COMPMID-881: RSH new arm_gemm interface.
Change-Id: I1e2a1a77097d8017c274af3f97eba6964f80f5fa
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/122592
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
diff --git a/src/core/NEON/kernels/arm_gemm/gemm_interleaved.hpp b/src/core/NEON/kernels/arm_gemm/gemm_interleaved.hpp
new file mode 100644
index 0000000..27e4e8d
--- /dev/null
+++ b/src/core/NEON/kernels/arm_gemm/gemm_interleaved.hpp
@@ -0,0 +1,535 @@
+/*
+ * Copyright (c) 2017-2018 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#pragma once
+
+#include <assert.h>
+#include <stdio.h>
+
+#include <algorithm>
+
+#include "arm_gemm.hpp"
+#include "utils.hpp"
+
+#include "buffer_manager.hpp"
+#include "mergeresults.hpp"
+#include "profiler.hpp"
+#include "transform.hpp"
+
+// Some macros used to decide how much working space to allocate.
+// Round allocations up to the next cache line.
+#define ALLOC_ROUND 64
+#define ROUND_UP(x) ((((x) + ALLOC_ROUND - 1) / ALLOC_ROUND) * ALLOC_ROUND)
+
+// Implementation of the GemmCommon abstract class.
+//
+// This implementation interleaves the source matrices in blocks - good for
+// larger matrices.
+namespace arm_gemm
+{
+template <typename strategy, typename To, typename Tr>
+class GemmInterleaved : public GemmCommon<To, Tr>
+{
+ typedef typename strategy::operand_type Toi;
+ typedef typename strategy::result_type Tri;
+
+ /* const properties set by constructor */
+ const CPUInfo *const _ci;
+
+ const unsigned int _Msize;
+ const unsigned int _Nsize;
+ const unsigned int _Ksize;
+
+ const bool _trA;
+ const bool _trB;
+
+ const Tr _alpha;
+ const Tr _beta;
+
+ const unsigned int _maxthreads;
+ const bool _pretransposed;
+
+ /* Blocking info */
+ unsigned int _k_block = 0;
+ unsigned int _x_block = 0;
+ unsigned int _Mround = 0;
+
+ /* Working space, pretransposed buffer, buffer manager */
+ const Toi *_B_transposed = nullptr;
+ BufferManager *_bm = nullptr;
+ void *_working_space = nullptr;
+
+ /* We will need to walk through the blocks of B in a few contexts, so
+ * factor that out. */
+ class blockwalker
+ {
+ private:
+ /* Loop parameters, we only block up N and K so don't worry about M. */
+ const unsigned int _Nsize, _Ksize, _x_block, _k_block;
+
+ /* K and X parameters for current iteration. */
+ unsigned int _k0 = 0, _x0 = 0;
+
+ unsigned int _index = 0;
+ bool _done = false;
+ bool _newkblock = true;
+
+ public:
+ blockwalker(const unsigned int K, const unsigned int k_block, const unsigned int N, const unsigned int x_block)
+ : _Nsize(N), _Ksize(K), _x_block(x_block), _k_block(k_block)
+ {
+ }
+
+ unsigned int xmax()
+ {
+ return std::min(_x0 + _x_block, _Nsize);
+ }
+
+ unsigned int kmax()
+ {
+ return std::min(_k0 + _k_block, _Ksize);
+ }
+
+ /* Advance to the next block, return false at the end. */
+ bool advance(void)
+ {
+ if(_done)
+ {
+ return false;
+ }
+
+ _newkblock = false;
+ _x0 += _x_block;
+ if(_x0 >= _Nsize)
+ {
+ _x0 = 0;
+ _k0 += _k_block;
+ if(_k0 >= _Ksize)
+ {
+ _done = true;
+ return false;
+ }
+ _newkblock = true;
+ }
+ _index++;
+
+ return true;
+ }
+
+ unsigned int k0(void)
+ {
+ return _k0;
+ }
+ unsigned int x0(void)
+ {
+ return _x0;
+ }
+ unsigned int index(void)
+ {
+ return _index;
+ }
+ bool done(void)
+ {
+ return _done;
+ }
+ bool newkblock(void)
+ {
+ return _newkblock;
+ }
+ };
+
+ // A working size: One of these needed, regardless of thread count. Divided according to window.
+ size_t get_a_working_size() const
+ {
+ return ROUND_UP(sizeof(Toi) * _k_block * _Mround);
+ }
+
+ // B working size: 0, 1 or 3 of these needed depending on pretransposed and threading settings.
+ size_t get_b_working_size() const
+ {
+ return ROUND_UP(sizeof(Toi) * _x_block * _k_block);
+ }
+
+ // C working size: One needed per thread.
