卷积函数的FPGA实现（五）对IPcore进行HLS及bug查找

最新推荐文章于 2024-05-31 15:46:53 发布
最新推荐文章于 2024-05-31 15:46:53 发布
阅读量5.1k 收藏 9
点赞数 5
分类专栏： FPGA MTCNN 机器学习
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/weixin_36474809/article/details/84832760
版权
机器学习 同时被 3 个专栏收录
133 篇文章 51 订阅
订阅专栏
FPGA
61 篇文章 102 订阅
订阅专栏
MTCNN
20 篇文章 7 订阅
订阅专栏

背景：我们编辑了3x3卷积的IPcore，并完成了预编译。程序通过调用3*3卷积的IPcore实现运行。并通过HLS预编译指令实现为硬件结构，现在我们需要对IPcore程序进行HLS。

目的：对卷积IPcore进行HLS

目录

一、testconvBench编写

1.1  linux下用cmake编译运行程序

1.2 隐患与BUG

1.3 testBench编写

  卷积尺寸

 卷积与结果对比

二、c-simulation

  出现bug更改流程

三、几个bug与解决

3.1 reg格式问题

3.2 关于DRAM接口的问题

3.3 DATAFLOW的错误

3.4 调试N_PE的问题

四、Bug位置查找

4.1 processInputChannel

 function instantiate

 WBRAM

  Loop 'L_CH_OUT'  in 'processAll_channelOut'

 OBRAM没有生成RTL端口

4.2 整个IPcore的HLS console

一、testconvBench编写

原程序需要调用OpenCV并且调用次数过多，无法当作HLS的testBench，我们需要编写简单的testBench，先确保IPcore无误且可用。

1.1  linux下用cmake编译运行程序

HLS_test文件夹，里面文件夹src放入相应程序。HLS_test文件夹创建CMakeList.txt文件

cmake_minimum_required(VERSION 2.8)
project(main)

set(CUDA_USE_STATIC_CUDA_RUNTIME OFF)
set(QMAKE_CXXFLAGS "-std=c++11")

AUX_SOURCE_DIRECTORY(./src DIR_SRCS)
add_executable(test_convBench ${DIR_SRCS})

第一个表示cmake的最低版本，project表示编译的是主程序文件来生成可执行文件。set是编译器的类型。add_executable表示生成的可执行文件的名字和位置。

xxr@gpu-SYS-7048GR-TR:~/Desktop/xxr2/HLS_test$ cmake .
-- The C compiler identification is GNU 5.4.0
-- The CXX compiler identification is GNU 5.4.0
-- Check for working C compiler: /usr/bin/cc
-- Check for working C compiler: /usr/bin/cc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Detecting C compile features
-- Detecting C compile features - done
-- Check for working CXX compiler: /usr/bin/c++
-- Check for working CXX compiler: /usr/bin/c++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Configuring done
-- Generating done
-- Build files have been written to: /home/xxr/Desktop/xxr2/HLS_test
xxr@gpu-SYS-7048GR-TR:~/Desktop/xxr2/HLS_test$ make
Scanning dependencies of target test_convBench
[ 25%] Building CXX object CMakeFiles/test_convBench.dir/src/pBox.cpp.o
[ 50%] Building CXX object CMakeFiles/test_convBench.dir/src/fpgaAcc.cpp.o
[ 75%] Building CXX object CMakeFiles/test_convBench.dir/src/test_convBench.cpp.o
[100%] Linking CXX executable test_convBench
[100%] Built target test_convBench
xxr@gpu-SYS-7048GR-TR:~/Desktop/xxr2/HLS_test$ ./test_convBench
test SUCCESS!

1.2 隐患与BUG

void convolution_3x3(const Weight *weightIn, const pBox *pboxIn, pBox *outpBox)

我们先确定我们卷积的形式。我们将卷积形式定为下表格式，输入和权重随机生成。在PS端和IPcore端都进行3*3卷积验证输出是否正确。

InputSizeKernelSizeStridePaddingOutputSize

25*25*64

3*3*64

2

Valid（no padding）

12*12*64

两个相关的知识点：

  • outputSize=（InSize-Kernel）/Stride   +1;这是一个不需要pad的卷积。
  • 卷积中bias的数量，与输出通道的个数相等。（代表着所有的三维的卷积核相乘累加之后，加了一个相同的bias，即使在输出通道的不同位置，只要输出通道相同，就加相同的bias）

出现BUG原因：

  • stride在两种卷积之中出现问题（此bug需要用tensorFlow输出的权重调试）
  • OBRAM的尺寸不够
  • IBRAM的尺寸不够
  • WBRAM的尺寸不够

1.3 testBench编写

经过上面bug，我们需要确定，测试的网络结构不能过大免得OBRAM不够用的情况。

  卷积尺寸

经过运行与检验（卷积的Stride有误conv3*3与convlution对不上，stride=1时可以对上，stride为2的时候不能对上。流程较长，难以排查，需要等到后面输出参数再重新调试）暂时将stride设为1看结果。

InputSizeKernelSizeStridePaddingOutputSize

24*24*64

3*3*64

1

Valid（no padding）

22*22*64
//conv parameters
int inputSize=23; int inChannelNum=32;
int outputSize=21;  int OutChannelNum=64;
int kernelSize=3; int Stride=1;
int Input_Pixels=inputSize*inputSize*inChannelNum;
int Output_Pixels=outputSize*outputSize*OutChannelNum;
int weightkernel_Pixels=9*inChannelNum*OutChannelNum;

//conv variable
Weight weightIn;
pBox featureIn;
pBox conv_PL_out;
pBox conv_PS_out;

//initialize conv weight variable
weightIn.out_ChannelNum=OutChannelNum;
weightIn.in_ChannelNum=inChannelNum;
weightIn.kernelSize=kernelSize;
weightIn.stride=Stride;
weightIn.leftPad=0;
weightIn.rightPad=0;
weightIn.pdata=(float *)malloc(sizeof(float)*weightkernel_Pixels);
//weightIn.pbias=(float *)malloc(sizeof(float)*OutChannelNum);
for (int i=0;i<weightkernel_Pixels;i++){
	weightIn.pdata[i]=(rand()%100)/100.0;
	//weightIn.pdata[i]=(i%10)/10;
}

//initialize conv Input variable
featureIn.width=inputSize;
featureIn.height=inputSize;
featureIn.channel=inChannelNum;
featureIn.pdata=(float*)malloc(sizeof(float)*Input_Pixels);
for (int i=0;i<Input_Pixels;i++){
	featureIn.pdata[i]=(rand()%100)/100.0;
	//featureIn.pdata[i]=(i%10)/10;
}

//initialize conv Output variable
conv_PL_out.width=outputSize;
conv_PL_out.height=outputSize;
conv_PL_out.channel=OutChannelNum;
conv_PS_out.width=outputSize;
conv_PS_out.height=outputSize;
conv_PS_out.channel=OutChannelNum;	
conv_PS_out.pdata=(float*)malloc(sizeof(float)*Output_Pixels);
conv_PL_out.pdata=(float*)malloc(sizeof(float)*Output_Pixels);

