1# 2# Copyright (C) 2016 Intel Corporation 3# 4# Permission is hereby granted, free of charge, to any person obtaining a 5# copy of this software and associated documentation files (the "Software"), 6# to deal in the Software without restriction, including without limitation 7# the rights to use, copy, modify, merge, publish, distribute, sublicense, 8# and/or sell copies of the Software, and to permit persons to whom the 9# Software is furnished to do so, subject to the following conditions: 10# 11# The above copyright notice and this permission notice (including the next 12# paragraph) shall be included in all copies or substantial portions of the 13# Software. 14# 15# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL 18# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING 20# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS 21# IN THE SOFTWARE. 22 23# Prior to Kaby Lake, The SIN and COS instructions on Intel hardware can 24# produce values slightly outside of the [-1.0, 1.0] range for a small set of 25# values. Obviously, this can break everyone's expectations about trig 26# functions. This appears to be fixed in Kaby Lake. 27# 28# According to an internal presentation, the COS instruction can produce 29# a value up to 1.000027 for inputs in the range (0.08296, 0.09888). One 30# suggested workaround is to multiply by 0.99997, scaling down the 31# amplitude slightly. Apparently this also minimizes the error function, 32# reducing the maximum error from 0.00006 to about 0.00003. 33 34import argparse 35import sys 36from math import pi 37 38TRIG_WORKAROUNDS = [ 39 (('fsin', 'x(is_not_const)'), ('fmul', ('fsin', 'x'), 0.99997)), 40 (('fcos', 'x(is_not_const)'), ('fmul', ('fcos', 'x'), 0.99997)), 41] 42 43LIMIT_TRIG_INPUT_RANGE_WORKAROUND = [ 44 (('fsin', 'x(is_not_const)'), ('fsin', ('fmod', 'x', 2.0 * pi))), 45 (('fcos', 'x(is_not_const)'), ('fcos', ('fmod', 'x', 2.0 * pi))), 46] 47 48def main(): 49 parser = argparse.ArgumentParser() 50 parser.add_argument('-p', '--import-path', required=True) 51 args = parser.parse_args() 52 sys.path.insert(0, args.import_path) 53 run() 54 55 56def run(): 57 import nir_algebraic # pylint: disable=import-error 58 59 print('#include "brw_nir.h"') 60 print(nir_algebraic.AlgebraicPass("brw_nir_apply_trig_workarounds", 61 TRIG_WORKAROUNDS).render()) 62 print(nir_algebraic.AlgebraicPass("brw_nir_limit_trig_input_range_workaround", 63 LIMIT_TRIG_INPUT_RANGE_WORKAROUND).render()) 64 65 66if __name__ == '__main__': 67 main() 68