thrust / dependencies /cub /docs /test_overview.rst
camenduru's picture
thanks to nvidia ❤
8ae5fc5
CUB Testing Overview
###########################
.. warning::
CUB is in the progress of migrating to [Catch2](https://github.com/catchorg/Catch2) framework.
CUB tests rely on `CPM <https://github.com/cpm-cmake/CPM.cmake>`_ to fetch
`Catch2 <https://github.com/catchorg/Catch2>`_ that's used as our main testing framework
along with `metal <https://github.com/brunocodutra/metal>`_ that's used as template metaprogramming
backend for some of the test macro implementation.
Currently,
legacy tests coexist with Catch2 ones.
This guide is focused on new tests.
.. important::
Instead of including ``<catch2/catch.hpp>`` directly, use ``catch2_test_helper.h``.
Also,
the helper header has to be included after all CUB headers,
otherwise the test won't catch some of the warnings in CUB kernels:
.. code-block:: c++
#include <cub/block/block_scan.cuh>
#include <thrust/host_vector.h>
// Has to go after all cub headers
#include "catch2_test_helper.h"
Directory and File Naming
*************************************
Our tests can be found in the ``test`` directory.
Legacy tests have the following naming scheme: ``test_SCOPE_FACILITY.cu``.
For instance, here are the reduce tests:
.. code-block:: c++
test/test_warp_reduce.cu
test/test_block_reduce.cu
test/test_device_reduce.cu
Catch2-based tests have a different naming scheme: ``catch2_test_SCOPE_FACILITY.cu``.
The prefix is essential since that's how CMake finds tests
and distinguishes new tests from legacy ones.
Test Structure
*************************************
Base case
=====================================
Let's start with a simple example.
Say there's no need to cover many types with your test.
.. code-block:: c++
// 0) Define test name and tags
CUB_TEST("SCOPE FACILITY works with CONDITION",
"[FACILITY][SCOPE]")
{
using type = std::int32_t;
constexpr int threads_per_block = 256;
constexpr int num_items = threads_per_block;
// 1) Allocate device input
thrust::device_vector<type> d_input(num_items);
// 2) Generate 3 random input arrays using Catch2 helper
c2h::gen(CUB_SEED(3), d_input);
// 3) Allocate output array
thrust::device_vector<type> d_output(d_input.size());
// 4) Copy device input to host
thrust::host_vector<key_t> h_reference = d_input;
// 5) Compute reference output
std::ALGORITHM(
thrust::raw_pointer_cast(h_reference.data()),
thrust::raw_pointer_cast(h_reference.data()) + h_reference.size());
// 6) Compute CUB output
SCOPE_ALGORITHM<threads_per_block>(d_input.data(),
d_output.data(),
d_input.size());
// 7) Compare device and host results
REQUIRE( d_input == d_output );
}
We introduce test cases with the ``CUB_TEST`` macro in (0).
This macro always takes two string arguments - a free-form test name and
one or more tags. Then, in (1), we allocate device memory using ``thrust::device_vector``.
The memory is filled with random data in (2).
Generator ``c2h::gen`` takes two parameters.
The first one is a random generator seed.
Instead of providing a single value, we use the ``CUB_SEED`` macro.
The macro expects a number of seeds that has to be generated.
In the example above, we require three random seeds to be generated.
This leads to the whole test being executed three times
with different seed values.
Later,
in (3) and (4),
we allocate device output and host reference.
Then, in (4),
we populate host input data and perform reference computation on the host in (5).
Then launch the CUB algorithm in (6).
At this point, we have a reference solution on CPU and CUB solution on GPU.
The two can be compared with ``REQUIRE`` assert.
.. warning::
Standard algorithms (``std::``) have to be used as much as possible when computing reference solutions.
If your test has to cover floating point types,
it's sufficient to replace ``REQUIRE( a == b )`` with ``REQUIRE_APPROX_EQ(a, b)``.
It's strongly advised to always use ``c2h::gen`` to produce input data.
Other data generation methods might be used
if absolutely necessary in tests of corner cases.
If a custom type has to be tested, the following helper should be used:
.. code-block:: c++
using type = c2h::custom_type_t<c2h::accumulateable_t,
c2h::equal_comparable_t>;
Here we enumerate all the type properties that we are interested in.
The produced type ends up having ``operator==`` and ``operator+``.
