[JS][WEB][BACKEND] Javascript(webassembly) backend.

[tqchen]

cc @Javelinjs maybe some good reference on setting up testcases

[tqchen]

Use TVM to Generate Javascript Library

The general idea is to use TVM as normally and set target to be llvm -target=asmjs-unknown-emscripten -system-lib.

The following code snippet from tests/web/prepare_test_libs.py demonstrate
the compilation process.

import tvm
from tvm.contrib import emscripten
import os
def prepare_test_libs(base_path):
    target = "llvm -target=asmjs-unknown-emscripten -system-lib"
    if not tvm.module.enabled(target):
        raise RuntimeError("Target %s is not enbaled" % target)
    n = tvm.var("n")
    A = tvm.placeholder((n,), name='A')
    B = tvm.compute(A.shape, lambda *i: A(*i) + 1.0, name='B')
    s = tvm.create_schedule(B.op)
    fadd1 = tvm.build(s, [A, B], target, name="add_one")
    obj_path = os.path.join(base_path, "test_add_one.bc")
    fadd1.save(obj_path)
    emscripten.create_js(os.path.join(base_path, "test_module.js"), obj_path)

if __name__ == "__main__":
    curr_path = os.path.dirname(os.path.abspath(os.path.expanduser(__file__)))
    prepare_test_libs(os.path.join(curr_path, "../../lib"))

In this workflow, we use TVM to generate a .bc file and statically link
that with the lib/libtvm_web_runtime.bc(emscripten.create_js will help you do that).
The result js library is a library that contains both TVM runtime and the compiled function.

Run the Generated Library

The following code snippet from tests/web/test_module_load.js demonstrate
how to run the compiled library.

// Load Emscripten Module, need to change path to root/lib
const path = require("path");
process.chdir(path.join(__dirname, "../../lib"));
var Module = require("../../lib/test_module.js");
// Bootstrap TVMruntime with emscripten module.
const tvm_runtime = require("../../web/tvm_runtime.js");
const tvm = tvm_runtime.create(Module);

// Load system library, the compiled functions is registered in sysLib.
var sysLib = tvm.systemLib();

function randomArray(length, max) {
  return Array.apply(null, Array(length)).map(function() {
    return Math.random() * max;
  });
}

function testAddOne() {
  // grab pre-loaded function
  var faddOne = sysLib.getFunction("add_one");
  tvm.assert(tvm.isPackedFunc(faddOne));
  var n = 124;
  var A = tvm.empty(n).copyFrom(randomArray(n, 1));
  var B = tvm.empty(n);
  // call the function.
  faddOne(A, B);
  // verify
  for (var i = 0; i < B.length; ++i) {
    tvm.assert(B[i] == A[i] + 1);
  }
  faddOne.release();
}

testAddOne();
sysLib.release();

[phara0hcom]

I have a question about the tests/web/prepare_test_libs.py after following all these steps:
https://github.com/dmlc/tvm/blob/master/web/README.md

I ran prepare_test_libs.py and got the following output:

Traceback (most recent call last):
  File "./tests/web/prepare_test_libs.py", line 21, in <module>
    prepare_test_libs(os.path.join(curr_path, "../../lib"))
  File "./tests/web/prepare_test_libs.py", line 9, in prepare_test_libs
    raise RuntimeError("Target %s is not enbaled" % target)
RuntimeError: Target llvm -target=asmjs-unknown-emscripten -system-lib is not enbaled

What can I do to fix this?
Can I use a different target?
If so can how can I get a list of targets?

I can make an issue out of this

[srkreddy1238]

What is the llvm version ?

try llc --version

[phara0hcom]

@srkreddy1238 this is what I get but only after I put in the following command:
source ./Documents/GitHub/emsdk/emsdk_env.sh
then I get llc as a bash command

LLVM (http://llvm.org/):
  LLVM version 6.0.1
  Optimized build.
  Default target: x86_64-apple-darwin17.7.0
  Host CPU: haswell

  Registered Targets:
    js     - JavaScript (asm.js, emscripten) backend
    x86    - 32-bit X86: Pentium-Pro and above
    x86-64 - 64-bit X86: EM64T and AMD64

else I will get an error:
-bash: llc: command not found

EDIT:

resolved the issue with bash -bash: llc: command not found by editing .bash_profile
but still have the same error as before