根据卷积生成相应的网络尺寸。并且开辟相应的内存空间。将权重与卷积的值设为随机生成的值。

 卷积与结果对比

将IPcore的卷积，与实际的卷积进行对比，来判断相应的结果是否一致。

//conv in PS
convolution(&weightIn,&featureIn,&conv_PS_out);

//conv in PL
convolution_3x3(&weightIn,&featureIn,&conv_PL_out);

//compare in PS and PL
int error=0;
for(int i=0;i<Output_Pixels;i++){
	if(conv_PS_out.pdata[i]!=conv_PL_out.pdata[i]){
		printf("Convolution ERROR!\n");
		printf("i is %d, value in PS= %f, in PL is %f\n",i,conv_PS_out.pdata[i],conv_PL_out.pdata[i]);
		error=1;
	}
}
printf("Compare DONE SUCCESS!\n");
if(error==0)
	printf("PS and PL conv match SUCCESS!\n");
else
	printf("PS and PL conv match FAILURE!\n");

二、c-simulation

相关内容：FPGA实践教程（一）用HLS将c程序生成IPcore https://blog.csdn.net/weixin_36474809/article/details/80597166

服务器上HLS-test，将相应IPcore运行成功，即可直接放入HLS进行c-simulation。

INFO: [HLS 200-10] Setting target device to 'xc7z035ffg676-2'
INFO: [SIM 211-2] *************** CSIM start ***************
INFO: [SIM 211-4] CSIM will launch GCC as the compiler.
   Compiling ../../../../src/test_convBench.cpp in debug mode
   Compiling ../../../../src/fpgaAcc.cpp in debug mode
   Generating csim.exe
Test Start SUCCESS!
Variable init SUCCESS!
Conv in PS SUCCESS!
33convConv in PL SUCCESS!
Compare DONE SUCCESS!
PS and PL conv match SUCCESS!
INFO: [SIM 211-1] CSim done with 0 errors.

  出现bug更改流程

  • 修改后的文件拷入服务器FPGA-mtcnn之中，编译运行成功。表明程序可以与原程序嵌套运行成功
  • 这个文件拷入服务器HLS_test之中，编译运行成功。表明可以与testBench嵌套运行成功
  • 这个文件放入虚拟机src文件夹之中，更新HLS之中c-simulation得到相同结果即可synthesis

三、几个bug与解决

3.1 reg格式问题

3 errors generated.
ERROR: [HLS 200-70] Compilation errors found:
Pragma processor failed: In file included from src/fpgaAcc.cpp:1:
src/fpgaAcc.cpp:166:11: error: use of undeclared identifier 'reg'
 float px=reg(input_ptr[load_pixel_offset+in_channel_pixel_offset]);
          ^
src/fpgaAcc.cpp:173:15: error: use of undeclared identifier 'reg'
 float read = reg(weight_DRAM_ptr[weight_loc]);
              ^
src/fpgaAcc.cpp:208:12: error: use of undeclared identifier 'reg'
  float px=reg(ImageCache::get_IBRAM_Pixel(IBRAM_line_offset,pixel_col_to_load,
           ^
3 errors generated.
Failed checking during preprocessing.
    while executing
"source /home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1/csynth.tcl"
    invoked from within
"hls::main /home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1/csynth.tcl"
    ("uplevel" body line 1)
    invoked from within
"uplevel 1 hls::main {*}$args"
    (procedure "hls_proc" line 5)
    invoked from within
"hls_proc $argv"

.h文件表示c文件，.hpp文件表示c++文件，可以将相应的.h文件改为.hpp文件。但是改了之后依然是此bug。

可能与reg的定义之前加了相应的#ifndef __SYNTHESIS__有关。不懂为什么zynqNet要加这个指令。我们将此指令删掉。此bug解决。

3.2 关于DRAM接口的问题

INFO: [XFORM 203-603] Inlining function 'MemoryController::writeBackOutputChannel' into 'convolution_3x3' (src/fpgaAcc.cpp:107).
INFO: [HLS 200-111] Finished Standard Transforms Time (s): cpu = 00:00:34 ; elapsed = 00:00:26 . Memory (MB): peak = 361.926 ; gain = 13.668 ; free physical = 607 ; free virtual = 33200
INFO: [HLS 200-10] Checking synthesizability ...
INFO: [XFORM 203-602] Inlining function 'ImageCache::writeNextChannelPixel_2_IBRAM' into 'convolution_3x3' (src/fpgaAcc.cpp:336->src/fpgaAcc.cpp:327->src/fpgaAcc.cpp:84) automatically.
ERROR: [SYNCHK 200-11] src/fpgaAcc.cpp:259: Argument 'weightIn.pdata' of function 'convolution_3x3' (src/fpgaAcc.cpp:45) has an unsynthesizable type (possible cause(s): pointer to pointer or global pointer).
ERROR: [SYNCHK 200-61] src/fpgaAcc.cpp:174: unsupported memory access on variable 'weightIn.pdata' which is (or contains) an array with unknown size at compile time.
INFO: [SYNCHK 200-10] 2 error(s), 0 warning(s).
ERROR: [HLS 200-70] Synthesizability check failed.
command 'ap_source' returned error code
    while executing

ERROR: [SYNCHK 200-11] src/fpgaAcc.cpp:259: Argument 'weightIn.pdata' of function 'convolution_3x3' (src/fpgaAcc.cpp:45) has an unsynthesizable type (possible cause(s): pointer to pointer or global pointer).

weightIn.pdata这个包含着不能被HLS综合的类型，例如指针指向的指针，或者全局变量指针。

ERROR: [SYNCHK 200-61] src/fpgaAcc.cpp:174: unsupported memory access on variable 'weightIn.pdata' which is (or contains) an array with unknown size at compile time.

weightIn.pdata是一个（或者包含）不知大小的数组。

因此我们必须添加预编译指令对接口进行综合。

 MTCNN的FPGA实现（四）接口的HLS https://blog.csdn.net/weixin_36474809/article/details/84940846

3.3 DATAFLOW的错误

WARNING: [XFORM 203-562] Loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'processAll_channelOu' has unknown bound because it has multiple exiting blocks.
WARNING: [XFORM 203-713] Function 'processInputChannel..1' (src/fpgaAcc.cpp:226:1) failed dataflow checking:  A dataflow region cannot be instantiated from with a pipelined loop  (src/fpgaAcc.cpp:226:1). Ignoring pipeline directive to allow the dataflow directive to take precedence. This behavior can be disabled by using 'config_compile -disable_dataflow_pipeline_check'.
Instruction does not dominate all uses!
  %tmp_60 = add i32 %WeightsCache_inChan_1, %tmp_59
  %memorybus_addr_rd_re = call i1 @_ssdm_op_ReadReq.m_axi.floatP(float* %memorybus_addr, i32 %tmp_60), !dbg !1031
Broken module found, compilation aborted!
Stack dump:
0.	Running pass 'Function Pass Manager' on module '/home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1/.autopilot/db/a.o.2.bc'.
1.	Running pass 'Module Verifier' on function '@convolution_3x3'
/mnt/workspace/Xilinx/Vivado/2017.4/bin/loader: line 194: 15937 Aborted                 (core dumped) "$RDI_PROG" "$@"
Finished C synthesis.

