#!/usr/bin/env python # Copyright 2021 Google LLC # # This source code is licensed under the BSD-style license found in the # LICENSE file in the root directory of this source tree. import argparse import codecs import math import os import re import sys import yaml sys.path.insert(0, os.path.dirname(os.path.abspath(__file__))) import xngen import xnncommon parser = argparse.ArgumentParser( description='Vector conversion operation microkernel test generator') parser.add_argument("-s", "--spec", metavar="FILE", required=True, help="Specification (YAML) file") parser.add_argument("-o", "--output", metavar="FILE", required=True, help='Output (C++ source) file') parser.set_defaults(defines=list()) def split_ukernel_name(name): match = re.fullmatch(r"xnn_(f16|f32|qs16|qs8|qu8)(_(f16|f32|qs8|qu8))?_vcvt_ukernel__(.+)_x(\d+)", name) if match is None: raise ValueError("Unexpected microkernel name: " + name) input_datatype = match.group(1) if match.group(2): output_datatype = match.group(3) else: output_datatype = input_datatype batch_tile = int(match.group(5)) arch, isa, assembly = xnncommon.parse_target_name(target_name=match.group(4)) return input_datatype, output_datatype, batch_tile, arch, isa CVT_TEST_TEMPLATE = """\ TEST(${TEST_NAME}, batch_eq_${BATCH_TILE}) { $if ISA_CHECK: ${ISA_CHECK}; VCvtMicrokernelTester() .batch_size(${BATCH_TILE}) $if INPUT_DATATYPE == "QS16": .input_zero_point(0) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } $if BATCH_TILE > 1: TEST(${TEST_NAME}, batch_div_${BATCH_TILE}) { $if ISA_CHECK: ${ISA_CHECK}; for (size_t batch_size = ${BATCH_TILE*2}; batch_size < ${BATCH_TILE*10}; batch_size += ${BATCH_TILE}) { VCvtMicrokernelTester() .batch_size(batch_size) $if INPUT_DATATYPE == "QS16": .input_zero_point(0) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } TEST(${TEST_NAME}, batch_lt_${BATCH_TILE}) { $if ISA_CHECK: ${ISA_CHECK}; for (size_t batch_size = 1; batch_size < ${BATCH_TILE}; batch_size++) { VCvtMicrokernelTester() .batch_size(batch_size) $if INPUT_DATATYPE == "QS16": .input_zero_point(0) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } TEST(${TEST_NAME}, batch_gt_${BATCH_TILE}) { $if ISA_CHECK: ${ISA_CHECK}; for (size_t batch_size = ${BATCH_TILE+1}; batch_size < ${10 if BATCH_TILE == 1 else BATCH_TILE*2}; batch_size++) { VCvtMicrokernelTester() .batch_size(batch_size) $if INPUT_DATATYPE == "QS16": .input_zero_point(0) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } $if INPUT_DATATYPE.startswith("Q") or OUTPUT_DATATYPE.startswith("Q"): TEST(${TEST_NAME}, scale) { $if ISA_CHECK: ${ISA_CHECK}; for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) .scale(50) $if INPUT_DATATYPE == "QS16": .input_zero_point(0) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .output_zero_point(100) .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } $if INPUT_DATATYPE in ["QS8", "QU8"]: TEST(${TEST_NAME}, input_zero_point) { $if ISA_CHECK: ${ISA_CHECK}; for (int16_t input_zero_point = 0; input_zero_point < 5; input_zero_point += 2) { for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) .input_zero_point(input_zero_point) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } } $if OUTPUT_DATATYPE in ["QS8", "QU8"]: TEST(${TEST_NAME}, output_zero_point) { $if ISA_CHECK: ${ISA_CHECK}; for (int16_t output_zero_point = 0; output_zero_point < 5; output_zero_point += 2) { for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) $if INPUT_DATATYPE == "QS16": .input_zero_point(0) .output_zero_point(output_zero_point) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } } $if INPUT_DATATYPE == "F32": TEST(${TEST_NAME}, saturation) { $if ISA_CHECK: ${ISA_CHECK}; for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) .scale(500) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .output_zero_point(128) .