Merge pull request PaddlePaddle#33 from NVIDIA/cutlass_1.2

CUTLASS 1.2

Author: kerrmudgeon

CHANGELOG.md

@@ -1,7 +1,11 @@
 # NVIDIA CUTLASS Changelog
 
+## [1.2.0](https://github.com/NVIDIA/cutlass/releases/tag/v1.2.0) (2018-10-26)
+ * Parallelized reductions across threadblocks ("Split-K")
+   * Improved IGEMM performance
+ * Batched strided WMMA GEMMs
 
-## 1.1.0 (2018-09-19)
+## [1.1.0](https://github.com/NVIDIA/cutlass/releases/tag/v1.1.0) (2018-09-19)
   * Turing Features
     * WMMA GEMM targeting TensorCores - INT8, INT4, 1-bit
   * Batched Strided GEMM
@@ -16,13 +20,13 @@
   * Examples
     * Basic GEMM, tensor views, CUTLASS utilities, batched GEMM, WMMA GEMM
 
-## 1.0.1 (2018-06-11)
+## [1.0.1](https://github.com/NVIDIA/cutlass/releases/tag/v1.0.1) (2018-06-11)
 
   * Intra-threadblock reduction added for small threadblock tile sizes
     * sgemm_64x128x16, sgemm_128x128x16, sgemm_128x64x16, sgemm_128x32x16, sgemm_64x64x16, sgemm_64x32x16
     * igemm_32x32x128
   * GEMM _K_ residue handled during prologue prior to mainloop
-  * Replaced Google Test copy with submodule. Use `git submodule init`
+  * Replaced Google Test copy with submodule. Use `git submodule init --recursive --update`
 
 ## [1.0.0](https://github.com/NVIDIA/cutlass/commit/2028ebe120aab22bfd0b2baf8902d4c9627eb33f) (2018-05-16)
 

CMakeLists.txt

@@ -141,6 +141,10 @@ else()
   string(APPEND NVCC_FLAGS " -lineinfo")
 endif()
 
+if (UNIX)
+  string(APPEND NVCC_FLAGS " -Xcompiler -Wconversion")
+endif()
+
 string(APPEND NVCC_FLAGS_DEBUG " -g")
 string(APPEND NVCC_FLAGS_RELWITHDEBINFO " -O3")
 string(APPEND NVCC_FLAGS_RELEASE " -O3")
@@ -169,6 +173,8 @@ file(GLOB CUTLASS_GEMM RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} cutlass/gemm/*.h)
 file(GLOB CUTLASS_UTIL RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} cutlass/util/*.h)
 file(GLOB CUTLASS_DEVICE RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} cutlass/device/*.h)
 file(GLOB CUTLASS_CORE RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} cutlass/*.h)
+file(GLOB CUTLASS_REDUCTION RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} cutlass/reduction/*.h )
+
 ###################################################################################################
 #
 # Define build targets
@@ -178,6 +184,7 @@ file(GLOB CUTLASS_CORE RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} cutlass/*.h)
 source_group("cutlass\\gemm" FILES ${CUTLASS_GEMM})
 source_group("cutlass\\util" FILES ${CUTLASS_UTIL})
 source_group("cutlass\\device" FILES ${CUTLASS_DEVICE})
+source_group("cutlass\\reduction" FILES ${CUTLASS_REDUCTION})
 source_group("cutlass" FILES ${CUTLASS_CORE})
 
 add_library(CUTLASS INTERFACE)
@@ -187,6 +194,7 @@ target_sources(CUTLASS INTERFACE
   ${CUTLASS_UTIL}
   ${CUTLASS_DEVICE}
   ${CUTLASS_CORE}
+  ${CUTLASS_REDUCTION}
 )
 
 target_include_directories(CUTLASS INTERFACE ${CMAKE_CURRENT_SOURCE_DIR})
@@ -197,6 +205,7 @@ add_custom_target(cutlass_ide SOURCES
   ${CUTLASS_UTIL}
   ${CUTLASS_DEVICE}
   ${CUTLASS_CORE}
+  ${CUTLASS_REDUCTION}
 )
 # Doxygen is available. Generate documentation
 if (DOXYGEN_FOUND)

CUTLASS.md

@@ -9,6 +9,7 @@ CUTLASS core components, and to identify their role in implementing GEMM computa
 2. [General Matrix Multiply](#S-general-matrix-multiply)
 3. [Core Components](#S-core-components)
 4. [Utilities](#S-utilities)
+5. [Optimization Strategies](#S-optimization-strategies)
 
 # <a name="S-design-patterns"></a> 1. Design Patterns
 
@@ -26,7 +27,7 @@ objectives. This section is intended to provide more detail.
 