重要报错：

WARNING: [XFORM 203-713] Function 'processInputChannel..1' (src/fpgaAcc.cpp:226:1) failed dataflow checking:  A dataflow region cannot be instantiated from with a pipelined loop  (src/fpgaAcc.cpp:226:1). Ignoring pipeline directive to allow the dataflow directive to take precedence. This behavior can be disabled by using 'config_compile -disable_dataflow_pipeline_check'.
Instruction does not dominate all uses!
  %tmp_60 = add i32 %WeightsCache_inChan_1, %tmp_59
  %memorybus_addr_rd_re = call i1 @_ssdm_op_ReadReq.m_axi.floatP(float* %memorybus_addr, i32 %tmp_60), !dbg !1031
Broken module found, compilation aborted!
Stack dump:
0.    Running pass 'Function Pass Manager' on module '/home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1/.autopilot/db/a.o.2.bc'.
1.    Running pass 'Module Verifier' on function '@convolution_3x3'
/mnt/workspace/Xilinx/Vivado/2017.4/bin/loader: line 194: 15937 Aborted                 (core dumped) "$RDI_PROG" "$@"

3.4 调试N_PE的问题

关于此，我们发现很有可能许多优化指令都没有添加成功。因为面板之中有一些这种报错，不知道是否添加成功。

 根据后面的console面板信息，我们发现可能是成功的，因为两点

  • zynqNet的IPcore在编译的过程之中也出现了上面的感叹号warning，但是成功输出RTL
  • 输出信息：INFO: [XFORM 203-101] Partitioning array 'OutputCache::OBRAM'  in dimension 1 with a cyclic factor 8.
WARNING: [XFORM 203-631] Renaming function 'ProcessingElement::processAll_channelOut' to 'processAll_channelOu' (src/fpgaAcc.cpp:192:43)
INFO: [XFORM 203-811] Inferring bus burst read of variable length on port 'memorybus' (src/fpgaAcc.cpp:178:15).
WARNING: [XFORM 203-562] Loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'processAll_channelOu' has unknown bound because it has multiple exiting blocks.
Instruction does not dominate all uses!
  %tmp_64 = add i32 %WeightsCache_inChan_1, %tmp_63
  %memorybus_addr_rd_re = call i1 @_ssdm_op_ReadReq.m_axi.floatP(float* %memorybus_addr, i32 %tmp_64), !dbg !1031
Broken module found, compilation aborted!
Stack dump:
0.	Running pass 'Function Pass Manager' on module '/home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1/.autopilot/db/a.o.2.bc'.
1.	Running pass 'Module Verifier' on function '@convolution_3x3'
/mnt/workspace/Xilinx/Vivado/2017.4/bin/loader: line 194: 35285 Aborted                 (core dumped) "$RDI_PROG" "$@"
Finished C synthesis.

