From 3c9769211856d5b89c1bd66f9e6999ba3834441b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20H=C3=B6nig?= <jan.hoenig@fau.de>
Date: Thu, 11 Nov 2021 18:03:31 +0100
Subject: [PATCH] All todos done so far
---
pystencils_benchmark/__init__.py | 1 +
pystencils_benchmark/benchmark.py | 139 +++++++++++-------
pystencils_benchmark/enums.py | 8 +
pystencils_benchmark/templates/Clang.mk | 19 +++
.../templates/{include_GCC.mk => GCC.mk} | 0
.../{include_GCCdebug.mk => GCCdebug.mk} | 4 +-
.../templates/{include_ICC.mk => ICC.mk} | 0
pystencils_benchmark/templates/Makefile | 21 +--
.../templates/include_Clang.mk | 24 ---
pystencils_benchmark/templates/main.c | 92 ++++++------
requirements.txt | 3 +-
test.ipynb | 130 ----------------
test.py | 45 ++++++
13 files changed, 215 insertions(+), 271 deletions(-)
create mode 100644 pystencils_benchmark/enums.py
create mode 100644 pystencils_benchmark/templates/Clang.mk
rename pystencils_benchmark/templates/{include_GCC.mk => GCC.mk} (100%)
rename pystencils_benchmark/templates/{include_GCCdebug.mk => GCCdebug.mk} (69%)
rename pystencils_benchmark/templates/{include_ICC.mk => ICC.mk} (100%)
delete mode 100644 pystencils_benchmark/templates/include_Clang.mk
delete mode 100644 test.ipynb
create mode 100755 test.py
diff --git a/pystencils_benchmark/__init__.py b/pystencils_benchmark/__init__.py
index 611cedd..6f5f32c 100644
--- a/pystencils_benchmark/__init__.py
+++ b/pystencils_benchmark/__init__.py
@@ -1 +1,2 @@
+from .enums import Compiler
from .benchmark import generate_benchmark, kernel_header, kernel_source
diff --git a/pystencils_benchmark/benchmark.py b/pystencils_benchmark/benchmark.py
index 1b4dcae..25c9044 100644
--- a/pystencils_benchmark/benchmark.py
+++ b/pystencils_benchmark/benchmark.py
@@ -1,6 +1,10 @@
+from typing import Union, List
+from collections import namedtuple
from pathlib import Path
from jinja2 import Environment, PackageLoader, StrictUndefined
+import numpy as np
+
from pystencils.backends.cbackend import generate_c, get_headers
from pystencils.astnodes import KernelFunction, PragmaBlock
from pystencils.enums import Backend
@@ -8,13 +12,13 @@ from pystencils.data_types import get_base_type
from pystencils.sympyextensions import prod
from pystencils.integer_functions import modulo_ceil
-import numpy as np
+from pystencils_benchmark.enums import Compiler
-def generate_benchmark(kernel_ast: KernelFunction,
+def generate_benchmark(kernel_asts: Union[KernelFunction, List[KernelFunction]],
path: Path = None,
*,
- dialect: Backend = Backend.C) -> None:
+ compiler: Compiler = Compiler.GCC) -> None:
if path is None:
path = Path('.')