 ## <a name="S-patterns-sequencing-nesting"></a> Sequencing and Nesting of Collective Primitives
 
-CUTLASS embodies a design paradigm exemplified by the [CUB library](https://nvlabs.github.io/cub/) for expressing collective operations. Objects expose an interface for a problem that is then decomposed into concurrent subtasks executed by cooperating threadblocks, warps, and threads. For example, a grid-level object may be constructed with base pointers to the start of a GEMM operation, add a threadblock-dependent offset to partition the problem, and then compute a per-threadblock GEMM. This in turn performs some operations as a collection of cooperating threads, while it may partition other parts of the task into warp-level subtasks. 
+CUTLASS embodies a design paradigm exemplified by the [CUB library](https://nvlabs.github.io/cub/) for expressing collective operations. Objects expose an interface for a problem that is then decomposed into concurrent subtasks executed by cooperating threadblocks, warps, and threads. For example, a grid-level object may be constructed with base pointers to the start of a GEMM operation, add a threadblock-dependent offset to partition the problem, and then compute a per-threadblock GEMM. This in turn performs some operations as a collection of cooperating threads, while it may partition other parts of the task into warp-level subtasks.
 
 ## <a name="S-patterns-tiles-iterators"></a> Tiles and Iterators
 
@@ -48,7 +49,7 @@ CUTLASS can take advantage of this CUDA grid-invariant property by constructing
 
 The design pattern in CUTLASS is for classes with nontrivial constructors to define `struct Params` as an inner class which contains grid-invariant state. These should define a constructor and an `initialize()` method. The `Params` structure should also include a data member corresponding to each data member in the parent class, so these too can be properly constructed in host code. The parent class should define a constructor which accepts `Params const &` as its first argument.
 
-For example, `cutlass::gemm::Gemm<>` should define `struct cutlass::gemm::Gemm::Params`. The latter should define data members for each data member in `cutlass::gemm::Gemm<>`. 
+For example, `cutlass::gemm::Gemm<>` should define `struct cutlass::gemm::Gemm::Params`. The latter should define data members for each data member in `cutlass::gemm::Gemm<>`.
 
 
 ## <a name="S-patterns-composable-shared-memory"></a> Composable shared memory allocation
@@ -94,7 +95,7 @@ multiply operation performed by each iteration of the mainloop is referred to as
 
 The threadblock loads a sequence of tiles from global memory and stores this data to shared memory. The iterative
 access and traversal of tiles in global memory are performed by a _TileLoadIterator_, and storing to a circular
-buffer in shared memory is performed by a _GlobalLoadIterator_. 
+buffer in shared memory is performed by a _GlobalLoadIterator_.
 
 **[Global Load Stream](cutlass/gemm/gemm_global_stream.h)** manages loading of the threadblock-scope multiplicands to the GEMM kernel. It owns an iterator into global memory for loading tiles of data, a TensorAllocation in shared memory to hold the resulting tile, and an iterator for writing the tile into this allocation. A transformer exists to optionally transform the data as it is loaded which may of use to perform type conversion or, in the case of int8 GEMM, transpose 4x4 tiles held in registers.
 
@@ -109,24 +110,24 @@ The Global Load Stream template contains members defined by the following templa
 The threadblock's _OutputTile_ is partitioned among the warps, and each computes a warp-level matrix product.
 Data is loaded from shared memory into registers, and math instructions are dispatched to CUDA Cores or Tensor Cores.
 
-[**Shared Load Stream**](cutlass/gemm/gemm_shared_stream.h) manages loading of warp-level multiplicands from shared memory into registers. This owns an iterator for fetching data and the destination fragments for holding the results. 
+[**Shared Load Stream**](cutlass/gemm/gemm_shared_stream.h) manages loading of warp-level multiplicands from shared memory into registers. This owns an iterator for fetching data and the destination fragments for holding the results.
 