在processAll_channelOut之中，是否展开系数N_PE这个优化指令能被运用上，我们直接将N_PE改为16，依然此报错。

将pipeline II=1删掉，依然同样报错。

四、Bug位置查找

嵌套IPcore过大，需要将其改小，单独的单元来进行测试。我们将processInputChannel设为top function，然后获得实验结果：

4.1 processInputChannel

Starting C synthesis ...
/mnt/workspace/Xilinx/Vivado/2017.4/bin/vivado_hls /home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1/csynth.tcl
INFO: [HLS 200-10] Running '/mnt/workspace/Xilinx/Vivado/2017.4/bin/unwrapped/lnx64.o/vivado_hls'
INFO: [HLS 200-10] For user 'osrc' on host 'osrc-virtual-machine' (Linux_x86_64 version 4.13.0-32-generic) on Wed Dec 12 10:37:09 CST 2018
INFO: [HLS 200-10] On os Ubuntu 16.04.3 LTS
INFO: [HLS 200-10] In directory '/home/osrc/Desktop/document/conv_Core/HLS_Conv'
INFO: [HLS 200-10] Opening project '/home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore'.
INFO: [HLS 200-10] Adding design file 'src/fpgaAcc.cpp' to the project
INFO: [HLS 200-10] Adding design file 'src/fpgaAcc.hpp' to the project
INFO: [HLS 200-10] Adding design file 'src/pBox.cpp' to the project
INFO: [HLS 200-10] Adding design file 'src/pBox.h' to the project
INFO: [HLS 200-10] Adding test bench file 'src/test_convBench.cpp' to the project
INFO: [HLS 200-10] Opening solution '/home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1'.
INFO: [SYN 201-201] Setting up clock 'default' with a period of 10ns.
INFO: [HLS 200-10] Setting target device to 'xc7z035ffg676-2'
INFO: [HLS 200-10] Analyzing design file 'src/pBox.cpp' ...
INFO: [HLS 200-10] Analyzing design file 'src/fpgaAcc.cpp' ...
INFO: [HLS 200-10] Validating synthesis directives ...
INFO: [HLS 200-111] Finished Checking Pragmas Time (s): cpu = 00:00:31 ; elapsed = 00:00:19 . Memory (MB): peak = 361.637 ; gain = 13.375 ; free physical = 395 ; free virtual = 32671
INFO: [HLS 200-111] Finished Linking Time (s): cpu = 00:00:33 ; elapsed = 00:00:21 . Memory (MB): peak = 361.637 ; gain = 13.375 ; free physical = 393 ; free virtual = 32671
INFO: [HLS 200-10] Starting code transformations ...
INFO: [XFORM 203-603] Inlining function 'ImageCache::calcu_IBRAM_row_offset' into 'ProcessingElement::loadPixel_buffer' (src/fpgaAcc.cpp:209).
INFO: [XFORM 203-603] Inlining function 'ImageCache::get_IBRAM_Pixel' into 'ProcessingElement::loadPixel_buffer' (src/fpgaAcc.cpp:213).
INFO: [XFORM 203-603] Inlining function 'ProcessingElement::loadPixel_buffer' into 'ProcessingElement::processInputChannel' (src/fpgaAcc.cpp:230).
INFO: [XFORM 203-603] Inlining function 'WeightsCache::get_WBRAM_addr' into 'WeightsCache::get_9_weights_to_buffer' (src/fpgaAcc.cpp:307).
INFO: [XFORM 203-603] Inlining function 'WeightsCache::get_9_weights_to_buffer' into 'ProcessingElement::processAll_channelOut' (src/fpgaAcc.cpp:247).
INFO: [XFORM 203-603] Inlining function 'ProcessingElement::macc2d' into 'ProcessingElement::processAll_channelOut' (src/fpgaAcc.cpp:249).
INFO: [XFORM 203-603] Inlining function 'OutputCache::setOutChannel' into 'OutputCache::accumulateChannel' (src/fpgaAcc.cpp:384).
INFO: [XFORM 203-603] Inlining function 'OutputCache::setOutChannel' into 'ProcessingElement::processAll_channelOut' (src/fpgaAcc.cpp:252).
INFO: [XFORM 203-603] Inlining function 'OutputCache::getOutChannel' into 'OutputCache::accumulateChannel' (src/fpgaAcc.cpp:382).
INFO: [XFORM 203-603] Inlining function 'OutputCache::accumulateChannel' into 'ProcessingElement::processAll_channelOut' (src/fpgaAcc.cpp:254).
WARNING: [XFORM 203-623] Cannot instantiate function 'ProcessingElement::processInputChannel'(src/fpgaAcc.cpp:225:1) for 'cur_ci' since none of the actual argument(s) of 'cur_ci' are constant or global.
INFO: [HLS 200-111] Finished Standard Transforms Time (s): cpu = 00:00:35 ; elapsed = 00:00:23 . Memory (MB): peak = 361.910 ; gain = 13.648 ; free physical = 383 ; free virtual = 32663
INFO: [HLS 200-10] Checking synthesizability ...
INFO: [HLS 200-111] Finished Checking Synthesizability Time (s): cpu = 00:00:35 ; elapsed = 00:00:23 . Memory (MB): peak = 361.910 ; gain = 13.648 ; free physical = 380 ; free virtual = 32661
INFO: [XFORM 203-502] Unrolling all sub-loops inside loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'ProcessingElement::processAll_channelOut' for pipelining.
INFO: [XFORM 203-501] Unrolling loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'ProcessingElement::processAll_channelOut' partially with a factor of 16.
INFO: [XFORM 203-501] Unrolling loop 'Loop-1.1' (src/fpgaAcc.cpp:308) in function 'ProcessingElement::processAll_channelOut' completely.
INFO: [XFORM 203-501] Unrolling loop 'L_MACC_multiply' (src/fpgaAcc.cpp:190) in function 'ProcessingElement::processAll_channelOut' completely.
INFO: [XFORM 203-501] Unrolling loop 'L_MACC_accumulate' (src/fpgaAcc.cpp:195) in function 'ProcessingElement::processAll_channelOut' completely.
INFO: [XFORM 203-101] Partitioning array 'pixel_buffer'  in dimension 1 completely.
INFO: [XFORM 203-101] Partitioning array 'weights_local' (src/fpgaAcc.cpp:244) in dimension 1 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM'  in dimension 1 completely.
INFO: [XFORM 203-101] Partitioning array 'multresult' (src/fpgaAcc.cpp:187) in dimension 1 completely.
INFO: [XFORM 203-101] Partitioning array 'OutputCache::OBRAM'  in dimension 1 with a cyclic factor 8.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.0'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.1'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.2'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.3'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.4'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.5'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.6'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.7'  in dimension 2 completely.
WARNING: [XFORM 203-623] Cannot instantiate function 'ProcessingElement::processInputChannel'(src/fpgaAcc.cpp:225:1) for 'cur_ci' since none of the actual argument(s) of 'cur_ci' are constant or global.
INFO: [HLS 200-111] Finished Pre-synthesis Time (s): cpu = 00:00:37 ; elapsed = 00:00:25 . Memory (MB): peak = 489.633 ; gain = 141.371 ; free physical = 353 ; free virtual = 32635
WARNING: [XFORM 203-631] Renaming function 'ProcessingElement::processAll_channelOut' to 'processAll_channelOu' (src/fpgaAcc.cpp:241:43)
WARNING: [XFORM 203-562] Loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'processAll_channelOu' has unknown bound because it has multiple exiting blocks.
INFO: [HLS 200-111] Finished Architecture Synthesis Time (s): cpu = 00:00:38 ; elapsed = 00:00:27 . Memory (MB): peak = 489.633 ; gain = 141.371 ; free physical = 349 ; free virtual = 32632
INFO: [HLS 200-10] Starting hardware synthesis ...
INFO: [HLS 200-10] Synthesizing 'ProcessingElement::processInputChannel' ...
WARNING: [SYN 201-103] Top function name 'ProcessingElement::processInputChannel' is not a legal RTL name.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-103] Top function name 'ProcessingElement::processInputChannel' is not a legal RTL name and is changed to 'ProcessingElement_processInputChannel'; this may result in automatic C/RTL co-simulation failure.
INFO: [HLS 200-10] ----------------------------------------------------------------
INFO: [HLS 200-42] -- Implementing module 'processAll_channelOu'
INFO: [HLS 200-10] ----------------------------------------------------------------
INFO: [SCHED 204-11] Starting scheduling ...
INFO: [SCHED 204-61] Pipelining loop 'L_CH_OUT'.
WARNING: [SCHED 204-69] Unable to schedule 'store' operation (src/fpgaAcc.cpp:395->src/fpgaAcc.cpp:384->src/fpgaAcc.cpp:254) of variable 'new_ch', src/fpgaAcc.cpp:383->src/fpgaAcc.cpp:254 on array 'OBRAM_0' due to limited memory ports. Please consider using a memory core with more ports or partitioning the array 'OBRAM_0'.
INFO: [SCHED 204-61] Pipelining result : Target II = 1, Final II = 2, Depth = 10.
INFO: [SCHED 204-11] Finished scheduling.
INFO: [HLS 200-111]  Elapsed time: 27.98 seconds; current allocated memory: 89.244 MB.
INFO: [BIND 205-100] Starting micro-architecture generation ...
INFO: [BIND 205-101] Performing variable lifetime analysis.
INFO: [BIND 205-101] Exploring resource sharing.
INFO: [BIND 205-101] Binding ...
INFO: [BIND 205-100] Finished micro-architecture generation.
INFO: [HLS 200-111]  Elapsed time: 0.59 seconds; current allocated memory: 90.872 MB.
INFO: [HLS 200-10] ----------------------------------------------------------------
INFO: [HLS 200-42] -- Implementing module 'ProcessingElement_processInputChannel'
INFO: [HLS 200-10] ----------------------------------------------------------------
INFO: [SCHED 204-11] Starting scheduling ...
INFO: [SCHED 204-11] Finished scheduling.
INFO: [HLS 200-111]  Elapsed time: 0.4 seconds; current allocated memory: 90.982 MB.
INFO: [BIND 205-100] Starting micro-architecture generation ...
INFO: [BIND 205-101] Performing variable lifetime analysis.
INFO: [BIND 205-101] Exploring resource sharing.
INFO: [BIND 205-101] Binding ...
INFO: [BIND 205-100] Finished micro-architecture generation.
INFO: [HLS 200-111]  Elapsed time: 0.08 seconds; current allocated memory: 91.045 MB.
INFO: [HLS 200-10] ----------------------------------------------------------------
INFO: [HLS 200-10] -- Generating RTL for module 'processAll_channelOu'
INFO: [HLS 200-10] ----------------------------------------------------------------
INFO: [SYN 201-210] Renamed object name 'processAll_channelOu_OBRAM_0' to 'processAll_channebkb' due to the length limit 20
INFO: [SYN 201-210] Renamed object name 'processAll_channelOu_OBRAM_1' to 'processAll_channecud' due to the length limit 20
INFO: [SYN 201-210] Renamed object name 'processAll_channelOu_OBRAM_2' to 'processAll_channedEe' due to the length limit 20
INFO: [SYN 201-210] Renamed object name 'processAll_channelOu_OBRAM_3' to 'processAll_channeeOg' due to the length limit 20
INFO: [SYN 201-210] Renamed object name 'processAll_channelOu_OBRAM_4' to 'processAll_channefYi' due to the length limit 20
INFO: [SYN 201-210] Renamed object name 'processAll_channelOu_OBRAM_5' to 'processAll_channeg8j' due to the length limit 20
INFO: [SYN 201-210] Renamed object name 'processAll_channelOu_OBRAM_6' to 'processAll_channehbi' due to the length limit 20
INFO: [SYN 201-210] Renamed object name 'processAll_channelOu_OBRAM_7' to 'processAll_channeibs' due to the length limit 20
INFO: [SYN 201-210] Renamed object name 'ProcessingElement_processInputChannel_fadd_32ns_32ns_32_4_full_dsp_1' to 'ProcessingElementjbC' due to the length limit 20
INFO: [RTGEN 206-100] Generating core module 'ProcessingElementjbC': 8 instance(s).
INFO: [RTGEN 206-100] Finished creating RTL model for 'processAll_channelOu'.
INFO: [HLS 200-111]  Elapsed time: 0.92 seconds; current allocated memory: 93.369 MB.
INFO: [HLS 200-10] ----------------------------------------------------------------
INFO: [HLS 200-10] -- Generating RTL for module 'ProcessingElement_processInputChannel'
INFO: [HLS 200-10] ----------------------------------------------------------------
INFO: [RTGEN 206-500] Setting interface mode on port 'ProcessingElement_processInputChannel/cur_row_times_stride' to 'ap_none'.
INFO: [RTGEN 206-500] Setting interface mode on port 'ProcessingElement_processInputChannel/cur_col_times_stride' to 'ap_none'.
INFO: [RTGEN 206-500] Setting interface mode on port 'ProcessingElement_processInputChannel/cur_ci' to 'ap_none'.
INFO: [RTGEN 206-500] Setting interface mode on port 'ProcessingElement_processInputChannel/out_channelNum' to 'ap_none'.
INFO: [RTGEN 206-500] Setting interface mode on function 'ProcessingElement_processInputChannel' to 'ap_ctrl_hs'.
WARNING: [RTGEN 206-101] Global array 'OBRAM_0' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Global array 'OBRAM_1' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Global array 'OBRAM_2' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Global array 'OBRAM_3' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Global array 'OBRAM_4' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Global array 'OBRAM_5' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Global array 'OBRAM_6' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Global array 'OBRAM_7' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Port 'ProcessingElement_processInputChannel/cur_row_times_stride' has no fanin or fanout and is left dangling.
               Please use C simulation to confirm this function argument can be read from or written to.
WARNING: [RTGEN 206-101] Port 'ProcessingElement_processInputChannel/cur_col_times_stride' has no fanin or fanout and is left dangling.
               Please use C simulation to confirm this function argument can be read from or written to.
INFO: [RTGEN 206-100] Finished creating RTL model for 'ProcessingElement_processInputChannel'.
INFO: [HLS 200-111]  Elapsed time: 1.04 seconds; current allocated memory: 97.566 MB.
INFO: [RTMG 210-278] Implementing memory 'processAll_channebkb_ram (RAM_T2P_BRAM)' using block RAMs with power-on initialization.
INFO: [HLS 200-111] Finished generating all RTL models Time (s): cpu = 00:00:42 ; elapsed = 00:00:32 . Memory (MB): peak = 489.633 ; gain = 141.371 ; free physical = 320 ; free virtual = 32614
INFO: [SYSC 207-301] Generating SystemC RTL for ProcessingElement_processInputChannel.
INFO: [VHDL 208-304] Generating VHDL RTL for ProcessingElement_processInputChannel.
INFO: [VLOG 209-307] Generating Verilog RTL for ProcessingElement_processInputChannel.
INFO: [HLS 200-112] Total elapsed time: 32.18 seconds; peak allocated memory: 97.566 MB.
Finished C synthesis.

