I have already installed cudnn and cuda in the ubuntu, and I
make all -j4
under the caffe-master directory, it passed well. but when I
make test
it shows:
CXX src/caffe/test/test_im2col_layer.cpp
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
from ./include/caffe/common.hpp:19,
from ./include/caffe/blob.hpp:8,
from src/caffe/test/test_im2col_layer.cpp:5:
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::createPoolingDesc(cudnnPoolingStruct**, caffe::PoolingParameter_PoolMethod, cudnnPoolingMode_t*, int, int, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:127:41: error: too few arguments to function ‘cudnnStatus_t cudnnSetPooling2dDescriptor(cudnnPoolingDescriptor_t, cudnnPoolingMode_t, cudnnNanPropagation_t, int, int, int, int, int, int)’
pad_h, pad_w, stride_h, stride_w));
^
./include/caffe/util/cudnn.hpp:15:28: note: in definition of macro ‘CUDNN_CHECK’
cudnnStatus_t status = condition; \
^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
from ./include/caffe/util/device_alternate.hpp:40,
from ./include/caffe/common.hpp:19,
from ./include/caffe/blob.hpp:8,
from src/caffe/test/test_im2col_layer.cpp:5:
/usr/local/cuda/include/cudnn.h:799:27: note: declared here
cudnnStatus_t CUDNNWINAPI cudnnSetPooling2dDescriptor(
^
Makefile:572: recipe for target '.build_release/src/caffe/test/test_im2col_layer.o' failed
make: *** [.build_release/src/caffe/test/test_im2col_layer.o] Error 1
I installed the newest version of cudnn(cudnn v5 library for linux) and cuda(cuda 7.5). Anyone could tell me how to solve the problem? Thanks a lot!
CuDNN v5 is incompatible with caffe. You can try making a fork and merging this: https://github.com/BVLC/caffe/pull/3919/files , but it is not officially supported.
Please note that the latest Caffe (November 8 2016) states that it supports CUDA 7+.
I still needed to fix this issue though as I was using an older branch of caffe linked to a faster-rcnn github repo. I was upgrading it to support CUDNN 5 and CUDA 8. Please note that I am not a Caffe expert, however all the tests ran successfully after I'd gotten it to compile successfully.
For the methods whose signatures appear to have changed, I found there are also _v3 and _v4 equivalents (for me I took a look in /usr/local/cuda/include/cudnn.h). I changed the method that was failing
FROM:
CUDNN_CHECK(cudnnSetPooling2dDescriptor(*pool_desc, *mode, h, w,
TO:
CUDNN_CHECK(cudnnSetPooling2dDescriptor_v3(*pool_desc, *mode, h, w,
4. src/caffe/layers/cudnn_conv_layer.cu
I needed to do similar things in the following files:
src/caffe/layers/cudnn_sigmoid_layer.cu
src/caffe/layers/cudnn_relu_layer.cu
src/caffe/layers/cudnn_conv_layer.cu
src/caffe/layers/cudnn_tanh_layer.cu
Hope that helps for you too!
try make clean --> make all --> make test --> make runtest. if you have a permission error use sudo.
Related
Using openzeppelin's UUPS upgradeable proxy contracts results in the following slither error. Are they false positives or should I be concerned?
Enviornment:
"#openzeppelin/contracts-upgradeable": "^4.5.2",
$ slither .