 * [GemmSharedLoadTile{A,B}](cutlass/gemm/gemm_shared_tile.h)
 
-**Matrix Multiply** computes a matrix product operation on data held in registers. Specializations exist for thread-level instructions such as single-precision fused multiply-add as well as warp-level matrix operations targeting TensorCores. 
+**Matrix Multiply** computes a matrix product operation on data held in registers. Specializations exist for thread-level instructions such as single-precision fused multiply-add as well as warp-level matrix operations targeting TensorCores.
 
 * [WMMA Multiply Add](cutlass/gemm/wmma_gemm_multiply_add.h)
 
 ## Thread-level GEMM
 
 SGEMM, IGEMM, HGEMM, and DGEMM are computed by SIMT math instructions issued by thread-level matrix multiply
-procedures. 
+procedures.
 
 * [ThreadMultiplyAdd](cutlass/gemm/thread_multiply_add.h)
 * [IGEMM specialization](cutlass/gemm/igemm_multiply_add.h)
 * [HGEMM specialization](cutlass/gemm/hgemm_multiply_add.h)
 
-## Epilogue 
+## Epilogue
 
 The [**epilogue**](cutlass/gemm/gemm_epilogue.h) iteratively selects a subset of accumulator elements held by a warp, writes them to shared memory, and loads them by different threads such that a threadblock-scoped tile store operation will make contiguous, striped accesses to global memory. Thus, the flow of data utilizes the following components:
 
@@ -227,7 +228,7 @@ must specify compile-time constant tile sizes.
 ## <a name="S-core-tile-structure"></a> Tile Structure
 
 Tiled structures express an arrangement of data in memory as well as a logical mapping of concurrent CUDA
-threads to the problem space. For example, the CUTLASS GEMM 
+threads to the problem space. For example, the CUTLASS GEMM
 
 Tiled structures can be defined using the `cutlass::TileTraits<>` concept which defines the following
 members. Collectively, these members offer a flexible way to define a 4-D subpartition of an integer
@@ -286,7 +287,7 @@ the next item in sequence.
 <img src="/media/images/cutlass-tile-iteration.png" alt="CUTLASS tile access and traversal" width="50%" />
 
 To offer a generic solution that spans numerous data types and layouts, CUTLASS defines the _TileIterator_ concept.
-This concept provides access to a sequence of _tiles_ embedded in a tensor in addressable memory. 
+This concept provides access to a sequence of _tiles_ embedded in a tensor in addressable memory.
 
 The canonical CUTLASS tile iterator template is defined in [cutlass/tile_iterator.h](cutlass/tile_iterator.h).
 
@@ -296,9 +297,9 @@ A fragment is analogous to `std::array<>` in that it is a constant-sized array o
 
 ## <a name="S-core-predicate-vector"></a> Predicate Vector
 
-SIMT architectures utilize predicated execution in place of control flow when conditional code sequences are fairly short, on the order of a few machine instructions. While CUDA C++ does not include constructs at the language level for predication, PTX makes this explicit, and compilation to SASS is assumed to aggressively utilize predication. Typical applications are to initialize a sequence of bits used to mask memory operations and use these bits as predicates guarding memory load and store instructions. 
+SIMT architectures utilize predicated execution in place of control flow when conditional code sequences are fairly short, on the order of a few machine instructions. While CUDA C++ does not include constructs at the language level for predication, PTX makes this explicit, and compilation to SASS is assumed to aggressively utilize predication. Typical applications are to initialize a sequence of bits used to mask memory operations and use these bits as predicates guarding memory load and store instructions.
 
-CUTLASS provides `PredicateVector` defined in [cutlass/predicate_vector.h](cutlass/predicate_vector.h) to manage a statically-sized bit vector, store them into general purpose registers, and efficiently access them in sequence. By storing four predicates per byte in hardware registers, the CUDA compiler is able to issue specialized instructions to achieve very efficient unpacking. 
+CUTLASS provides `PredicateVector` defined in [cutlass/predicate_vector.h](cutlass/predicate_vector.h) to manage a statically-sized bit vector, store them into general purpose registers, and efficiently access them in sequence. By storing four predicates per byte in hardware registers, the CUDA compiler is able to issue specialized instructions to achieve very efficient unpacking.
 
