Keep pkt header (Xilinx#1125)

Signed-off-by: Abhishek Varma <abhvarma@amd.com>
Co-authored-by: Javier Setoain <jsetoain@users.noreply.github.com>
Co-authored-by: James Newling <james.newling@gmail.com>
Co-authored-by: Maksim Levental <maksim.levental@gmail.com>
Co-authored-by: Joseph Melber <jgmelber@gmail.com>
Co-authored-by: Abhishek Varma <abhvarma@amd.com>
Co-authored-by: erwei-xilinx <erweiw@xilinx.com>
Co-authored-by: AndraBisca <andrab@amd.com>

Started writing the objfifo intro tutorial

Vectorize vec scalar (Xilinx#1135)

Added new programming guide section placedholders (Xilinx#1138)

[EXAMPLE] An element-wise add example (Xilinx#1148)

Co-authored-by: pjr <pjr@xilinx.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Jack Lo <jack.lo@amd.com>

Moved objfifo design example 1 to programming_guide sections 3.

Started objectFifo programming guide

Continue section 3 guide.

ObjFifo guide: access patterns

Add example to objFifo guide

Update objfifo guide

Separate section 3 of the guide into 3 subsections

[ASPLOS] Weight expand asplos (Xilinx#1158)

Co-authored-by: pjr <pjr@xilinx.com>
Co-authored-by: Phil James-Roxby <pjr@amd.com>

Reorganize subsections in section 3 of the guide.

Updated sections 3a and 3b

Pjr vector exp (Xilinx#1166)

Co-authored-by: pjr <pjr@xilinx.com>
Co-authored-by: Joseph Melber <jgmelber@gmail.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

[SOFTMAX] Single column rapid test (Xilinx#1168)

Co-authored-by: pjr <pjr@xilinx.com>
Co-authored-by: Joseph Melber <jgmelber@gmail.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

Vector softmax (Xilinx#1172)

Co-authored-by: pjr <pjr@xilinx.com>

[MERGE] This has gone horribly wrong, so fixing up in place

Update exp.cc

Merge resolved

Added new programming guide section placedholders (Xilinx#1174)

Swapped section-2 and section-3 (Xilinx#1177)

Update section-1 and section-3 examples (Xilinx#1179)

Update section-3 (Xilinx#1181)

Reorganize tutorials and reference_designs to programming_examples (Xilinx#1182)

Fix for mmult lit (Xilinx#1187)

Revert "Fix for mmult lit" (Xilinx#1188)

[ASPLOS][WIP] python host code example (Xilinx#1185)

Extract arg parse from test.py (Xilinx#1192)

Add finished write-up for sections 2a and 2b

Add objectfifo bindings to quick references

Update section2 subsection list

[ASPLOS][WIP] initial version of asplos24 tutorial description (Xilinx#1184)

Co-authored-by: Jack Lo <jack.lo@amd.com>

Fix for lit tests (Xilinx#1189)

Rename tutorials folder (Xilinx#1190) (Xilinx#1197)

Co-authored-by: AndraBisca <andrab@amd.com>
Co-authored-by: Jack Lo <36210336+jackl-xilinx@users.noreply.github.com>

[ASPLOS] Rename directories (Xilinx#1196)

Co-authored-by: Jeff Fifield <jeff.fifield@amd.com>
Co-authored-by: AndraBisca <andrab@amd.com>
Co-authored-by: Jack Lo <36210336+jackl-xilinx@users.noreply.github.com>

Add section 2c. Update tiles in sections 2a and 2b.