'npx hardhat compile --force' running
hardhat solidity version 0.8.9
Error 1:
ERC1967UpgradeUpgradeable._functionDelegateCall(address,bytes) (node_modules/#openzeppelin/contracts-upgradeable/proxy/ERC1967/ERC1967UpgradeUpgradeable.sol#198-204) uses delegatecall to a input-controlled function id
- (success,returndata) = target.delegatecall(data) (node_modules/#openzeppelin/contracts-upgradeable/proxy/ERC1967/ERC1967UpgradeUpgradeable.sol#202)
Reference: https://github.com/crytic/slither/wiki/Detector-Documentation#controlled-delegatecall
Error 2:
ERC1967UpgradeUpgradeable._upgradeToAndCallUUPS(address,bytes,bool).slot (node_modules/#openzeppelin/contracts-upgradeable/proxy/ERC1967/ERC1967UpgradeUpgradeable.sol#98) is a local variable never initialized
Reference: https://github.com/crytic/slither/wiki/Detector-Documentation#uninitialized-local-variables
Error 3:
ERC1967UpgradeUpgradeable._upgradeToAndCallUUPS(address,bytes,bool) (node_modules/#openzeppelin/contracts-upgradeable/proxy/ERC1967/ERC1967UpgradeUpgradeable.sol#87-105) ignores return value by IERC1822ProxiableUpgradeable(newImplementation).proxiableUUID() (node_modules/#openzeppelin/contracts-upgradeable/proxy/ERC1967/ERC1967UpgradeUpgradeable.sol#98-102)
ERC721Upgradeable._checkOnERC721Received(address,address,uint256,bytes) (node_modules/#openzeppelin/contracts-upgradeable/token/ERC721/ERC721Upgradeable.sol#393-414) ignores return value by IERC721ReceiverUpgradeable(to).onERC721Received(_msgSender(),from,tokenId,_data) (node_modules/#openzeppelin/contracts-upgradeable/token/ERC721/ERC721Upgradeable.sol#400-410)
Reference: https://github.com/crytic/slither/wiki/Detector-Documentation#unused-return
No, slither often flags OpenZepplin contracts, because they often use low-level functions like delegatecall or use assembly to deal with the return data.
So it's fine just keep in mind that while slither is an amazing piece of software it can't detect everything correctly.
I know this has been posted all over the place but I can not find what I need. I have a Arduino project that I must have vector or another way of allocating a dynamic array safely. I am trying to get the "StandardCplusplus" library to compile but I am getting an error
'Starting combiner'
"C:\Eclipse\eclipse\arduinoPlugin\tools\arduino\avr-gcc\4.8.1-arduino5/bin/avr-gcc" -Os -Wl,--gc-sections,--relax -mmcu=atmega2560 -o "D:/Development/MegaTesting/Release/MegaTesting.elf" ./.ino.cpp.o ./libraries/StandardCplusplus/algorithm.cpp.o ./libraries/StandardCplusplus/associative_base.cpp.o ./libraries/StandardCplusplus/bitset.cpp.o ./libraries/StandardCplusplus/char_traits.cpp.o ./libraries/StandardCplusplus/complex.cpp.o ./libraries/StandardCplusplus/del_op.cpp.o ./libraries/StandardCplusplus/del_opnt.cpp.o ./libraries/StandardCplusplus/del_opv.cpp.o ./libraries/StandardCplusplus/del_opvnt.cpp.o ./libraries/StandardCplusplus/deque.cpp.o ./libraries/StandardCplusplus/eh_alloc.cpp.o ./libraries/StandardCplusplus/eh_globals.cpp.o ./libraries/StandardCplusplus/exception.cpp.o ./libraries/StandardCplusplus/fstream.cpp.o ./libraries/StandardCplusplus/func_exception.cpp.o ./libraries/StandardCplusplus/iomanip.cpp.o ./libraries/StandardCplusplus/ios.cpp.o ./libraries/StandardCplusplus/iostream.cpp.o ./libraries/StandardCplusplus/istream.cpp.o ./libraries/StandardCplusplus/iterator.cpp.o ./libraries/StandardCplusplus/limits.cpp.o ./libraries/StandardCplusplus/list.cpp.o ./libraries/StandardCplusplus/locale.cpp.o ./libraries/StandardCplusplus/map.cpp.o ./libraries/StandardCplusplus/new_handler.cpp.o ./libraries/StandardCplusplus/new_op.cpp.o ./libraries/StandardCplusplus/new_opnt.cpp.o ./libraries/StandardCplusplus/new_opv.cpp.o ./libraries/StandardCplusplus/new_opvnt.cpp.o ./libraries/StandardCplusplus/numeric.cpp.o ./libraries/StandardCplusplus/ostream.cpp.o ./libraries/StandardCplusplus/queue.cpp.o ./libraries/StandardCplusplus/set.cpp.o ./libraries/StandardCplusplus/sstream.cpp.o ./libraries/StandardCplusplus/stack.cpp.o ./libraries/StandardCplusplus/stdexcept.cpp.o ./libraries/StandardCplusplus/streambuf.cpp.o ./libraries/StandardCplusplus/string.cpp.o ./libraries/StandardCplusplus/support.cpp.o ./libraries/StandardCplusplus/typeinfo.cpp.o ./libraries/StandardCplusplus/utility.cpp.o ./libraries/StandardCplusplus/valarray.cpp.o ./libraries/StandardCplusplus/vector.cpp.o arduino.ar "-LD:/Development/MegaTesting/Release" -lm