 
 # <a name="S-utilities"></a> 4. Utilities
@@ -310,6 +311,46 @@ framework offering features such as:
 * Components for allocating and initializing [host-side and device-side tensors](tools/util/host_tensor.h) usable by CUTLASS
 * Reference implementations of [GEMM](tools/util/reference/host/gemm.h) and [element-wise operations](tools/util/reference/host/tensor_elementwise.h)
 
+
+# <a name="S-optimization-strategies"></a>5. Optimization Strategies
+
+This section describes several strategies taken to increase performance beyond what is achievable with
+a basic implementation of the hierarchical GEMM structure.
+
+
+## Threadblock Rasterization
+
+To maximize reuse of data held in the last level cache, CUTLASS defines several functions to
+affect the mapping of threadblocks to logical partitions of the GEMM problem. These map
+consecutively launched threadblocks to packed two-dimensional regions of the partitioned GEMM
+problem to increase the probability that these will access the same tiles of global memory at
+approximately the same time.
+
+Several functions are defined in [cutlass/gemm/threadblock_swizzle.h](cutlass/gemm/threadblock_swizzle.h).
+
+
+## Parallel Reductions across GEMM _K_
+
+Matrix product computations expose parallelism among _O(MN)_ independent inner product
+computations. For sufficiently large problem sizes, a GEMM kernel in CUTLASS may approach
+the theoretical maximum computational throughput. For small problems, however, there are
+too few threadblocks to efficiently occupy the entire GPU.
+
+As a recourse, parallelizing the reduction performed during the inner product computation
+enables more threadblocks to execute concurrently while still taking advantage of the throughput
+benefits of large threadblock-level GEMM tiles.
+
+CUTLASS implements parallel reductions across threadblocks by partitioning the GEMM _K_ dimension
+and launching an additional set of threadblocks for each partition. Consequently, we refer to
+this strategy within CUTLASS as "parallel reduction splitK." The "parallel reduction splitK" in cutlass requires the execution of 2 kernels. The first one is called partitionedK GEMM. The second one is called batched reduction.
+
+The partitionedK GEMM is very similar to one flavor of batched strided GEMM. Instead of requiring users to specify the problem size of each batch, partitionedK GEMM asks for the overall problem size and the number of partition that will be applied along K dimension for operand A and B. For example, parameters of m=128, n=128, k=4096 and partition=16 will result in 16 batched strided GEMMs with each batch of m=128, n=128, k=256. PartitionedK also allows scenario where k is not divisible by partition count. For example, parameters of m=128, n=128, k=4096 and partition=20 will result in 20 batched strided GEMMs with the first 19 batches of m=128, n=128, k=4096/20=204 and the last batch of m=128, n=128, k=220.
+
+The batched reduction kernel will further perform reduction along the K-dimension. Thus, the input of the batched reduction kernel is the output (C) of partitionedK GEMM. An workspace memory is managed by the users to store this intermediate results.
+
+An example of splitK usage can be found [here](examples/06_splitK_gemm/splitK_gemm.cu).
+
+
 # Copyright
 
 Copyright (c) 2017-2018, NVIDIA CORPORATION.  All rights reserved.
@@ -335,4 +376,3 @@ Copyright (c) 2017-2018, NVIDIA CORPORATION.  All rights reserved.
   STRICT LIABILITY, OR TOR (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 ```
-

README.md

@@ -1,10 +1,10 @@
 ![ALT](/media/images/gemm-hierarchy-with-epilogue-no-labels.png "Complete CUDA GEMM decomposition")
 
-# CUTLASS 1.1
+# CUTLASS 1.2
 
-_CUTLASS 1.1.0 - September 2018_
+_CUTLASS 1.2.0 - October 2018_
 
-CUTLASS 1.1 is a collection of CUDA C++ template abstractions for implementing
+CUTLASS is a collection of CUDA C++ template abstractions for implementing
 high-performance matrix-multiplication (GEMM) at all levels and scales within CUDA.
 It incorporates strategies for hierarchical decomposition and data movement similar
 to those used to implement cuBLAS.  CUTLASS decomposes these "moving parts" into
@@ -22,12 +22,19 @@ point (FP64) types.  Furthermore, CUTLASS demonstrates CUDA's WMMA API for targe
 the programmable, high-throughput _Tensor Cores_ provided by NVIDIA's Volta architecture
 and beyond.
 