Add generic aie array description paragraph (Xilinx#1191)

Co-authored-by: Joseph Melber <jgmelber@gmail.com>
Co-authored-by: Jack Lo <jack.lo@amd.com>

ReLU with tracing (Xilinx#1204)

ReLU example with tracing

Co-authored-by: pjr <pjr@xilinx.com>
Co-authored-by: Joseph Melber <jgmelber@gmail.com>

Ml eltwise add and mul (Xilinx#1207)

Move around of the eltwise add (put it in ml) and a new eltwise mul kernel

Co-authored-by: pjr <pjr@xilinx.com>
Co-authored-by: Jeff Fifield <jeff.fifield@amd.com>

Moved test_lib to runtime_lib/test_lib for now

Pjr reduce (Xilinx#1222) Reduce programming examples

Co-authored-by: pjr <pjr@xilinx.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

[ASPLOS][WIP] Passthrough kernel in basic examples (Xilinx#1216)

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

fix paths run.lit passthrough _kernel (Xilinx#1225)

Fixed CMakeLists.txt reference to test_utils.h (Xilinx#1223)

Minor CMakeLists.txt and Makefile fixes for programming_examples (Xilinx#1227)

aie_kernels/aie2/add.cc

@@ -0,0 +1,61 @@
+//===- scale.cc -------------------------------------------------*- C++ -*-===//
+//
+// This file is licensed under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+// Copyright (C) 2023, Advanced Micro Devices, Inc.
+//
+//===----------------------------------------------------------------------===//
+
+#define __AIENGINE__ 2
+#define NOCPP
+#define __AIEARCH__ 20
+
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <type_traits>
+
+#include <aie_api/aie.hpp>
+
+template <typename T_in, typename T_out, const int N>
+void eltwise_add(T_in *a, T_in *b, T_out *c) {
+  for (int i = 0; i < N; i++) {
+    c[i] = a[i] + b[i];
+  }
+}
+
+template <typename T_in, typename T_out, const int N>
+void eltwise_vadd(T_in *a, T_in *b, T_out *c) {
+
+  constexpr int vec_factor = 16;
+  event0();
+  T_in *__restrict pA1 = a;
+  T_in *__restrict pB1 = b;
+  T_out *__restrict pC1 = c;
+  const int F = N / vec_factor;
+  for (int i = 0; i < F; i++)
+    chess_prepare_for_pipelining chess_loop_range(16, ) {
+      aie::vector<T_in, vec_factor> A0 = aie::load_v<vec_factor>(pA1);
+      pA1 += vec_factor;
+      aie::vector<T_in, vec_factor> B0 = aie::load_v<vec_factor>(pB1);
+      pB1 += vec_factor;
+      aie::vector<T_out, vec_factor> cout = aie::add(A0, B0);
+      aie::store_v(pC1, cout);
+      pC1 += vec_factor;
+    }
+  event1();
+}
+
+extern "C" {
+
+void eltwise_add_bf16_scalar(bfloat16 *a_in, bfloat16 *b_in, bfloat16 *c_out) {
+  eltwise_add<bfloat16, bfloat16, 1024>(a_in, b_in, c_out);
+}
+
+void eltwise_add_bf16_vector(bfloat16 *a_in, bfloat16 *b_in, bfloat16 *c_out) {
+  eltwise_vadd<bfloat16, bfloat16, 1024>(a_in, b_in, c_out);
+}
+
+} // extern "C"

programming_examples/vision/vision_kernels/passThrough.cc → aie_kernels/aie_generic/passThrough.cc

@@ -15,30 +15,21 @@
 #include <stdio.h>
 #include <stdlib.h>
 
-#define REL_WRITE 0
-#define REL_READ 1
-
 #include <aie_api/aie.hpp>
 
 template <typename T, int N>
 __attribute__((noinline)) void passThrough_aie(T *restrict in, T *restrict out,
                                                const int32_t height,
                                                const int32_t width) {
-  //::aie::vector<T, N> data_out;
-  //::aie::mask<N> temp_val;
+  event0();
+
   v64uint8 *restrict outPtr = (v64uint8 *)out;
   v64uint8 *restrict inPtr = (v64uint8 *)in;
 
   for (int j = 0; j < (height * width); j += N) // Nx samples per loop
-    chess_prepare_for_pipelining chess_loop_range(6, ) {
-      //::aie::vector<T, N> tmpVector = ::aie::load_v(in);
-      //::aie::store_v(out, tmpVector);
-
-      *outPtr++ = *inPtr++;
+    chess_prepare_for_pipelining chess_loop_range(6, ) { *outPtr++ = *inPtr++; }
 
