raze-gles/polymer/eduke32/source/lunatic/test/test_geom.lua

86 lines
2 KiB
Lua
Raw Normal View History

#!/usr/bin/env luajit
local os = require("os")
local xmath = require("xmath")
-- XXX: perf regression? (See below PERF_REGRESSION)
-- No, happens only with Clang build. (Why?)
local N = os.exit and (arg[1] and tostring(arg[1])) or 1e5 --1e6
local A,B = {}, {}
local V,W = {}, {}
local randvec
local args = { ... }
local ourname = args[1]
if (os.exit) then
local math = require("math")
randvec = function()
return xmath.vec2(math.random(), math.random())
end
print("Running stand-alone. ourname: "..tostring(ourname))
else
local randgen = require("randgen")
local s = randgen.new(true)
-- NOTE: factoring out the inner s:getdbl() into a separate function
-- reduces performance seriously (about an order of magnitude!)
randvec = function()
return xmath.vec2(s:getdbl(), s:getdbl())
end
-- Test optional arguments from our_require().
printf("Running %s embedded with opt arg %s", ourname, tostring(args[2]))
end
local t1 = os.clock()
if (os.exit == nil) then
local randgen = require("randgen")
local r = randgen.new(true)
for i=1,4*2*N do
-- This is to test the performance compared to a direct
-- ffiC.rand_jkiss_dbl() call in randgen.lua
r:getdbl()
end
end
local t2 = os.clock()
-- init random points and vectors
for i=1,N do
A[i] = randvec()
B[i] = randvec()
V[i] = randvec()
W[i] = randvec()
end
local t3 = os.clock()
local v = xmath.vec2(0, 0)
for i=1,N do
local intersp = xmath.intersect(A[i],V[i], B[i],W[i], true)
if (intersp ~= nil) then
v = v + intersp
end
end
local t4 = os.clock()
-- x86_64 (embedded): approx. 200 ms (vs. the 100 ms of direct
-- ffiC.rand_jkiss_dbl()):
-- x86: 170 ms
print("getdbl: ".. 1000*(t2-t1))
print("genpoints: ".. 1000*(t3-t2)) -- x86_64: 500 ms, x86: 700 ms
print("intersect: ".. 1000*(t4-t3)) -- x86_64, x86: about 35 ms <- thanks to allocation sinking (else, about 500 ms?)
print("result: ".. tostring(v))
-- PERF_REGRESSION: with N==1e6 getdbl, genpoints now about 1000ms from EDuke32!