+ size_t get_c_working_size() const
+ {
+ return ROUND_UP(sizeof(Tri) * _x_block * strategy::out_height);
+ }
+
+ // Internal execute function.
+ // This supports both the "pretransposed" and "standard" interfaces via the template parameter.
+ template <bool pretransposed>
+ void execute_internal(unsigned int start, unsigned int end, int threadid)
+ {
+ profiler prof;
+ strategy strat(_ci);
+
+ blockwalker current(_Ksize, _k_block, _Nsize, _x_block);
+ blockwalker next = current;
+
+ /* Compute the M values to operate on */
+ unsigned int m_0 = start * strategy::out_height;
+ unsigned int m_max = std::min(end * strategy::out_height, _Msize);
+
+ /* Make sure we've been set up correctly. */
+ if(pretransposed)
+ {
+ assert(_B_transposed);
+ }
+ else
+ {
+ assert(_bm);
+ }
+
+ assert(_working_space);
+ int8_t *working_space_bytes = reinterpret_cast<int8_t *>(_working_space);
+
+ // Private buffers. Treat working_space as an array of C buffers (one per thread) first, followed by the (window-divided) A buffer.
+ Toi *const a_panel = reinterpret_cast<Toi *>(working_space_bytes + (_maxthreads * get_c_working_size()) + (m_0 * _k_block * sizeof(Toi)));
+ Tri *const c_panel = reinterpret_cast<Tri *>(working_space_bytes + (threadid * get_c_working_size()));
+
+ // Shared buffers - these come either from BufferManager or _B_transposed.
+ const Toi *b_panel;
+
+ if(pretransposed)
+ {
+ b_panel = _B_transposed;
+ }
+
+ //printf("Starting GEMM loop, x_block=%d, k_block=%d\n", _x_block, _k_block);
+
+ // newkblock() is always true on the first iteration, so this will be set properly on the first loop.
+ int kern_k = 0;
+
+ for(; !current.done(); current.advance())
+ {
+ if(current.newkblock())
+ {
+ prof(PROFILE_PREPA, ((m_max - m_0) * (current.kmax() - current.k0()) * sizeof(Toi)), [&](void)
+ {
+ if(_trA ^ strategy::A_transpose)
+ {
+ Transform<strategy::A_interleave, strategy::A_block, true>(a_panel, this->_Aptr, this->_lda, m_0, m_max, current.k0(), current.kmax());
+ }
+ else
+ {
+ Transform<strategy::A_interleave, strategy::A_block, false>(a_panel, this->_Aptr, this->_lda, m_0, m_max, current.k0(), current.kmax());
+ }
+ });
+
+ // Figure out how many "K" the kernel will actually process.
+ kern_k = iceildiv(current.kmax() - current.k0(), strategy::k_unroll);
+ kern_k *= strat.k_unroll;
+ }
+
+ int bblocks = iceildiv(current.xmax() - current.x0(), strategy::out_width);
+
+ if(!pretransposed)
+ {
+ /* Look ahead to the next block and populate it if necessary.
+ * This avoids the populate operation becoming a bottleneck, and
+ * helps keep the threads synchronized (the first thread to get
+ * here will populate while the rest will advance).
+ *
+ * If we are running single threaded, bm->try_populate() will do
+ * nothing.