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } TEST(${TEST_NAME}, overflow) { $if ISA_CHECK: ${ISA_CHECK}; for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) .scale(4294967296.0f) $if INPUT_DATATYPE == "QS16": .input_zero_point(0) $if OUTPUT_DATATYPE == "QS8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) $elif OUTPUT_DATATYPE == "QU8": .qmin(std::numeric_limits::min()) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } $if INPUT_DATATYPE == "F32" and OUTPUT_DATATYPE == "QS8": TEST(${TEST_NAME}, qmin) { $if ISA_CHECK: ${ISA_CHECK}; for (int16_t qmin = -128; qmin < 127; qmin += 51) { for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) .scale(500) .qmin(qmin) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } } TEST(${TEST_NAME}, qmax) { $if ISA_CHECK: ${ISA_CHECK}; for (int16_t qmax = -127; qmax <= 127; qmax += 51) { for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) .scale(500) $if INPUT_DATATYPE == "QS16": .input_zero_point(0) .qmin(std::numeric_limits::min()) .qmax(qmax) .Test(${", ".join(TEST_ARGS)}); } } } $if INPUT_DATATYPE == "F32" and OUTPUT_DATATYPE == "QU8": TEST(${TEST_NAME}, qmin) { $if ISA_CHECK: ${ISA_CHECK}; for (int16_t qmin = 0; qmin < 255; qmin += 51) { for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) .scale(500) .output_zero_point(128) .qmin(qmin) .qmax(std::numeric_limits::max()) .Test(${", ".join(TEST_ARGS)}); } } } TEST(${TEST_NAME}, qmax) { $if ISA_CHECK: ${ISA_CHECK}; for (int16_t qmax = 1; qmax <= 255; qmax += 51) { for (size_t batch_size = 1; batch_size <= ${BATCH_TILE*5}; batch_size += ${max(1, BATCH_TILE-1)}) { VCvtMicrokernelTester() .batch_size(batch_size) .scale(500) .output_zero_point(128) .qmin(std::numeric_limits::min()) .qmax(qmax) .Test(${", ".join(TEST_ARGS)}); } } } """ def generate_test_cases(ukernel, init_fn, input_datatype, output_datatype, batch_tile, isa): """Generates all tests cases for a Vector Convert Operation micro-kernel. Args: ukernel: C name of the micro-kernel function. init_fn: C name of the function to initialize microkernel parameters. input_datatype: input conversion data type. output_datatype: output conversion data type. batch_tile: Number of batch elements processed per one iteration of the inner loop of the micro-kernel. isa: instruction set required to run the micro-kernel. Generated unit test will skip execution if the host processor doesn't support this ISA. Returns: Code for the test case. """ _, test_name = ukernel.split("_", 1) test_args = [ukernel] if init_fn: test_args.append(init_fn) return xngen.preprocess(CVT_TEST_TEMPLATE, { "TEST_NAME": test_name.upper().replace("UKERNEL_", ""), "TEST_ARGS": test_args, "BATCH_TILE": batch_tile, "INPUT_DATATYPE": input_datatype.upper(), "OUTPUT_DATATYPE": output_datatype.upper(), "ISA_CHECK": xnncommon.generate_isa_check_macro(isa), }) def main(args): options = parser.parse_args(args) with codecs.open(options.spec, "r", encoding="utf-8") as spec_file: spec_yaml = yaml.safe_load(spec_file) if not isinstance(spec_yaml, list): raise ValueError("expected a list of micro-kernels in the spec") tests = """\ // Copyright 2021 Google LLC // // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. // // Auto-generated file. Do not edit! // Specification: {specification} // Generator: {generator} #include #include #include #include #include #include "vcvt-microkernel-tester.h" """.format(specification=options.spec, generator=sys.argv[0]) for ukernel_spec in spec_yaml: name = ukernel_spec["name"] init_fn = ukernel_spec.get("init") input_datatype, output_datatype, batch_tile, arch, isa = \ split_ukernel_name(name) test_case = generate_test_cases( name, init_fn, input_datatype, output_datatype, batch_tile, isa) tests += "\n\n" + xnncommon.postprocess_test_case(test_case, arch, isa) txt_changed = True if os.path.exists(options.output): with codecs.open(options.output, "r", encoding="utf-8") as output_file: txt_changed = output_file.read() != tests if txt_changed: with codecs.open(options.output, "w", encoding="utf-8") as output_file: output_file.write(tests) if __name__ == "__main__": main(sys.argv[1:])