else:
@@ -24,20 +28,39 @@ def generate_benchmark(kernel_ast: KernelFunction,
include_path = path / 'include'
include_path.mkdir(parents=True, exist_ok=True)
- kernel_name = kernel_ast.function_name
+ if isinstance(kernel_asts, KernelFunction):
+ kernel_asts = [kernel_asts]
+
+ for kernel_ast in kernel_asts:
+ kernel_name = kernel_ast.function_name
- header = kernel_header(kernel_ast, dialect)
- with open(include_path / f'{kernel_name}.h', 'w+') as f:
- f.write(header)
+ header = kernel_header(kernel_ast)
+ with open(include_path / f'{kernel_name}.h', 'w+') as f:
+ f.write(header)
- source = kernel_source(kernel_ast, dialect)
- with open(src_path / f'{kernel_name}.c', 'w+') as f:
- f.write(source)
+ source = kernel_source(kernel_ast)
+ with open(src_path / f'{kernel_name}.c', 'w+') as f:
+ f.write(source)
with open(src_path / 'main.c', 'w+') as f:
- f.write(kernel_main(kernel_ast))
+ f.write(kernel_main(kernel_asts))
copy_static_files(path)
+ compiler_toolchain(path, compiler)
+
+
+def compiler_toolchain(path: Path, compiler: Compiler) -> None:
+ name = compiler.name
+ jinja_context = {
+ 'compiler': name,
+ }
+
+ env = Environment(loader=PackageLoader('pystencils_benchmark'), undefined=StrictUndefined)
+ files = ['Makefile', f'{name}.mk']
+ for file_name in files:
+ with open(path / file_name, 'w+') as f:
+ template = env.get_template(file_name).render(**jinja_context)
+ f.write(template)
def copy_static_files(path: Path) -> None:
@@ -47,8 +70,7 @@ def copy_static_files(path: Path) -> None:
include_path.mkdir(parents=True, exist_ok=True)
env = Environment(loader=PackageLoader('pystencils_benchmark'), undefined=StrictUndefined)
- files = ['Makefile', 'aligned_malloc.h', 'timing.h', 'timing.c', 'include_Clang.mk', 'include_GCC.mk',
- 'include_ICC.mk', 'include_GCCdebug.mk']
+ files = ['aligned_malloc.h', 'timing.h', 'timing.c']
for file_name in files:
template = env.get_template(file_name).render()
if file_name[-1] == 'h':
@@ -61,61 +83,64 @@ def copy_static_files(path: Path) -> None:
f.write(template)
-def kernel_main(kernel: KernelFunction, timing: bool = True):
+def kernel_main(kernels_ast: List[KernelFunction], timing: bool = True):
"""
Return C code of a benchmark program for the given kernel.
Args:
- kernel: the pystencils AST object as returned by create_kernel
+ kernels_ast: A list of the pystencils AST object as returned by create_kernel for benchmarking
timing: add timing output to the code, prints time per iteration to stdout
Returns:
C code as string
"""
- kernel_name = kernel.function_name
- accessed_fields = {f.name: f for f in kernel.fields_accessed}
- constants = []
- fields = []
- call_parameters = []
- for p in kernel.get_parameters():
- if not p.is_field_parameter:
- constants.append((p.symbol.name, str(p.symbol.dtype)))
- call_parameters.append(p.symbol.name)
- else:
- assert p.is_field_pointer, "Benchmark implemented only for kernels with fixed loop size"
- field = accessed_fields[p.field_name]
- dtype = str(get_base_type(p.symbol.dtype))
- np_dtype = get_base_type(p.symbol.dtype).numpy_dtype
- size_data_type = np_dtype.itemsize
-
- dim0_size = field.shape[-1]
- dim1_size = np.prod(field.shape[:-1])
- elements = prod(field.shape)
-
- if kernel.instruction_set:
- align = kernel.instruction_set['width'] * size_data_type
- padding_elements = modulo_ceil(dim0_size, kernel.instruction_set['width']) - dim0_size
- padding_bytes = padding_elements * size_data_type
- ghost_layers = max(max(kernel.ghost_layers))
-
- size = dim1_size * padding_bytes + np.prod(field.shape) * size_data_type
-
- assert align % np_dtype.itemsize == 0