+ */
+ if(next.advance())
+ {
+ _bm->try_populate(next.index(), [&](void *buffer)
+ {
+ prof(PROFILE_PREPB, (next.xmax() - next.x0()) * (next.kmax() - next.k0()) * sizeof(Toi), [&](void)
+ {
+ Toi *b_panel = reinterpret_cast<Toi *>(buffer);
+ if(_trB ^ strategy::B_transpose)
+ {
+ Transform<strategy::B_interleave, strategy::B_block, true>(b_panel, this->_Bptr, this->_ldb, next.x0(), next.xmax(), next.k0(), next.kmax());
+ }
+ else
+ {
+ Transform<strategy::B_interleave, strategy::B_block, false>(b_panel, this->_Bptr, this->_ldb, next.x0(), next.xmax(), next.k0(), next.kmax());
+ }
+ });
+ });
+ }
+
+ /* Get the buffer for this iteration from the BufferManager. */
+ b_panel = reinterpret_cast<Toi *>(_bm->get(current.index(), [&](void *bpv)
+ {
+ prof(PROFILE_PREPB, (current.xmax() - current.x0()) * (current.kmax() - current.k0()) * sizeof(Toi), [&](void)
+ {
+ Toi *b_panel = reinterpret_cast<Toi *>(bpv);
+ if(_trB ^ strategy::B_transpose)
+ {
+ Transform<strategy::B_interleave, strategy::B_block, true>(b_panel, this->_Bptr, this->_ldb, current.x0(), current.xmax(), current.k0(), current.kmax());
+ }
+ else
+ {
+ Transform<strategy::B_interleave, strategy::B_block, false>(b_panel, this->_Bptr, this->_ldb, current.x0(), current.xmax(), current.k0(), current.kmax());
+ }
+ });
+ }));
+ }
+
+ /* Do the actual work. */
+ for(unsigned int y = m_0; y < m_max; y += strategy::out_height)
+ {
+ unsigned int ymax = std::min(_Msize, y + strategy::out_height);
+
+ prof(PROFILE_KERNEL, (strategy::out_height * bblocks * strategy::out_width * kern_k), [&](void)
+ {
+ strat.kernel(a_panel + ((y - m_0) * kern_k), b_panel, c_panel, 1, bblocks, kern_k);
+ });
+ prof(PROFILE_MERGE, (strategy::out_height * bblocks * strategy::out_width * sizeof(Tr)), [&](void)
+ {
+ MergeResults<strategy::out_width, strategy::out_height>(this->_Cptr, c_panel, this->_ldc, y, ymax,
+ current.x0(), current.xmax(), _alpha, (current.k0() == 0 ? _beta : static_cast<Tr>(1)));
+ });
+ }
+
+ if(pretransposed)
+ {
+ b_panel += (bblocks * strat.out_width * kern_k);
+ }
+ else
+ {
+ _bm->release(current.index());
+ }
+ }
+ }
+
+public:
+ GemmInterleaved(GemmInterleaved &) = delete;
+ GemmInterleaved &operator=(GemmInterleaved &) = delete;
+
+ /* Constructor */
+ GemmInterleaved(const CPUInfo *ci, const unsigned int M, const unsigned int N, const unsigned int K,
+ const bool trA, const bool trB, const Tr alpha, const Tr beta, const int maxthreads,
+ const bool pretransposed)
+ : _ci(ci), _Msize(M), _Nsize(N), _Ksize(K), _trA(trA), _trB(trB), _alpha(alpha), _beta(beta), _maxthreads(maxthreads), _pretransposed(pretransposed)
+ {
+ const unsigned int L1_size = ci->get_L1_cache_size();
+ const unsigned int L2_size = ci->get_L2_cache_size();
+
+ assert(maxthreads > 0);
+
+ // Work out blocking parameters
+
+ // k_block: Find out how much of the larger array can be loaded into half the cache.
+ // This should account for associative caches.
+ _k_block = (L1_size / 2) / (sizeof(Toi) * (std::max(strategy::out_width, strategy::out_height)));
+
+ // Needs to be (at least a single) multiple of the K unroll level.
+ _k_block /= strategy::k_unroll;
+ _k_block = std::max(_k_block, 1U) * strategy::k_unroll;
+
+ // Now tune to presented problem size; this is how many blocks we need.
+ int num_k_blocks = iceildiv(K, _k_block);
+
+ // So divide the space equally into that many blocks.
+ _k_block = iceildiv(K, num_k_blocks);
+
+ // And round UP to the K unroll level required.
+ _k_block = iceildiv(_k_block, strategy::k_unroll);
+ _k_block *= strategy::k_unroll;
+
+ // x_block: Work out how many rows (of length k_block) will fit in the L2
+ // Don't allocate more than 90% of the L2 to allow for overheads, and subtract off the L1 contents.
+ _x_block = (((L2_size * 9) / 10) - (_k_block * sizeof(Toi) * (strategy::out_width + strategy::out_height))) / (sizeof(Toi) * _k_block);
+
+ // Needs to be (at least a single) multiple of the kernel output width.
+ _x_block /= strategy::out_width;
+ _x_block = std::max(_x_block, 1U) * strategy::out_width;
+
+ // And tune to the presented problem size.
+ int num_x_blocks = iceildiv(N, _x_block);
+ _x_block = iceildiv(N, num_x_blocks);
+
+ _x_block = iceildiv(_x_block, strategy::out_width);
+ _x_block *= strategy::out_width;
+
+ // Work out the rounded size of M - needed for some buffers.
+ _Mround = iceildiv(M, strategy::out_height);
+ _Mround *= strategy::out_height;
+ }
+
+ // Interface implementation - Compulsory functions
+
+ // Window size: Only the last thread should do a ragged block, so dole out work in units of out_height */
+ unsigned int get_window_size() const override
+ {
+ // _Mround is a multiple of out_height by definition.
+ return _Mround / strategy::out_height;
+ }
+
+ // set_nthreads: pass on to buffer manager to avoid it waiting for non-existant threads.