There are more properties implemented.
If some property is missing,
it'd be better to add one in ``c2h``
instead of writing a custom type from scratch.
Type Lists
=====================================
Since CUB is a generic library,
it's often required to test CUB algorithms against many types.
To do so,
it's sufficient to define a type list and provide it to the ``CUB_TEST`` macro.
.. code-block:: c++
// 0) Define type list
using types = c2h::type_list<std::uint8_t, std::int32_t>;
CUB_TEST("SCOPE FACILITY works with CONDITION",
"[FACILITY][SCOPE]",
types) // 1) Provide it to the test case
{
// 2) Access current type with `c2h::get`
using type = typename c2h::get<0, TestType>;
// ...
}
This will lead to the test running two times.
The first run will cause the ``type`` to be ``std::uint8_t``.
The second one will cause ``type`` to be ``std::uint32_t``.
.. warning::
It's important to use types in ``std::`` instead of primitive types like ``char`` and ``int``.
Multidimensional Configuration Spaces
=====================================
In most cases, the input data type is not the only compile-time parameter we want to vary.
For instance, you might need to test a block algorithm for different data types
**and** different thread block sizes.
To do so, you can add another type list as follows:
.. code-block:: c++
using block_sizes = c2h::enum_type_list<int, 128, 256>;
using types = c2h::type_list<std::uint8_t, std::int32_t>;
CUB_TEST("SCOPE FACILITY works with CONDITION",
"[FACILITY][SCOPE]",
types,
block_sizes)
{
using type = typename c2h::get<0, TestType>;
constexpr int threads_per_block = c2h::get<1, TestType>::value;
// ...
}
The code above leads to the following combinations being compiled:
- ``type = std::uint8_t``, ``threads_per_block = 128``
- ``type = std::uint8_t``, ``threads_per_block = 256``
- ``type = std::int32_t``, ``threads_per_block = 128``
- ``type = std::int32_t``, ``threads_per_block = 256``
Speedup Compilation Time
=====================================
Since type lists in the ``CUB_TEST`` form a Cartesian product,
compilation time grows quickly with every new dimension.
To keep the compilation process parallelized,
it's possible to rely on ``%PARAM%`` machinery:
.. code-block:: c++
// %PARAM% BLOCK_SIZE bs 128:256
using block_sizes = c2h::enum_type_list<int, BLOCK_SIZE>;
using types = c2h::type_list<std::uint8_t, std::int32_t>;
CUB_TEST("SCOPE FACILITY works with CONDITION",
"[FACILITY][SCOPE]",
types,
block_sizes)
{
using type = typename c2h::get<0, TestType>;
constexpr int threads_per_block = c2h::get<1, TestType>::value;
// ...
}
The comment with ``%PARAM%`` is recognized at CMake level.
It leads to multiple executables being produced from a single test source.
.. code-block:: bash
bin/cub.test.scope_algorithm.bs_128
bin/cub.test.scope_algorithm.bs_256
Multiple ``%PARAM%`` comments can be specified forming another Cartesian product.
Final Test
=====================================
Let's consider the final test that illustrates all of the tools we discussed above:
.. code-block:: c++
// %PARAM% BLOCK_SIZE bs 128:256
using block_sizes = c2h::enum_type_list<int, BLOCK_SIZE>;
using types = c2h::type_list<std::uint8_t, std::int32_t>;
CUB_TEST("SCOPE FACILITY works with CONDITION",
"[FACILITY][SCOPE]",
types,
block_sizes)
{
using type = typename c2h::get<0, TestType>;
constexpr int threads_per_block = c2h::get<1, TestType>::value;
constexpr int max_num_items = threads_per_block;
thrust::device_vector<type> d_input(
GENERATE_COPY(take(2, random(0, max_num_items))));
c2h::gen(CUB_SEED(3), d_input);
thrust::device_vector<type> d_output(d_input.size());
SCOPE_ALGORITHM<threads_per_block>(d_input.data(),
d_output.data(),
d_input.size());
REQUIRE( d_input == d_output );
}
Apart from discussed tools, here we also rely on ``Catch2`` to generate random input sizes in ``[0, max_num_items]`` range.
Overall, the test will produce two executables.
Each of these executables is going to generate ``2`` input problem sizes.
For each problem size, ``3`` random vectors are generated.
As a result, we have ``12`` different tests.