-      // in += N;
-      // out += N;
-    }
+  event1();
 }
 
 extern "C" {

aie_kernels/generic/vector_max.cc

@@ -0,0 +1,53 @@
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <type_traits>
+
+#include <aie_api/aie.hpp>
+
+void vector(int32_t *restrict in, int32_t *restrict out) {
+
+  v16int32 tiny = broadcast_to_v16int32((int32_t)-2147483648);
+  int32_t input_size = 1024;
+  int32_t vector_size = 16;
+  v16int32 after_vector;
+  v16int32 running_max = tiny;
+  for (int32_t i = 0; i < input_size; i += vector_size)
+    chess_prepare_for_pipelining chess_loop_range(64, 64) {
+      v16int32 next = *(v16int32 *)(in + i);
+      v16int32 test = max(running_max, next);
+      running_max = test;
+    }
+  after_vector = running_max;
+  v16int32 first = shift_bytes(after_vector, after_vector, 32);
+  v16int32 second = max(after_vector, first);
+  v16int32 second_shift = shift_bytes(second, second, 16);
+  v16int32 third = max(second, second_shift);
+  v16int32 third_shift = shift_bytes(third, third, 8);
+  v16int32 fourth = max(third, third_shift);
+  v16int32 fourth_shift = shift_bytes(fourth, fourth, 4);
+  v16int32 fifth = max(fourth, fourth_shift);
+  int32_t last = extract_elem(fifth, 0);
+  *(int32_t *)out = last;
+  return;
+}
+
+void scalar(int32_t *restrict in, int32_t *restrict out) {
+  size_t input_size = 1024;
+  int32_t running_max = (int32_t)-2147483648;
+  for (int32_t i = 0; i < input_size; i++) {
+    if (in[i] > running_max)
+      running_max = in[i];
+  }
+  *(int32_t *)out = running_max;
+
+  return;
+}
+
+extern "C" {
+
+void vector_max(int32_t *a_in, int32_t *c_out) { vector(a_in, c_out); }
+
+void scalar_max(int32_t *a_in, int32_t *c_out) { scalar(a_in, c_out); }
+
+} // extern "C"

aie_kernels/generic/vector_min.cc

@@ -0,0 +1,53 @@
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <type_traits>
+
+#include <aie_api/aie.hpp>
+
+void vector(int32_t *restrict in, int32_t *restrict out) {
+
+  v16int32 massive = broadcast_to_v16int32((int32_t)2147483647);
+  int32_t input_size = 1024;
+  int32_t vector_size = 16;
+  v16int32 after_vector;
+  v16int32 running_min = massive;
+  for (int32_t i = 0; i < input_size; i += vector_size)
+    chess_prepare_for_pipelining chess_loop_range(64, 64) {
+      v16int32 next = *(v16int32 *)(in + i);
+      v16int32 test = min(running_min, next);
+      running_min = test;
+    }
+  after_vector = running_min;
+  v16int32 first = shift_bytes(after_vector, after_vector, 32);
+  v16int32 second = min(after_vector, first);
+  v16int32 second_shift = shift_bytes(second, second, 16);
+  v16int32 third = min(second, second_shift);
+  v16int32 third_shift = shift_bytes(third, third, 8);
+  v16int32 fourth = min(third, third_shift);
+  v16int32 fourth_shift = shift_bytes(fourth, fourth, 4);
+  v16int32 fifth = min(fourth, fourth_shift);
+  int32_t last = extract_elem(fifth, 0);
+  *(int32_t *)out = last;
+  return;
+}
+
+void scalar(int32_t *restrict in, int32_t *restrict out) {
+  size_t input_size = 1024;
+  int32_t running_min = (int32_t)2147483647;
+  for (int32_t i = 0; i < input_size; i++) {
+    if (in[i] < running_min)
+      running_min = in[i];
+  }
+  *(int32_t *)out = running_min;
+
+  return;
+}
+
+extern "C" {
+
+void vector_min(int32_t *a_in, int32_t *c_out) { vector(a_in, c_out); }
+
+void scalar_min(int32_t *a_in, int32_t *c_out) { scalar(a_in, c_out); }
+
+} // extern "C"

