annoy-gpu
This project is derived from Annoy. The original project can use multi-thread to accelerate build process. In this project GPU is used to accelerate the build process. This project is still under development. Currently it only support the 'angular' metrics.
Install
First set the environment variable "CUDAHOME" to CUDA installation location, e.g. export CUDAHOME=/usr/local/cuda-10.2, then run pip install annoy_gpu to pull down the latest version from PyPI.
Python code example
Add items and then save at 'test-1e5.tree'.
from annoy_gpu import AnnoyIndex
import random, time
n, f = 100000, 768
t = AnnoyIndex(f, 'angular')
t.fill_items('test-1e5.tree')
for i in xrange(n):
print(f"{i} / {n}")
v = []
for z in xrange(f):
v.append(random.gauss(0, 1))
t.add_item(i, v)
t.save_items()Load the saved items 'test-1e5.tree' and build 5 index trees, then perform precision test.
q = AnnoyIndex(f, 'angular')
q.load_items('test-1e5.tree')
q.build(5)
limits = [10, 100, 1000, 10000]
k = 10
prec_sum = {}
prec_n = 10
time_sum = {}
for i in xrange(prec_n):
j = random.randrange(0, n)
closest = set(q.get_nns_by_item(j, k, n))
for limit in limits:
t0 = time.time()
toplist = q.get_nns_by_item(j, k, limit)
T = time.time() - t0
found = len(closest.intersection(toplist))
hitrate = 1.0 * found / k
prec_sum[limit] = prec_sum.get(limit, 0.0) + hitrate
time_sum[limit] = time_sum.get(limit, 0.0) + T
for limit in limits:
print('limit: %-9d precision: %6.2f%% avg time: %.6fs'
% (limit, 100.0 * prec_sum[limit] / (i + 1),
time_sum[limit] / (i + 1)))Currently it only allows you to build on disk.