- offset = ((dim0_size + padding_elements + ghost_layers) % kernel.instruction_set['width']) * size_data_type
-
- fields.append((p.field_name, dtype, elements, size, offset, align))
- call_parameters.append(p.field_name)
+ Kernel = namedtuple('Kernel', ['name', 'constants', 'fields', 'call_parameters', 'call_argument_list'])
+ kernels = []
+ includes = set()
+ for kernel in kernels_ast:
+ name = kernel.function_name
+ accessed_fields = {f.name: f for f in kernel.fields_accessed}
+ constants = []
+ fields = []
+ call_parameters = []
+ for p in kernel.get_parameters():
+ if not p.is_field_parameter:
+ constants.append((p.symbol.name, str(p.symbol.dtype)))
+ call_parameters.append(p.symbol.name)
else:
- size = elements * size_data_type
- fields.append((p.field_name, dtype, elements, size, 0, 0))
- call_parameters.append(p.field_name)
+ assert p.is_field_pointer, "Benchmark implemented only for kernels with fixed loop size"
+ field = accessed_fields[p.field_name]
+ dtype = str(get_base_type(p.symbol.dtype))
+ np_dtype = get_base_type(p.symbol.dtype).numpy_dtype
+ size_data_type = np_dtype.itemsize
+
+ dim0_size = field.shape[-1]
+ dim1_size = np.prod(field.shape[:-1])
+ elements = prod(field.shape)
+
+ if kernel.instruction_set:
+ align = kernel.instruction_set['width'] * size_data_type
+ padding_elements = modulo_ceil(dim0_size, kernel.instruction_set['width']) - dim0_size
+ padding_bytes = padding_elements * size_data_type
+ ghost_layers = max(max(kernel.ghost_layers))
+
+ size = dim1_size * padding_bytes + np.prod(field.shape) * size_data_type
+
+ assert align % np_dtype.itemsize == 0
+ offset = ((dim0_size + padding_elements + ghost_layers) % kernel.instruction_set['width']) * size_data_type
+
+ fields.append((p.field_name, dtype, elements, size, offset, align))
+ call_parameters.append(p.field_name)
+ else:
+ size = elements * size_data_type
+ fields.append((p.field_name, dtype, elements, size, 0, 0))
+ call_parameters.append(p.field_name)
+ kernels.append(Kernel(name=name, fields=fields, constants=constants, call_parameters=call_parameters,
+ call_argument_list=",".join(call_parameters)))
+
+ includes.add(f'#include "{name}.h"\n')
- includes = f'#include "{kernel_name}.h"\n'
jinja_context = {
- 'kernel_code': generate_c(kernel, dialect=Backend.C),
- 'kernel_name': kernel_name,
- 'fields': fields,
- 'constants': constants,
- 'call_argument_list': ",".join(call_parameters),
+ 'kernels': kernels,
'includes': includes,
'timing': timing,
}
diff --git a/pystencils_benchmark/enums.py b/pystencils_benchmark/enums.py
new file mode 100644
index 0000000..ec56c8a
--- /dev/null
+++ b/pystencils_benchmark/enums.py
@@ -0,0 +1,8 @@
+from enum import Enum, auto
+
+
+class Compiler(Enum):
+ GCC = auto()
+ GCCdebug = auto()
+ Clang = auto()
+ ICC = auto()
diff --git a/pystencils_benchmark/templates/Clang.mk b/pystencils_benchmark/templates/Clang.mk
new file mode 100644
index 0000000..61eee7f
--- /dev/null
+++ b/pystencils_benchmark/templates/Clang.mk
@@ -0,0 +1,19 @@
+CC = clang
+LINKER = $(CC)
+
+ANSI_CFLAGS = -ansi
+ANSI_CFLAGS += -std=c99
+ANSI_CFLAGS += -pedantic
+ANSI_CFLAGS += -Wextra
+
+CFLAGS = -O3 -Wno-format -Wall $(ANSI_CFLAGS) -fopenmp
+# More warning pls
+#CFLAGS += -Wfloat-equal -Wundef -Wshadow -Wpointer-arith -Wcast-align -Wstrict-overflow=5 -Wwrite-strings -Waggregate-return
+# Maybe too much warnings
+#CFLAGS += -Wcast-qual -Wswitch-default -Wconversion -Wunreachable-code
+# Specific C flags
+CFLAGS := $(CFLAGS) -Wstrict-prototypes
+LFLAGS = -fopenmp
+DEFINES = -D_GNU_SOURCE -DNDEBUG
+INCLUDES =
+LIBS =
diff --git a/pystencils_benchmark/templates/include_GCC.mk b/pystencils_benchmark/templates/GCC.mk