aie_kernels/relu.cc

@@ -0,0 +1,41 @@
+//===- scale.cc -------------------------------------------------*- C++ -*-===//
+//
+// This file is licensed under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+// Copyright (C) 2023, Advanced Micro Devices, Inc.
+//
+//===----------------------------------------------------------------------===//
+
+#define __AIENGINE__ 2
+#define NOCPP
+#define __AIEARCH__ 20
+
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <type_traits>
+
+#include <aie_api/aie.hpp>
+
+void relu(bfloat16 *restrict a, bfloat16 *restrict c, const int TILE_SIZE) {
+  const int v_factor = 32;
+  v32bfloat16 zeroes = broadcast_zero_bfloat16();
+
+  event0();
+  for (size_t i = 0; i < TILE_SIZE; i += v_factor)
+    chess_prepare_for_pipelining chess_loop_range(32, 32) {
+      v32bfloat16 input = *(v32bfloat16 *)(a + i);
+      v32bfloat16 output = max(input, zeroes);
+      *(v32bfloat16 *)(c + i) = output;
+    }
+  event1();
+  return;
+}
+
+extern "C" {
+
+void bf16_relu(bfloat16 *a_in, bfloat16 *c_out) { relu(a_in, c_out, 1024); }
+
+} // extern "C"

programming_examples/basic/eltwise_exp/Makefile

@@ -0,0 +1,53 @@
+##===- Makefile -----------------------------------------------------------===##
+# 
+# This file licensed under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+# 
+##===----------------------------------------------------------------------===##
+
+include ../../makefile-common
+
+all: build/final.xclbin
+
+targetname = eltwise_exp
+
+build/lut_based_ops.o:
+	mkdir -p ${@D}
+	cd ${@D} &&	xchesscc_wrapper ${CHESSCCWRAP2_FLAGS} -I. -c ../../../../aie_runtime_lib/AIE2/lut_based_ops.cpp -o ${@F}
+
+build/exp.o: 
+	mkdir -p ${@D}
+	cd ${@D} && xchesscc_wrapper ${CHESSCCWRAP2_FLAGS} -I. -I ../../../../aie_runtime_lib/AIE2 -c ../../../../aie_kernels/aie2/bf16_exp.cc -o ${@F}
+
+build/kernels.a: build/exp.o build/lut_based_ops.o
+	ar rvs $@ $+
+
+build/aie.mlir: aie2.py
+	mkdir -p ${@D}
+	python3 $< > $@
+
+build/final.xclbin: build/aie.mlir build/kernels.a
+	mkdir -p ${@D}
+	cd ${@D} && aiecc.py --aie-generate-cdo --aie-generate-ipu --no-compile-host \
+		--xclbin-name=${@F} --ipu-insts-name=insts.txt ${<F}
+
+${targetname}.exe: test.cpp
+	rm -rf _build
+	mkdir -p _build
+	cd _build && ${powershell} cmake .. -DTARGET_NAME=${targetname} 
+	cd _build && ${powershell} cmake --build . --config Release
+ifeq "${powershell}" "powershell.exe"
+	cp _build/${targetname}.exe $@
+else
+	cp _build/${targetname} $@ 
+endif
+
+run: ${targetname}.exe build/final.xclbin build/insts.txt 
+	${powershell} ./$< -x build/final.xclbin -i build/insts.txt -k MLIR_AIE
+
+run_py: build/final.xclbin build/insts.txt
+	${powershell} python3 test.py -x build/final.xclbin -i build/insts.txt -k MLIR_AIE
+
+clean:
+	rm -rf build _build ${targetname}.exe

programming_examples/basic/eltwise_mul/CMakeLists.txt

@@ -13,9 +13,14 @@
 # cmake needs this line
 cmake_minimum_required(VERSION 3.1)
 