其中需要注意的问题：

 function instantiate

WARNING: [XFORM 203-623] Cannot instantiate function 'ProcessingElement::processInputChannel'(src/fpgaAcc.cpp:225:1) for 'cur_ci' since none of the actual argument(s) of 'cur_ci' are constant or global.

此报错出现了两次，但是zynqNet在只HLS inputchannel函数时候也会出此报错。

在整个IPcore实现时没有出现此报错。

 WBRAM

WBRAM的报错相同。

WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.
WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.

包括在进行数组分开的时候，WBRAM已经与zynqNet展现出不同，下标少了。

INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.7'  in dimension 2 completely.

zynqNet的INFO为：

INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.15.2'  in dimension 2 completely.

但是后面出现了同样的报错：可能是进行分开之后WBRAM的名字发生了改变。

WARNING: [SYN 201-303] Cannot apply memory assignment of 'RAM_S2P_BRAM' (src/fpgaAcc.cpp:305->src/fpgaAcc.cpp:247): 'WBRAM_0_0' does not exist or is optimized away.

  Loop 'L_CH_OUT'  in 'processAll_channelOut'

MTCNN会比zynqNet多了一个警告：

WARNING: [XFORM 203-562] Loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'processAll_channelOu' has unknown bound because it has multiple exiting blocks.

 OBRAM没有生成RTL端口

可能因为OBRAM分成的与尺寸不匹配，最终OBRAM比zynqNet的console多了：

WARNING: [RTGEN 206-101] Global array 'OBRAM_0' will not be exposed as RTL port.
WARNING: [RTGEN 206-101] Global array 'OBRAM_1' will not be exposed as RTL port. 。。。