similarity index 100%
rename from pystencils_benchmark/templates/include_GCC.mk
rename to pystencils_benchmark/templates/GCC.mk
diff --git a/pystencils_benchmark/templates/include_GCCdebug.mk b/pystencils_benchmark/templates/GCCdebug.mk
similarity index 69%
rename from pystencils_benchmark/templates/include_GCCdebug.mk
rename to pystencils_benchmark/templates/GCCdebug.mk
index 1685617..f453f0d 100644
--- a/pystencils_benchmark/templates/include_GCCdebug.mk
+++ b/pystencils_benchmark/templates/GCCdebug.mk
@@ -5,10 +5,8 @@ ANSI_CFLAGS = -ansi
ANSI_CFLAGS += -std=c99
ANSI_CFLAGS += -pedantic
ANSI_CFLAGS += -Wextra
-ANSI_CFLAGS += -O0
-ANSI_CFLAGS += -g
-CFLAGS = -Wno-format -Wall $(ANSI_CFLAGS)
+CFLAGS = -O0 -g -Wno-format -Wall $(ANSI_CFLAGS)
FCFLAGS =
LFLAGS =
DEFINES = -D_GNU_SOURCE
diff --git a/pystencils_benchmark/templates/include_ICC.mk b/pystencils_benchmark/templates/ICC.mk
similarity index 100%
rename from pystencils_benchmark/templates/include_ICC.mk
rename to pystencils_benchmark/templates/ICC.mk
diff --git a/pystencils_benchmark/templates/Makefile b/pystencils_benchmark/templates/Makefile
index 5490559..b9b8cfc 100644
--- a/pystencils_benchmark/templates/Makefile
+++ b/pystencils_benchmark/templates/Makefile
@@ -1,4 +1,4 @@
-TAG = GCC
+TAG = {{compiler}}
#CONFIGURE BUILD SYSTEM
TARGET = benchmark-$(TAG)
@@ -8,13 +8,13 @@ MAKE_DIR = ./
Q ?= @
#DO NOT EDIT BELOW
-include $(MAKE_DIR)/include_$(TAG).mk
+include $(MAKE_DIR)/$(TAG).mk
INCLUDES += -I./include
VPATH = $(SRC_DIR)
ASM = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.s,$(wildcard $(SRC_DIR)/*.c))
OBJ = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.o,$(wildcard $(SRC_DIR)/*.c))
-CPPFLAGS := $(CPPFLAGS) $(DEFINES) $(INCLUDES)
+CFLAGS := $(CFLAGS) $(DEFINES) $(INCLUDES)
${TARGET}: $(BUILD_DIR) $(OBJ)
@@ -25,8 +25,8 @@ asm: $(BUILD_DIR) $(ASM)
$(BUILD_DIR)/%.o: %.c
@echo "===> COMPILE $@"
- $(Q)$(CC) -c $(CPPFLAGS) $(CFLAGS) $< -o $@
- $(Q)$(CC) $(CPPFLAGS) -MT $(@:.d=.o) -MM $< > $(BUILD_DIR)/$*.d
+ $(Q)$(CC) -c $(CFLAGS) $< -o $@
+ $(Q)$(CC) $(CFLAGS) -MT $(@:.d=.o) -MM $< > $(BUILD_DIR)/$*.d
$(BUILD_DIR)/%.s: %.c
@echo "===> GENERATE ASM $@"
@@ -44,16 +44,7 @@ ifeq ($(findstring $(MAKECMDGOALS),clean),)
-include $(OBJ:.o=.d)
endif
-.PHONY: clean distclean debug debug-clean debug-distclean
-
-debug:
- @make TAG=GCCdebug
-
-debug-clean:
- @make clean TAG=GCCdebug
-
-debug-distclean:
- @make distclean TAG=GCCdebug
+.PHONY: clean distclean
clean:
@echo "===> CLEAN"
diff --git a/pystencils_benchmark/templates/include_Clang.mk b/pystencils_benchmark/templates/include_Clang.mk
deleted file mode 100644
index dab3ce3..0000000
--- a/pystencils_benchmark/templates/include_Clang.mk
+++ /dev/null
@@ -1,24 +0,0 @@
-CC = clang
-CXX = clang++
-FC = gfortran
-LINKER = $(CXX)
-
-ANSI_CFLAGS = -ansi
-ANSI_CFLAGS += -std=c++0x
-ANSI_CFLAGS += -pedantic
-ANSI_CFLAGS += -Wextra
-
-CFLAGS = -O3 -Wno-format -Wall $(ANSI_CFLAGS)
-# More warning pls
-CFLAGS += -Wfloat-equal -Wundef -Wshadow -Wpointer-arith -Wcast-align -Wstrict-overflow=5 -Wwrite-strings -Waggregate-return
-# Maybe too much warnings
-CFLAGS += -Wcast-qual -Wswitch-default -Wconversion -Wunreachable-code
-CXXFLAGS := $(CFLAGS)
-# Specific C flags
-CFLAGS := $(CFLAGS) -Wstrict-prototypes
-FCFLAGS =
-CPPFLAGS = -std=c++0x
-LFLAGS = #-fopenmp
-DEFINES = -D_GNU_SOURCE -DNDEBUG
-INCLUDES = #-I/usr/lib/gcc/x86_64-linux-gnu/4.8/include/
-LIBS = #-L/usr/lib/x86_64-linux-gnu/libomp.so.5
diff --git a/pystencils_benchmark/templates/main.c b/pystencils_benchmark/templates/main.c
index f15e9ab..7f8ea99 100644