+set(CMAKE_CXX_STANDARD 23) 
+set(CMAKE_CXX_STANDARD_REQUIRED YES)
+
 find_program(WSL NAMES powershell.exe)
 
 if (NOT WSL)
+    set(CMAKE_C_COMPILER gcc-13)
+    set(CMAKE_CXX_COMPILER g++-13)
     set(BOOST_ROOT /usr/include/boost CACHE STRING "Path to Boost install")
     set(XRT_INC_DIR /opt/xilinx/xrt/include CACHE STRING "Path to XRT cloned repo")
     set(XRT_LIB_DIR /opt/xilinx/xrt/lib CACHE STRING "Path to xrt_coreutil.lib")
@@ -40,6 +45,7 @@ project(${ProjectName})
 find_package(Boost REQUIRED)
 
 add_executable(${currentTarget}
+    ${CMAKE_CURRENT_SOURCE_DIR}/../../../runtime_lib/test_lib/test_utils.cpp
     test.cpp
 )
 
@@ -48,7 +54,7 @@ target_compile_definitions(${currentTarget} PUBLIC DISABLE_ABI_CHECK=1)
 target_include_directories (${currentTarget} PUBLIC 
     ${XRT_INC_DIR}
     ${Boost_INCLUDE_DIRS}
-    ../../../programming_examples/utils
+    ${CMAKE_CURRENT_SOURCE_DIR}/../../../runtime_lib/test_lib
 )
 
 target_link_directories(${currentTarget} PUBLIC

programming_examples/basic/eltwise_mul/Makefile

@@ -6,15 +6,15 @@
 # 
 ##===----------------------------------------------------------------------===##
 
-include ../../../programming_examples/basic/makefile-common
+include ../../makefile-common
 
 all: build/final.xclbin
 
 targetname = myEltwiseMul
 
 build/mul.o:
 	mkdir -p ${@D}
-	cd ${@D} &&	xchesscc_wrapper ${CHESSCCWRAP2_FLAGS} -I. -c ${REPO_ROOT}/aie_kernels/aie2/mul.cc -o ${@F}
+	cd ${@D} &&	xchesscc_wrapper ${CHESSCCWRAP2_FLAGS} -I. -c ../../../../aie_kernels/aie2/mul.cc -o ${@F}
 
 build/aie.mlir: aie2.py
 	mkdir -p ${@D}
@@ -28,8 +28,7 @@ build/final.xclbin: build/aie.mlir build/mul.o
 ${targetname}.exe: test.cpp
 	rm -rf _build
 	mkdir -p _build
-#	cd _build && ${powershell} cmake .. -DTARGET_NAME=${targetname}
-	cd _build && ${powershell} cmake -E env CXXFLAGS="-std=c++23 -ggdb" cmake .. -D CMAKE_C_COMPILER=gcc-13 -D CMAKE_CXX_COMPILER=g++-13 -DTARGET_NAME=${targetname} -Dsubdir=${subdir}
+	cd _build && ${powershell} cmake .. -DTARGET_NAME=${targetname}
 	cd _build && ${powershell} cmake --build . --config Release
 ifeq "${powershell}" "powershell.exe"
 	cp _build/${targetname}.exe $@

programming_examples/basic/eltwise_mul/test.py

@@ -9,7 +9,7 @@
 import sys
 import time
 
-sys.path.append("../../programming_examples/utils")
+sys.path.append("../../python")
 import test_utils
 
 # ------------------------------------------------------