4.2 整个IPcore的HLS console

Starting C synthesis ...
/mnt/workspace/Xilinx/Vivado/2017.4/bin/vivado_hls /home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1/csynth.tcl
INFO: [HLS 200-10] Running '/mnt/workspace/Xilinx/Vivado/2017.4/bin/unwrapped/lnx64.o/vivado_hls'
INFO: [HLS 200-10] For user 'osrc' on host 'osrc-virtual-machine' (Linux_x86_64 version 4.13.0-32-generic) on Tue Dec 11 18:46:57 CST 2018
INFO: [HLS 200-10] On os Ubuntu 16.04.3 LTS
INFO: [HLS 200-10] In directory '/home/osrc/Desktop/document/conv_Core/HLS_Conv'
INFO: [HLS 200-10] Opening project '/home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore'.
INFO: [HLS 200-10] Adding design file 'src/fpgaAcc.cpp' to the project
INFO: [HLS 200-10] Adding design file 'src/fpgaAcc.hpp' to the project
INFO: [HLS 200-10] Adding design file 'src/pBox.cpp' to the project
INFO: [HLS 200-10] Adding design file 'src/pBox.h' to the project
INFO: [HLS 200-10] Adding test bench file 'src/test_convBench.cpp' to the project
INFO: [HLS 200-10] Opening solution '/home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1'.
INFO: [SYN 201-201] Setting up clock 'default' with a period of 10ns.
INFO: [HLS 200-10] Setting target device to 'xc7z035ffg676-2'
INFO: [HLS 200-10] Analyzing design file 'src/pBox.cpp' ...
INFO: [HLS 200-10] Analyzing design file 'src/fpgaAcc.cpp' ...
INFO: [HLS 200-10] Validating synthesis directives ...
INFO: [HLS 200-111] Finished Checking Pragmas Time (s): cpu = 00:00:31 ; elapsed = 00:00:20 . Memory (MB): peak = 361.641 ; gain = 13.375 ; free physical = 320 ; free virtual = 32652
INFO: [HLS 200-111] Finished Linking Time (s): cpu = 00:00:33 ; elapsed = 00:00:21 . Memory (MB): peak = 361.641 ; gain = 13.375 ; free physical = 318 ; free virtual = 32651
INFO: [HLS 200-10] Starting code transformations ...
INFO: [XFORM 203-603] Inlining function 'MemoryController::setLayerConfig' into 'convolution_3x3' (src/fpgaAcc.cpp:77).
INFO: [XFORM 203-603] Inlining function 'ImageCache::setLayerConfig' into 'convolution_3x3' (src/fpgaAcc.cpp:78).
INFO: [XFORM 203-603] Inlining function 'WeightsCache::setLayerConfig' into 'convolution_3x3' (src/fpgaAcc.cpp:79).
INFO: [XFORM 203-603] Inlining function 'WeightsCache::get_WBRAM_addr' into 'WeightsCache::get_9_weights_to_buffer' (src/fpgaAcc.cpp:307).
INFO: [XFORM 203-603] Inlining function 'WeightsCache::get_WBRAM_addr' into 'WeightsCache::load_WBRAM_from_DRAM' (src/fpgaAcc.cpp:284).
INFO: [XFORM 203-603] Inlining function 'MemoryController::load_weight_2_reg' into 'WeightsCache::load_WBRAM_from_DRAM' (src/fpgaAcc.cpp:291).
INFO: [XFORM 203-603] Inlining function 'WeightsCache::load_WBRAM_from_DRAM' into 'convolution_3x3' (src/fpgaAcc.cpp:83).
INFO: [XFORM 203-603] Inlining function 'MemoryController::setPixelLoadRowOffset' into 'convolution_3x3' (src/fpgaAcc.cpp:94).
INFO: [XFORM 203-603] Inlining function 'MemoryController::setPixelLoadRowOffset' into 'convolution_3x3' (src/fpgaAcc.cpp:87).
INFO: [XFORM 203-603] Inlining function 'MemoryController::setPixelLoadRowOffset' into 'convolution_3x3' (src/fpgaAcc.cpp:85).
INFO: [XFORM 203-603] Inlining function 'MemoryController::setPixelLoadOffset' into 'ImageCache::loadRowDRAM_2_IBRAM' (src/fpgaAcc.cpp:330).
INFO: [XFORM 203-603] Inlining function 'MemoryController::loadInputChannelPixel' into 'ImageCache::loadPixelDRAM_2_IBRAM' (src/fpgaAcc.cpp:339).
INFO: [XFORM 203-603] Inlining function 'ImageCache::loadPixelDRAM_2_IBRAM' into 'ImageCache::loadRowDRAM_2_IBRAM' (src/fpgaAcc.cpp:331).
INFO: [XFORM 203-603] Inlining function 'ImageCache::loadRowDRAM_2_IBRAM' into 'convolution_3x3' (src/fpgaAcc.cpp:95).
INFO: [XFORM 203-603] Inlining function 'ImageCache::loadRowDRAM_2_IBRAM' into 'convolution_3x3' (src/fpgaAcc.cpp:88).
INFO: [XFORM 203-603] Inlining function 'ImageCache::loadRowDRAM_2_IBRAM' into 'convolution_3x3' (src/fpgaAcc.cpp:86).
INFO: [XFORM 203-603] Inlining function 'MemoryController::setPixelOutOffset' into 'convolution_3x3' (src/fpgaAcc.cpp:99).
INFO: [XFORM 203-603] Inlining function 'ImageCache::calcu_IBRAM_row_offset' into 'ProcessingElement::loadPixel_buffer' (src/fpgaAcc.cpp:209).
INFO: [XFORM 203-603] Inlining function 'ImageCache::get_IBRAM_Pixel' into 'ProcessingElement::loadPixel_buffer' (src/fpgaAcc.cpp:213).
INFO: [XFORM 203-603] Inlining function 'ProcessingElement::loadPixel_buffer' into 'ProcessingElement::processInputChannel' (src/fpgaAcc.cpp:230).
INFO: [XFORM 203-603] Inlining function 'WeightsCache::get_9_weights_to_buffer' into 'ProcessingElement::processAll_channelOut' (src/fpgaAcc.cpp:247).
INFO: [XFORM 203-603] Inlining function 'ProcessingElement::macc2d' into 'ProcessingElement::processAll_channelOut' (src/fpgaAcc.cpp:249).
INFO: [XFORM 203-603] Inlining function 'OutputCache::setOutChannel' into 'OutputCache::accumulateChannel' (src/fpgaAcc.cpp:384).
INFO: [XFORM 203-603] Inlining function 'OutputCache::setOutChannel' into 'ProcessingElement::processAll_channelOut' (src/fpgaAcc.cpp:252).
INFO: [XFORM 203-603] Inlining function 'OutputCache::getOutChannel' into 'OutputCache::accumulateChannel' (src/fpgaAcc.cpp:382).
INFO: [XFORM 203-603] Inlining function 'OutputCache::accumulateChannel' into 'ProcessingElement::processAll_channelOut' (src/fpgaAcc.cpp:254).
INFO: [XFORM 203-603] Inlining function 'MemoryController::writeBackOutputChannel' into 'convolution_3x3' (src/fpgaAcc.cpp:109).
INFO: [HLS 200-111] Finished Standard Transforms Time (s): cpu = 00:00:34 ; elapsed = 00:00:23 . Memory (MB): peak = 361.930 ; gain = 13.664 ; free physical = 307 ; free virtual = 32642
INFO: [HLS 200-10] Checking synthesizability ...
INFO: [XFORM 203-602] Inlining function 'ImageCache::writeNextChannelPixel_2_IBRAM' into 'convolution_3x3' (src/fpgaAcc.cpp:340->src/fpgaAcc.cpp:331->src/fpgaAcc.cpp:86) automatically.
INFO: [HLS 200-111] Finished Checking Synthesizability Time (s): cpu = 00:00:35 ; elapsed = 00:00:23 . Memory (MB): peak = 361.930 ; gain = 13.664 ; free physical = 303 ; free virtual = 32639
INFO: [XFORM 203-502] Unrolling all sub-loops inside loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'ProcessingElement::processAll_channelOut' for pipelining.
INFO: [XFORM 203-501] Unrolling loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'ProcessingElement::processAll_channelOut' partially with a factor of 8.
INFO: [XFORM 203-501] Unrolling loop 'Loop-1.1' (src/fpgaAcc.cpp:308) in function 'ProcessingElement::processAll_channelOut' completely.
INFO: [XFORM 203-501] Unrolling loop 'L_MACC_multiply' (src/fpgaAcc.cpp:190) in function 'ProcessingElement::processAll_channelOut' completely.
INFO: [XFORM 203-501] Unrolling loop 'L_MACC_accumulate' (src/fpgaAcc.cpp:195) in function 'ProcessingElement::processAll_channelOut' completely.
INFO: [XFORM 203-101] Partitioning array 'pixel_buffer' (src/fpgaAcc.cpp:228) in dimension 1 completely.
INFO: [XFORM 203-101] Partitioning array 'weights_local' (src/fpgaAcc.cpp:244) in dimension 1 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM'  in dimension 1 completely.
INFO: [XFORM 203-101] Partitioning array 'multresult' (src/fpgaAcc.cpp:187) in dimension 1 completely.
INFO: [XFORM 203-101] Partitioning array 'OutputCache::OBRAM'  in dimension 1 with a cyclic factor 8.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.0'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.1'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.2'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.3'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.4'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.5'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.6'  in dimension 2 completely.
INFO: [XFORM 203-101] Partitioning array 'WeightsCache::WBRAM.7'  in dimension 2 completely.
INFO: [XFORM 203-602] Inlining function 'ImageCache::writeNextChannelPixel_2_IBRAM' into 'convolution_3x3' (src/fpgaAcc.cpp:340->src/fpgaAcc.cpp:331->src/fpgaAcc.cpp:86) automatically.
INFO: [XFORM 203-622] Instantiating function 'ProcessingElement::processInputChannel'(src/fpgaAcc.cpp:221) to 'ProcessingElement::processInputChannel.0' at call site (src/fpgaAcc.cpp:103) by setting 'cur_ci' to 'cur_channel_in'.
INFO: [XFORM 203-721] Changing loop 'Loop_load_pixel_2_PE_row_loop_proc' (src/fpgaAcc.cpp:207) to a process function for dataflow in function 'ProcessingElement::processInputChannel.0'.
INFO: [XFORM 203-712] Applying dataflow to function 'ProcessingElement::processInputChannel.0' (src/fpgaAcc.cpp:224:1), detected/extracted 2 process function(s): 
	 'ProcessingElement::processInputChannel.0_Loop_load_pixel_2_PE_row_loop_proc5'
	 'ProcessingElement::processAll_channelOut'.
INFO: [HLS 200-111] Finished Pre-synthesis Time (s): cpu = 00:00:37 ; elapsed = 00:00:26 . Memory (MB): peak = 489.637 ; gain = 141.371 ; free physical = 275 ; free virtual = 32614
WARNING: [XFORM 203-542] Cannot flatten a loop nest 'Loop-1.1' (src/fpgaAcc.cpp:283:18) in function 'convolution_3x3' : 


