing module
    backing_mod = llvm.parse_assembly("")
    engine = llvm.create_mcjit_compiler(backing_mod, target_machine)
    return engine


def compile_ir(engine, llvm_ir):
    """
    Compile the LLVM IR string with the given engine.
    The compiled module object is returned.
    """
    # Create a LLVM module object from the IR
    mod = llvm.parse_assembly(llvm_ir)
    mod.verify()
    # Now add the module and make sure it is ready for execution
#     engine.add_module(mod)
#     engine.finalize_object()
#     engine.run_static_constructors()
    return mod


engine = create_execution_engine()
mod = compile_ir(engine, llvm_ir)

# Look up the function pointer (a Python int)
# func_ptr = engine.get_function_address("fpadd")

# # Run the function via ctypes
# cfunc = CFUNCTYPE(c_double, c_double, c_double)(func_ptr)
# res = cfunc(1.0, 3.5)
# print("fpadd(...) =", res)

target = llvm.Target.from_triple("wasm32")
target_machine = target.create_target_machine()
asm = str(target_machine.emit_assembly(mod))
print(asm)

[DerThorsten]

Test script:

from ctypes import CFUNCTYPE, c_double

import llvmlite.binding as llvm


# All these initializations are required for code generation!
llvm.initialize()
llvm.initialize_all_targets()
llvm.initialize_all_asmprinters()  # yes, even this one

llvm_ir = """
   ; ModuleID = "examples/ir_fpadd.py"
   target triple = "wasm32"
   target datalayout = ""

   define double @"fpadd"(double %".1", double %".2")
   {
   entry:
     %"res" = fadd double %".1", %".2"
     ret double %"res"
   }
   """

def create_execution_engine():
    """
    Create an ExecutionEngine suitable for JIT code generation on
    the host CPU.  The engine is reusable for an arbitrary number of
    modules.
    """
    # Create a target machine representing the host
    target = llvm.Target.from_triple("wasm32")
    target_machine = target.create_target_machine()
    # And an execution engine with an empty backing module
    backing_mod = llvm.parse_assembly("")
    engine = llvm.create_mcjit_compiler(backing_mod, target_machine)
    return engine


def compile_ir(engine, llvm_ir):
    """
    Compile the LLVM IR string with the given engine.
    The compiled module object is returned.
    """
    # Create a LLVM module object from the IR
    mod = llvm.parse_assembly(llvm_ir)
    mod.verify()
    # Now add the module and make sure it is ready for execution
#     engine.add_module(mod)
#     engine.finalize_object()
#     engine.run_static_constructors()
    return mod


engine = create_execution_engine()
mod = compile_ir(engine, llvm_ir)

# Look up the function pointer (a Python int)
# func_ptr = engine.get_function_address("fpadd")

# # Run the function via ctypes
# cfunc = CFUNCTYPE(c_double, c_double, c_double)(func_ptr)
# res = cfunc(1.0, 3.5)
# print("fpadd(...) =", res)

target = llvm.Target.from_triple("wasm32")
target_machine = target.create_target_machine()
asm = str(target_machine.emit_assembly(mod))
print(asm)

@wolfv can we make this an actual pytest test?

[DerThorsten]

@wolfv good pkg to check if the numba stuff works: https://pypi.org/project/umap-learn/ =)

[wolfv]

of course, but I just wanted to keep it there to not loose this valuable content :)

[jbednar]

Datashader is also a great test case for these packages when they are available. We are eagerly awaiting trying out Datashader standalone in the browser!

[twiecki]

Is this blocked by anything? Would be awesome to get numba support into pyodide.

[jbednar]

@DerThorsten , can you provide some context here? Is this being closed for lack of progress, or for some other reason?

[DerThorsten]

@DerThorsten , can you provide some context here? Is this being closed for lack of progress, or for some other reason?

Lack of progress, but I'll reopen it in the hope someone wants to continue

[DerThorsten]

this needs to be converted to the new rattler-recipe format