--- a/pystencils_benchmark/templates/main.c
+++ b/pystencils_benchmark/templates/main.c
@@ -5,56 +5,66 @@
#include <stdio.h>
#include <assert.h>
-{{ includes }}
+#include "timing.h"
+
+{%- for include in includes %}
+{{ include }}
+{%- endfor %}
#define RESTRICT __restrict__
#define FUNC_PREFIX
-void timing(double* wcTime, double* cpuTime);
-
int main(int argc, char **argv)
{
- {%- for field_name, dataType, elements, size, offset, alignment in fields %}
- // Initialization {{field_name}}
- {%- if alignment > 0 %}
- {{dataType}} * {{field_name}} = ({{dataType}} *) aligned_malloc_with_offset({{size}}, {{alignment}}, {{offset}});
- {%- else %}
- {{dataType}} * {{field_name}} = ({{dataType}} *) malloc({{size}});
- {%- endif %}
- for (unsigned long long i = 0; i < {{elements}}; ++i)
- {{field_name}}[i] = 0.23;
- {%- endfor %}
-
- {%- for constantName, dataType in constants %}
- // Constant {{constantName}}
- {{dataType}} {{constantName}};
- {{constantName}} = 0.23;
- {%- endfor %}
-
- for(int warmup = 1; warmup >= 0; --warmup) {
- int repeat = 2;
- if(warmup == 0) {
- repeat = atoi(argv[1]);
+ if(argc < 2) {
+ printf("Usage: %s <n_repeat>\n", argv[0]);
+ return -1;
}
+ int n_repeat = atoi(argv[1]);
+ {%- for kernel in kernels %}
+ {
+ {%- for field_name, dataType, elements, size, offset, alignment in kernel.fields %}
+ // Initialization {{field_name}}
+ {%- if alignment > 0 %}
+ {{dataType}} * {{field_name}} = ({{dataType}} *) aligned_malloc_with_offset({{size}}, {{alignment}}, {{offset}});
+ {%- else %}
+ {{dataType}} * {{field_name}} = ({{dataType}} *) malloc({{size}});
+ {%- endif %}
+ for (unsigned long long i = 0; i < {{elements}}; ++i)
+ {{field_name}}[i] = 0.23;
+ {%- endfor %}
- {%- if timing %}
- double wcStartTime, cpuStartTime, wcEndTime, cpuEndTime;
- timing(&wcStartTime, &cpuStartTime);
- {%- endif %}
+ {%- for constantName, dataType in kernel.constants %}
+ // Constant {{constantName}}
+ {{dataType}} {{constantName}};
+ {{constantName}} = 0.23;
+ {%- endfor %}
- for (; repeat > 0; --repeat)
- {
- {{kernel_name}}({{call_argument_list}});
- }
- {%- if timing %}
- timing(&wcEndTime, &cpuEndTime);
- if( warmup == 0)
- printf("%e\n", (wcEndTime - wcStartTime) / atoi(argv[1]) );
- {%- endif %}
+ for(int warmup = 1; warmup >= 0; --warmup) {
+ int repeat = 2;
+ if(warmup == 0) {
+ repeat = n_repeat;
+ }
+
+ {%- if timing %}
+ double wcStartTime, cpuStartTime, wcEndTime, cpuEndTime;
+ timing(&wcStartTime, &cpuStartTime);
+ {%- endif %}
- }
+ for (; repeat > 0; --repeat)
+ {
+ {{kernel.name}}({{kernel.call_argument_list}});
+ }
+ {%- if timing %}
+ timing(&wcEndTime, &cpuEndTime);
+ if( warmup == 0)
+ printf("%s\t%e\n", "{{kernel.name}}",(wcEndTime - wcStartTime) / n_repeat );
+ {%- endif %}
+ }
- {%- for field_name, dataType, elements, size, offset, alignment in fields %}
- free({{field_name}});
- {%- endfor %}
+ {%- for field_name, dataType, elements, size, offset, alignment in kernel.fields %}
+ free({{field_name}});
+ {%- endfor %}
+ }
+ {%- endfor %}
}
diff --git a/requirements.txt b/requirements.txt
index ef5420a..14f8297 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,3 +1,4 @@
Jinja2>=3.0.2
pystencils>=0.4.1
-setuptools>=44.0.0
\ No newline at end of file
+setuptools>=44.0.0
+numpy>=1.21.4
\ No newline at end of file
diff --git a/test.ipynb b/test.ipynb
deleted file mode 100644
index e6ac951..0000000
--- a/test.ipynb
+++ /dev/null
@@ -1,130 +0,0 @@
-{
- "cells": [
- {
- "cell_type": "code",
- "execution_count": 1,
- "metadata": {
- "collapsed": true
- },
- "outputs": [],
- "source": [
- "import numpy as np\n",
- "import pystencils as ps\n",