the outer loop is not a perfect loop.
WARNING: [XFORM 203-542] Cannot flatten a loop nest 'Loop-1' (src/fpgaAcc.cpp:280:18) in function 'convolution_3x3' : 


the outer loop is not a perfect loop.
WARNING: [XFORM 203-542] Cannot flatten a loop nest 'L_DRAM_PRELOADROW_X' (src/fpgaAcc.cpp:329:77) in function 'convolution_3x3' : 


the outer loop is not a perfect loop.
WARNING: [XFORM 203-542] Cannot flatten a loop nest 'L_DRAM_PRELOADROW_X' (src/fpgaAcc.cpp:329:77) in function 'convolution_3x3' : 


the outer loop is not a perfect loop.
WARNING: [XFORM 203-542] Cannot flatten a loop nest 'L_DRAM_PRELOADROW_X' (src/fpgaAcc.cpp:329:77) in function 'convolution_3x3' : 


the outer loop is not a perfect loop.
WARNING: [XFORM 203-542] Cannot flatten a loop nest 'Loop-4.1' (src/fpgaAcc.cpp:93:3) in function 'convolution_3x3' : 


the outer loop is not a perfect loop.
WARNING: [XFORM 203-542] Cannot flatten a loop nest 'row_loop' (src/fpgaAcc.cpp:91:85) in function 'convolution_3x3' : 


more than one sub loop.
WARNING: [XFORM 203-631] Renaming function 'ProcessingElement::processInputChannel.0_Loop_load_pixel_2_PE_row_loop_proc5' to 'processInputChannel.' (src/fpgaAcc.cpp:207:3)
WARNING: [XFORM 203-631] Renaming function 'ProcessingElement::processInputChannel.0' to 'processInputChannel..1' (src/fpgaAcc.cpp:226:1)
WARNING: [XFORM 203-631] Renaming function 'ProcessingElement::processAll_channelOut' to 'processAll_channelOu' (src/fpgaAcc.cpp:192:43)
INFO: [XFORM 203-811] Inferring bus burst read of variable length on port 'memorybus' (src/fpgaAcc.cpp:178:15).
WARNING: [XFORM 203-562] Loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'processAll_channelOu' has unknown bound because it has multiple exiting blocks.
WARNING: [XFORM 203-713] Function 'processInputChannel..1' (src/fpgaAcc.cpp:226:1) failed dataflow checking:  A dataflow region cannot be instantiated from with a pipelined loop  (src/fpgaAcc.cpp:226:1). Ignoring pipeline directive to allow the dataflow directive to take precedence. This behavior can be disabled by using 'config_compile -disable_dataflow_pipeline_check'.
Instruction does not dominate all uses!
  %tmp_60 = add i32 %WeightsCache_inChan_1, %tmp_59
  %memorybus_addr_rd_re = call i1 @_ssdm_op_ReadReq.m_axi.floatP(float* %memorybus_addr, i32 %tmp_60), !dbg !1031
Broken module found, compilation aborted!
Stack dump:
0.	Running pass 'Function Pass Manager' on module '/home/osrc/Desktop/document/conv_Core/HLS_Conv/conv3x3_IPcore/solution1/.autopilot/db/a.o.2.bc'.
1.	Running pass 'Module Verifier' on function '@convolution_3x3'
/mnt/workspace/Xilinx/Vivado/2017.4/bin/loader: line 194: 15937 Aborted                 (core dumped) "$RDI_PROG" "$@"
Finished C synthesis.

重要的错误来自两点：

unknow bound和dataflow不能实现。

WARNING: [XFORM 203-562] Loop 'L_CH_OUT' (src/fpgaAcc.cpp:241) in function 'processAll_channelOu' has unknown bound because it has multiple exiting blocks.
WARNING: [XFORM 203-713] Function 'processInputChannel..1' (src/fpgaAcc.cpp:226:1) failed dataflow checking:  A dataflow region cannot be instantiated from with a pipelined loop  (src/fpgaAcc.cpp:226:1). Ignoring pipeline directive to allow the dataflow directive to take precedence. This behavior can be disabled by using 'config_compile -disable_dataflow_pipeline_check'.