+ void set_nthreads(int nthreads) override
+ {
+ if(_bm)
+ {
+ _bm->set_nthreads(nthreads);
+ }
+ }
+
+ // Execute
+ void execute(unsigned int start, unsigned int end, int threadid) override
+ {
+ if(_pretransposed)
+ {
+ execute_internal<true>(start, end, threadid);
+ }
+ else
+ {
+ execute_internal<false>(start, end, threadid);
+ }
+ }
+
+ // Interface implementation - working space
+ size_t get_working_size() const override
+ {
+ // In all cases, we need one A buffer plus a C buffer per thread.
+ size_t size = get_a_working_size() + (get_c_working_size() * _maxthreads);
+
+ // For pretransposed case, there is no working space needed for B.
+ // Otherwise, we need a BufferManager.
+ if(!_pretransposed)
+ {
+ size += BufferManager::get_storage_requirement(_maxthreads, get_b_working_size());
+ }
+
+ size += 64; // Add on a cache line extra for alignment.
+
+ return size;
+ }
+
+ void set_working_space(void *working_space) override
+ {
+ // Make sure everything ends up cache line aligned
+ int8_t *working_space_bytes = reinterpret_cast<int8_t *>(working_space);
+ intptr_t working_space_int = reinterpret_cast<intptr_t>(working_space);
+
+ size_t diff = 0;
+
+ if(working_space_int & 0x3F)
+ {
+ diff = 0x40 - (working_space_int & 0x3F);
+ }
+
+ working_space_bytes += diff;
+
+ if(_pretransposed)
+ {
+ // Pretransposed case: just set internal pointer to parameter value.
+ _working_space = reinterpret_cast<void *>(working_space_bytes);
+ }
+ else
+ {
+ // Otherwise, use the first part of the working space for the buffer manager.
+ // It's legal to call this again so don't leak a buffer manager if it already existed.
+ delete _bm;
+
+ _bm = new BufferManager(_maxthreads, get_b_working_size(), reinterpret_cast<void *>(working_space_bytes));
+
+ working_space_bytes += BufferManager::get_storage_requirement(_maxthreads, get_b_working_size());
+
+ _working_space = reinterpret_cast<void *>(working_space_bytes);
+ }
+ }
+
+ // Interface implementation - pretransposed
+ bool B_is_pretransposed() const override
+ {
+ return _pretransposed;
+ }
+
+ bool B_pretranspose_required() const override
+ {
+ return _pretransposed && (_B_transposed == nullptr);
+ }
+
+ // TODO: this could almost certainly be considerably simpler.
+ size_t get_B_pretransposed_array_size() const override
+ {
+ size_t total = 0;
+ blockwalker current(_Ksize, _k_block, _Nsize, _x_block);
+
+ do
+ {
+ /* Figure out the size of each block. */
+ size_t x_size = (current.xmax() - current.x0());
+ size_t k_size = (current.kmax() - current.k0());
+
+ /* Round sizes up as needed. */
+ x_size = iceildiv(x_size, strategy::out_width);
+ x_size *= strategy::out_width;
+
+ k_size = iceildiv(k_size, strategy::k_unroll);
+ k_size *= strategy::k_unroll;
+
+ total += x_size * k_size * sizeof(Toi);
+ }
+ while(current.advance());
+
+ return total;
+ }
+
+ void pretranspose_B_array(void *in_buffer, const To *B, const int ldb) override
+ {
+ blockwalker current(_Ksize, _k_block, _Nsize, _x_block);
+ Toi *buffer = reinterpret_cast<Toi *>(in_buffer);
+ _B_transposed = buffer;
+
+ do
+ {
+ /* Figure out the size of each block. */
+ size_t x_size = (current.xmax() - current.x0());
+ size_t k_size = (current.kmax() - current.k0());
+
+ /* Round sizes up as needed. */
+ x_size = iceildiv(x_size, strategy::out_width);
+ x_size *= strategy::out_width;
+
+ k_size = iceildiv(k_size, strategy::k_unroll);
+ k_size *= strategy::k_unroll;
+
+ if(_trB ^ strategy::B_transpose)
+ {
+ Transform<strategy::B_interleave, strategy::B_block, true>(buffer, B, ldb, current.x0(), current.xmax(), current.k0(), current.kmax());
+ }
+ else
+ {
+ Transform<strategy::B_interleave, strategy::B_block, false>(buffer, B, ldb, current.x0(), current.xmax(), current.k0(), current.kmax());
+ }
+
+ buffer += (x_size * k_size);
+ }
+ while(current.advance());
+ }
+
+ ~GemmInterleaved() override
+ {
+ delete _bm;
+ }
+};
+
+} // namespace arm_gemm