- "from pystencils_benchmark import generate_benchmark\n",
- "from pathlib import Path"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 2,
- "outputs": [],
- "source": [
- "config = ps.CreateKernelConfig(function_name='vadd')"
- ],
- "metadata": {
- "collapsed": false,
- "pycharm": {
- "name": "#%%\n"
- }
- }
- },
- {
- "cell_type": "code",
- "execution_count": 3,
- "outputs": [],
- "source": [
- "a, b, c = ps.fields(a=np.ones(4000000), b=np.ones(4000000), c=np.ones(4000000))"
- ],
- "metadata": {
- "collapsed": false,
- "pycharm": {
- "name": "#%%\n"
- }
- }
- },
- {
- "cell_type": "code",
- "execution_count": 4,
- "outputs": [],
- "source": [
- "@ps.kernel\n",
- "def vadd():\n",
- " a[0] @= b[0] + c[0]"
- ],
- "metadata": {
- "collapsed": false,
- "pycharm": {
- "name": "#%%\n"
- }
- }
- },
- {
- "cell_type": "code",
- "execution_count": 5,
- "outputs": [
- {
- "data": {
- "text/plain": "<IPython.core.display.HTML object>",
- "text/html": "<style>pre { line-height: 125%; }\ntd.linenos .normal { color: inherit; background-color: transparent; padding-left: 5px; padding-right: 5px; }\nspan.linenos { color: inherit; background-color: transparent; padding-left: 5px; padding-right: 5px; }\ntd.linenos .special { color: #000000; background-color: #ffffc0; padding-left: 5px; padding-right: 5px; }\nspan.linenos.special { color: #000000; background-color: #ffffc0; padding-left: 5px; padding-right: 5px; }\n.highlight .hll { background-color: #ffffcc }\n.highlight { background: #f8f8f8; }\n.highlight .c { color: #408080; font-style: italic } /* Comment */\n.highlight .err { border: 1px solid #FF0000 } /* Error */\n.highlight .k { color: #008000; font-weight: bold } /* Keyword */\n.highlight .o { color: #666666 } /* Operator */\n.highlight .ch { color: #408080; font-style: italic } /* Comment.Hashbang */\n.highlight .cm { color: #408080; font-style: italic } /* Comment.Multiline */\n.highlight .cp { color: #BC7A00 } /* Comment.Preproc */\n.highlight .cpf { color: #408080; font-style: italic } /* Comment.PreprocFile */\n.highlight .c1 { color: #408080; font-style: italic } /* Comment.Single */\n.highlight .cs { color: #408080; font-style: italic } /* Comment.Special */\n.highlight .gd { color: #A00000 } /* Generic.Deleted */\n.highlight .ge { font-style: italic } /* Generic.Emph */\n.highlight .gr { color: #FF0000 } /* Generic.Error */\n.highlight .gh { color: #000080; font-weight: bold } /* Generic.Heading */\n.highlight .gi { color: #00A000 } /* Generic.Inserted */\n.highlight .go { color: #888888 } /* Generic.Output */\n.highlight .gp { color: #000080; font-weight: bold } /* Generic.Prompt */\n.highlight .gs { font-weight: bold } /* Generic.Strong */\n.highlight .gu { color: #800080; font-weight: bold } /* Generic.Subheading */\n.highlight .gt { color: #0044DD } /* Generic.Traceback */\n.highlight .kc { color: #008000; font-weight: bold } /* Keyword.Constant */\n.highlight .kd { color: #008000; font-weight: bold } /* Keyword.Declaration */\n.highlight .kn { color: #008000; font-weight: bold } /* Keyword.Namespace */\n.highlight .kp { color: #008000 } /* Keyword.Pseudo */\n.highlight .kr { color: #008000; font-weight: bold } /* Keyword.Reserved */\n.highlight .kt { color: #B00040 } /* Keyword.Type */\n.highlight .m { color: #666666 } /* Literal.Number */\n.highlight .s { color: #BA2121 } /* Literal.String */\n.highlight .na { color: #7D9029 } /* Name.Attribute */\n.highlight .nb { color: #008000 } /* Name.Builtin */\n.highlight .nc { color: #0000FF; font-weight: bold } /* Name.Class */\n.highlight .no { color: #880000 } /* Name.Constant */\n.highlight .nd { color: #AA22FF } /* Name.Decorator */\n.highlight .ni { color: #999999; font-weight: bold } /* Name.Entity */\n.highlight .ne { color: #D2413A; font-weight: bold } /* Name.Exception */\n.highlight .nf { color: #0000FF } /* Name.Function */\n.highlight .nl { color: #A0A000 } /* Name.Label */\n.highlight .nn { color: #0000FF; font-weight: bold } /* Name.Namespace */\n.highlight .nt { color: #008000; font-weight: bold } /* Name.Tag */\n.highlight .nv { color: #19177C } /* Name.Variable */\n.highlight .ow { color: #AA22FF; font-weight: bold } /* Operator.Word */\n.highlight .w { color: #bbbbbb } /* Text.Whitespace */\n.highlight .mb { color: #666666 } /* Literal.Number.Bin */\n.highlight .mf { color: #666666 } /* Literal.Number.Float */\n.highlight .mh { color: #666666 } /* Literal.Number.Hex */\n.highlight .mi { color: #666666 } /* Literal.Number.Integer */\n.highlight .mo { color: #666666 } /* Literal.Number.Oct */\n.highlight .sa { color: #BA2121 } /* Literal.String.Affix */\n.highlight .sb { color: #BA2121 } /* Literal.String.Backtick */\n.highlight .sc { color: #BA2121 } /* Literal.String.Char */\n.highlight .dl { color: #BA2121 } /* Literal.String.Delimiter */\n.highlight .sd { color: #BA2121; font-style: italic } /* Literal.String.Doc */\n.highlight .s2 { color: #BA2121 } /* Literal.String.Double */\n.highlight .se { color: #BB6622; font-weight: bold } /* Literal.String.Escape */\n.highlight .sh { color: #BA2121 } /* Literal.String.Heredoc */\n.highlight .si { color: #BB6688; font-weight: bold } /* Literal.String.Interpol */\n.highlight .sx { color: #008000 } /* Literal.String.Other */\n.highlight .sr { color: #BB6688 } /* Literal.String.Regex */\n.highlight .s1 { color: #BA2121 } /* Literal.String.Single */\n.highlight .ss { color: #19177C } /* Literal.String.Symbol */\n.highlight .bp { color: #008000 } /* Name.Builtin.Pseudo */\n.highlight .fm { color: #0000FF } /* Name.Function.Magic */\n.highlight .vc { color: #19177C } /* Name.Variable.Class */\n.highlight .vg { color: #19177C } /* Name.Variable.Global */\n.highlight .vi { color: #19177C } /* Name.Variable.Instance */\n.highlight .vm { color: #19177C } /* Name.Variable.Magic */\n.highlight .il { color: #666666 } /* Literal.Number.Integer.Long */</style>"
- },
- "metadata": {},
- "output_type": "display_data"
- },
- {
- "data": {
- "text/plain": "FUNC_PREFIX void vadd(double * RESTRICT _data_a, double * RESTRICT const _data_b, double * RESTRICT const _data_c)\n{\n for (int64_t ctr_0 = 0; ctr_0 < 4000000; ctr_0 += 1)\n {\n _data_a[ctr_0] = _data_b[ctr_0] + _data_c[ctr_0];\n }\n}",
- "text/html": "<div class=\"highlight\"><pre><span></span><span class=\"n\">FUNC_PREFIX</span><span class=\"w\"> </span><span class=\"kt\">void</span><span class=\"w\"> </span><span class=\"n\">vadd</span><span class=\"p\">(</span><span class=\"kt\">double</span><span class=\"w\"> </span><span class=\"o\">*</span><span class=\"w\"> </span><span class=\"n\">RESTRICT</span><span class=\"w\"> </span><span class=\"n\">_data_a</span><span class=\"p\">,</span><span class=\"w\"> </span><span class=\"kt\">double</span><span class=\"w\"> </span><span class=\"o\">*</span><span class=\"w\"> </span><span class=\"n\">RESTRICT</span><span class=\"w\"> </span><span class=\"k\">const</span><span class=\"w\"> </span><span class=\"n\">_data_b</span><span class=\"p\">,</span><span class=\"w\"> </span><span class=\"kt\">double</span><span class=\"w\"> </span><span class=\"o\">*</span><span class=\"w\"> </span><span class=\"n\">RESTRICT</span><span class=\"w\"> </span><span class=\"k\">const</span><span class=\"w\"> </span><span class=\"n\">_data_c</span><span