-CUTLASS 1.1 is described in the [CUTLASS Documentation](CUTLASS.md) and the accompanying
+CUTLASS 1.2 is described in the [CUTLASS Documentation](CUTLASS.md) and the accompanying
 [Doxygen documentation](https://nvidia.github.io/cutlass).
 We describe the structure of an efficient GEMM in our talk at the
 [GPU Technology Conference 2018](http://on-demand.gputechconf.com/gtc/2018/presentation/s8854-cutlass-software-primitives-for-dense-linear-algebra-at-all-levels-and-scales-within-cuda.pdf).
 
+# What's New in CUTLASS 1.2
+_October 2018_
+* [Parallelized Reductions](CUTLASS.md#parallel-reductions-across-gemm-k)
+* Batched strided WMMA GEMM
+
+
 # What's New in CUTLASS 1.1
+_September 2018_
 
 * [CUTLASS Documentation](CUTLASS.md)
 * [Examples](examples/)

cutlass/coord.h

@@ -313,6 +313,56 @@ struct Coord {
 
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
+/// Scalar multiplication
+template <typename T, int Rank, typename Index>
+CUTLASS_HOST_DEVICE
+Coord<Rank, Index> operator*(T s, Coord<Rank, Index> coord) {
+  CUTLASS_PRAGMA_UNROLL
+  for (int i = 0; i < Rank; ++i) {
+    coord[i] *= s;
+  }
+  return coord;
+}
+
+/// Scalar multiplication
+template <typename T, int Rank, typename Index>
+CUTLASS_HOST_DEVICE
+Coord<Rank, Index> operator*(Coord<Rank, Index> coord, T s) {
+  CUTLASS_PRAGMA_UNROLL
+  for (int i = 0; i < Rank; ++i) {
+    coord[i] *= s;
+  }
+  return coord;
+}
+
+/// Scalar division
+template <typename T, int Rank, typename Index>
+CUTLASS_HOST_DEVICE
+Coord<Rank, Index> operator/(T s, Coord<Rank, Index> coord) {
+  CUTLASS_PRAGMA_UNROLL
+  for (int i = 0; i < Rank; ++i) {
+    coord[i] = s / coord[i];
+  }
+  return coord;
+}
+
+/// Scalar division
+template <typename T, int Rank, typename Index>
+CUTLASS_HOST_DEVICE
+Coord<Rank, Index> operator/(Coord<Rank, Index> coord, T s) {
+  CUTLASS_PRAGMA_UNROLL
+  for (int i = 0; i < Rank; ++i) {
+    coord[i] /= s;
+  }
+  return coord;
+}
+
+////////////////////////////////////////////////////////////////////////////////////////////////////
+//
+// Integer-valued make_Coord
+//
+////////////////////////////////////////////////////////////////////////////////////////////////////
+
 /// Helper to make a 2-element coordinate
 CUTLASS_HOST_DEVICE
 Coord<1> make_Coord(int _0) {

cutlass/cutlass.h

@@ -32,7 +32,7 @@
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
 #define CUTLASS_MAJOR 1
-#define CUTLASS_MINOR 1
+#define CUTLASS_MINOR 2
 #define CUTLASS_PATCH 0
 #define CUTLASS_VERSION ((CUTLASS_MAJOR)*100 + (CUTLASS_MINOR)*10 + CUTLASS_PATCH)
 
@@ -49,21 +49,7 @@
 
 #define CUTLASS_ASSERT(x) assert(x)
 
-// CUTLASS_PRAGMA_(UNROLL|NO_UNROLL) optimization directives for the CUDA compiler.
-#if defined(__CUDA_ARCH__)
-#if defined(_MSC_VER)
-#define CUTLASS_PRAGMA_UNROLL __pragma("unroll")
-#define CUTLASS_PRAGMA_NO_UNROLL __pragma("unroll 1")
-#else
-#define CUTLASS_PRAGMA_UNROLL _Pragma("unroll")
-#define CUTLASS_PRAGMA_NO_UNROLL _Pragma("unroll 1")
-#endif
-#else
-#define CUTLASS_PRAGMA_UNROLL
-#define CUTLASS_PRAGMA_NO_UNROLL
-#endif
-
-#define CUTLASS_GEMM_LOOP CUTLASS_PRAGMA_NO_UNROLL
+#include "cutlass/util/performance_tuning.h"
 
 // A small helper class to dump a type at compile time
 // Usage:: DumpType<Class>::Class