后续需要对这些BUG进行调试。初步判断BUG为processAll_channelOut之中的for循环展开的问题与OBRAM的分开有差别的问题。

祥瑞Coding
关注
  • 5
    点赞
  • 踩
  • 9
    收藏
    觉得还不错? 一键收藏
  • 打赏
    打赏
  • 1
    评论
专栏目录
基于HLS的高效深度卷积神经网络FPGA实现方法1
08-03
在基于HLS（硬件描述语言）的FPGA（现场可编程门阵列）实现深度卷积神经网络（CNN）时，设计思路需兼顾硬件效率和网络的紧凑性。以下是几个关键知识点： 1.1 考虑硬件友好和网络小巧的设计思路 1.1.1 填充方式的...
卷积编码及Viterbi 解码的FPGA 实现及应用
01-19
介绍了卷积编码及Viterbi串行解码的原理及其FPGA的实现。在保证系统性能的前提下讨论了分帧式编解码在实际系统中的应用。 　0 引言 　在现代通信系统中，信道编码技术得到了广泛的应用。卷积码结构简单，硬件实现...
vivado 启动过程中报错
weixin_43509834的博客
05-16 3288
以前装过一次 vivado ，安装教程如下：vivado安装教程1-视频，安装教程2-CSDN，但是完全没有出现过任何错误。但是这次安装莫名其妙的报错，报错信息为： /opt/Xilinx/Vivado/2015.2/bin/loader: line 157: 4133 Segmentation fault (core dumped) “$ RDI_PROG” “$@” 搜索以后，解...
Frambuffer_rd在Windows下的vivado 2020.1编译报错解决方法
CookieNGU的博客
05-19 2937
vivado2020.1 HLS生成的IP在Windows下报错Frambuffer_rd在Windows下的vivado 2020.1编译报错解决方法我的解决方法：编译过程如下之后重新对工程generate ，这样就可以了。也可以单独对IP进行编译（没试过）另外一种笨方法mixer ip编译不过，而且没报错是内存不足 Frambuffer_rd在Windows下的vivado 2020.1编译报错解决方法 解决方法参考Xilinx官方论坛的解决方法，帖子链接如下： https://forums.xilin
[BUG历险记] ERROR: [SIM 211-100] CSim failed with errors
最新发布
qq_34022877的博客
05-31 560
在开发HLS过程中，我碰到一个奇怪的现象，同样的工程，在我重装完系统后，不能进行C仿真了，但是综合实现都是可以正常运作的。真实，惜字如金呀。
在vivado HLS2018.2中使用xfOpenCV
wd1603926823的专栏
09-14 1万+
本来写了一篇关于在HLS中使用xfOpenCV的记录，一步一步，但不小心删掉了。只能重新大致回忆下，肯定没有之前的全面。 我是在github上下载xfOpenCV https://github.com/Xilinx/xfopencv/tree/master/HLS_Use_Model 并仔细查看这个pdf文档。我跑了他这个例子，在HLS上跑起来了： 因为我是将整个xfOpenCV文件夹放在工程...
Vivado HLS | Export RTL报错 “ERROR: [IMPL 213-28] Failed to generate IP.
weixin_46279376的博客
08-17 759
之前查看其他人的解决办法是在Xlinx官网下载补丁，但是对于安装方法没有详细说明，自己解决时也遇到了问题，所以上传上来帮助大家
ERROR: [SYNCHK 200-61] unsupported memory access on variable
weixin_48437318的博客
12-26 439
这是在hls2018.3时其指针没有表示接口，需要我们指定，然后就可以综合出来了。在高版本的hls里就比较智能了，不需要指定。附一张变量类型接口的图。
基于HLS的高效深度学习卷积神经网络FPGA实现方法项目全部数据.zip
11-30
基于HLS的高效深度学习卷积神经网络FPGA实现方法项目全部数据.zip本文通过对现有相关研究的分析、总结和改进，给出了一系列在软件层面上如何构建和训练小巧高效且利于硬件加速的网络方法，在FPGA实现时如何减少资源...
基于Python HLS的高效深度卷积神经网络FPGA实现方法【100011343】
03-22
通过对现有相关研究的分析、总结和改进，给出了一系列在软件层面上如何构建和训练小巧高效且利于硬件加速的网络方法，在FPGA实现时如何减少资源、降低功耗及提高速度的方法，以及在HLS中如何增加设计灵活性、可移植...
基于FPGA的线性卷积的实时实现.docx
07-11
总结来说，基于 FPGA 的线性卷积实时实现利用了 FFT 和 IFFT 的高效性，通过转换为循环卷积降低了计算复杂度，并借助 FPGA 的并行计算能力，实现了对大规模离散序列卷积的快速处理，满足了实时信号处理的需求。...
ERROR: [SYNCHK 200-61] conv_core/conv_core.cpp:55: unsupported memory access on variable ‘bias‘
weixin_48437318的博客
11-02 69
以mnist手写数字体为例，看的是IC菌大佬的视频，用hls工具进行综合报错原因是没有添加约束文件。
利用Vitis开发基于ZCU106的神经网络加速器（一）——Vitis概述及XRT编译
GaleZhang的博客
11-19 4171
前言 毕设要用到Xilinx家的ZCU106这块板子，了解到最近Xilinx统一了Vivado，XilinxSDK，并集成了常用开源IP核，推出了Vitis统一软件平台，使我们不再需要关注底层的Verilog实现，因此尝试使用Vitis开发一个神经网络加速器，作为毕设的基础。 Vitis架构 XRT 由上图可以看到，Xilinx为我们提供了各种各样的加速库，而XRT(Xilinx Runtime Library)则是我们的代码与Xilinx硬件交互的中间件和媒介。有了XRT，我们就可以用更上层的开发方
[XSIM 43-3225] Cannot find design unit xil_defaultlib.simple_ram_tb in library work located at xsim.
nature_forest的博客
05-05 9777
2015.03 vivado，编写了一个仿真vivado simple_ram_dual IP核，想检测下伪双口ram的实际功能。结果报了如下错误 [XSIM 43-3225] Cannot find design unit xil_defaultlib.simple_ram_tb in library work located at xsim.dir/work. 直接说结果，就是
Vivado HLS error: Cannot find ISE in the PATH variable or it's an unsupported version
weixin_30483697的博客
04-18 222
在Vivado HLS中export rtl时选择pcores for edk并选中Evaluate VHDL,时会出现该错误,提示ISE找不到. 解决方案: 在path中添加ISE 和Vivado的路径后,重启Vivado HLS. 我添加的路径如下,G:\Xilinx\14.7\ISE_DS\ISE\bin\nt64;D:\Xilinx\vivado\Vivado\2013.2\b...
使用CCS时遇到的的一些麻烦
热门推荐
kyq___的博客
11-19 2万+
学习CCS编程是遇到的问题
基于HLS的高效深度卷积神经网络fpga实现方法
04-28
基于HLS的高效深度卷积神经网络FPGA实现方法通常包括以下几个步骤： 1. 数据预处理：将输入数据进行预处理，包括数据归一化、图像增强等。 2. 网络模型设计：设计卷积神经网络模型，包括网络的结构、参数等。 3. HLS代码生成：使用高层次综合（HLS）工具生成FPGA可执行的代码。 4. 硬件平台搭建：搭建FPGA硬件平台，包括硬件开发板、外设等。 5. 数据传输与计算：将预处理后的数据传输到FPGA上进行计算，并将计算结果传输回主机端。 6. 性能优化：针对算法和硬件平台进行优化，提高性能和能耗效率。 需要注意的是，HLS的使用需要熟悉硬件描述语言（HDL）和计算机体系结构等相关知识，因此需要具备一定的技术背景和经验。

“相关推荐”对你有帮助么？

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论 1
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

祥瑞Coding

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付：¥1
获取中
扫码支付

您的余额不足，请更换扫码支付或充值

打赏作者

实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值