class=\"p\">)</span><span class=\"w\"></span>\n<span class=\"p\">{</span><span class=\"w\"></span>\n<span class=\"w\"> </span><span class=\"k\">for</span><span class=\"w\"> </span><span class=\"p\">(</span><span class=\"kt\">int64_t</span><span class=\"w\"> </span><span class=\"n\">ctr_0</span><span class=\"w\"> </span><span class=\"o\">=</span><span class=\"w\"> </span><span class=\"mi\">0</span><span class=\"p\">;</span><span class=\"w\"> </span><span class=\"n\">ctr_0</span><span class=\"w\"> </span><span class=\"o\"><</span><span class=\"w\"> </span><span class=\"mi\">4000000</span><span class=\"p\">;</span><span class=\"w\"> </span><span class=\"n\">ctr_0</span><span class=\"w\"> </span><span class=\"o\">+=</span><span class=\"w\"> </span><span class=\"mi\">1</span><span class=\"p\">)</span><span class=\"w\"></span>\n<span class=\"w\"> </span><span class=\"p\">{</span><span class=\"w\"></span>\n<span class=\"w\"> </span><span class=\"n\">_data_a</span><span class=\"p\">[</span><span class=\"n\">ctr_0</span><span class=\"p\">]</span><span class=\"w\"> </span><span class=\"o\">=</span><span class=\"w\"> </span><span class=\"n\">_data_b</span><span class=\"p\">[</span><span class=\"n\">ctr_0</span><span class=\"p\">]</span><span class=\"w\"> </span><span class=\"o\">+</span><span class=\"w\"> </span><span class=\"n\">_data_c</span><span class=\"p\">[</span><span class=\"n\">ctr_0</span><span class=\"p\">];</span><span class=\"w\"></span>\n<span class=\"w\"> </span><span class=\"p\">}</span><span class=\"w\"></span>\n<span class=\"p\">}</span><span class=\"w\"></span>\n</pre></div>\n"
- },
- "metadata": {},
- "output_type": "display_data"
- }
- ],
- "source": [
- "kernel_vadd = ps.create_kernel(vadd, config=config)\n",
- "ps.show_code(kernel_vadd)"
- ],
- "metadata": {
- "collapsed": false,
- "pycharm": {
- "name": "#%%\n"
- }
- }
- },
- {
- "cell_type": "code",
- "execution_count": 6,
- "outputs": [],
- "source": [
- "example_path = Path.cwd() / 'example'\n",
- "generate_benchmark(kernel_vadd, example_path)"
- ],
- "metadata": {
- "collapsed": false,
- "pycharm": {
- "name": "#%%\n"
- }
- }
- }
- ],
- "metadata": {
- "kernelspec": {
- "display_name": "Python 3",
- "language": "python",
- "name": "python3"
- },
- "language_info": {
- "codemirror_mode": {
- "name": "ipython",
- "version": 2
- },
- "file_extension": ".py",
- "mimetype": "text/x-python",
- "name": "python",
- "nbconvert_exporter": "python",
- "pygments_lexer": "ipython2",
- "version": "2.7.6"
- }
- },
- "nbformat": 4,
- "nbformat_minor": 0
-}
\ No newline at end of file
diff --git a/test.py b/test.py
new file mode 100755
index 0000000..d516678
--- /dev/null
+++ b/test.py
@@ -0,0 +1,45 @@
+#!/usr/bin/env python
+# coding: utf-8
+import subprocess
+import numpy as np
+import sympy as sp
+import pystencils as ps
+from pystencils_benchmark import generate_benchmark
+from pathlib import Path
+
+
+def generate(path: Path):
+ a, b, c = ps.fields(a=np.ones(4000000), b=np.ones(4000000), c=np.ones(4000000))
+ alpha = sp.symbols('alpha')
+
+ @ps.kernel_config(ps.CreateKernelConfig())
+ def vadd():
+ a[0] @= b[0] + c[0]
+ kernel_vadd = ps.create_kernel(**vadd)
+
+ @ps.kernel_config(ps.CreateKernelConfig())
+ def daxpy():
+ b[0] @= alpha * a[0] + b[0]
+ kernel_daxpy = ps.create_kernel(**daxpy)
+
+ generate_benchmark([kernel_vadd, kernel_daxpy], path)
+
+
+def make(path: Path):
+ subprocess.run(['make'])
+
+
+def execute(path: Path):
+ subprocess.run(['./benchmark-GCC', '200'])
+
+
+def main():
+ path = Path.cwd()
+ generate(path)
+ make(path)
+ execute(path)
+
+
+if __name__ == '__main__':
+ main()
+
--
GitLab