cgarciae

defmodule Streamer do
  def async(enum) do
    Stream.resource(
      #Start
      fn ->
        origin = self
        spawn fn ->
          for x <- enum do
            send origin, {:elem, x}
          end

cgarciae

cudnn install ubuntu + tensorflow

1. Install cuda
2. Export variables in ~/.bashrc or ~/.zshrc.

export PATH=/usr/local/cuda/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export CUDA_HOME=/usr/local/cuda

2. Download cudnn. Note: check TensorFlow install guide for the correct version.
3. Extract cudnn, this extracts a folder named cuda, cd into it.

cgarciae

# client-async-as-completed.py

from aiohttp import ClientSession, TCPConnector
import asyncio
from itertools import islice
import sys

def limited_as_completed(coros, limit):
    futures = [
        asyncio.ensure_future(c)

cgarciae

# server.py

from aiohttp import web
import asyncio
import random

async def handle(request):
    await asyncio.sleep(random.randint(0, 3))
    return web.Response(text="Hello, World!")

cgarciae

nvidia-docker with docker-compose

Setup

1. Install nvidia-docker-plugin
2. Reboot (for some reason)
3. On the folder with this docker-compose.yml file do

docker-compose up

cgarciae

import asyncio

class TaskPool(object):

    def __init__(self, workers):
        self._semaphore = asyncio.Semaphore(workers)
        self._tasks = set()
        

    async def put(self, coro):

cgarciae

using Base.Threads
using Distributions
using BenchmarkTools

ENV["PYCALL_JL_RUNTIME_PYTHON"] = Sys.which("python")

using PyCall

py"""
import sys

cgarciae

import typing as tp

from jax import numpy as jnp
import jax
import numpy as np
import time

@jax.jit
def _distances_jax(data1, data2):

cgarciae

using Base.Threads
using LoopVectorization
using BenchmarkTools

const None = [CartesianIndex()] 

function distances(data1, data2)
    data1 = deg2rad.(data1)
    data2 = deg2rad.(data2)
    lat1 = @view data1[:, 1]

cgarciae

def get_model(params) -> tf.keras.Model:

    x0 = tf.keras.Input(shape=(1,), name="x0")
    x1 = tf.keras.Input(shape=(1,), name="x1")

    inputs = [x0, x1]

    # x0 embeddings