arduino.ar(new.cpp.o): In function `operator new[](unsigned int)':
C:/Eclipse/eclipse/arduinoPlugin/packages/arduino/hardware/avr/1.6.10/cores/arduino/new.cpp:26: multiple definition of `operator new[](unsigned int)'
./libraries/StandardCplusplus/new_opv.cpp.o:C:/Eclipse/eclipse/arduinoPlugin/packages/arduino/hardware/avr/1.6.10/libraries/StandardCplusplus/new_opv.cpp:24: first defined here
c:/eclipse/eclipse/arduinoplugin/tools/arduino/avr-gcc/4.8.1-arduino5/bin/../lib/gcc/avr/4.8.1/../../../../avr/bin/ld.exe: Disabling relaxation: it will not work with multiple definitions
arduino.ar(new.cpp.o): In function `operator new(unsigned int)':
C:/Eclipse/eclipse/arduinoPlugin/packages/arduino/hardware/avr/1.6.10/cores/arduino/new.cpp:22: multiple definition of `operator delete[](void*)'
makefile:83: recipe for target 'MegaTesting.elf' failed
./libraries/StandardCplusplus/del_opv.cpp.o:C:/Eclipse/eclipse/arduinoPlugin/packages/arduino/hardware/avr/1.6.10/libraries/StandardCplusplus/del_opv.cpp:25: first defined here
collect2.exe: error: ld returned 1 exit status
make: *** [MegaTesting.elf] Error 1
06:23:57 Build Finished (took 21s.666ms)
I have searched here and online and there seems to be a bunch of fixes but none of them work. There was some mention of including a pnew.cpp in one of your files but this file is not in the library anywhere.
All my code contains that I am trying to compile is
#include "StandardCplusplus.h"
#include "Arduino.h"
void setup(void) {
Serial.begin(57600);
}
void loop(void) {
}
There are a couple of versions of this library out there and I have tried the two main forks that I can find and both of them generate this error.
Also I am using Eclipse with the Arduino AVR plugin to compile.
Edit
I thought that commenting out the lines did work until I tried to use a vector. Once I did the StandardCplusplus library got really mad saying that the function new was not defined.
I think that there needs to be something done to the StandardCplusplus library like in new/new_opv delete/delete_opv files? Or is there a better way?
Edit - 2
Alright so I went into the files:
StandardCplusplus/new_opv.h
StandardCplusplus/del_opv.h
And commented out the bodies for the functions and everything seems to be happy. I did a simple test and everything looks like it is working but now I am concerned that I might of removed something that something else in the library might need, can anyone tell me if my changes will hurt anything?
Edit - 3
Well the last fix didn't work so I tried going to another library all together and I am allot closer. I Downloaded this STL-AVR library and had to make a minor edit to vector.h now I am having problems with the pnew.cpp file. Looks like the template is calling a "new" function that is now defined in arduino but if I try to comment out this body I get that it is not defined. I am sure this is some sort of order of the includes or missing one can anyone help?
Those STL like libary versions you have were put together before operator new was added to the Arduino core. Now they are in the core you are getting a multiple definition error.
You can go into your core and comment out the bits causing the multiple definition error.
You can probably just get away with commenting out the function definitions in the .cpp:
C:/Eclipse/eclipse/arduinoPlugin/packages/arduino/hardware/avr/1.6.10/cores/arduino/new.cpp
If you still get the errors, also comment out the function declarations in new.h, but you shouldn't need to as multiple declarations are allowed, only definitions must be unique.
In cython when my code is compiled with
#cython.wraparound(True)
and I use the following cast function to convert (cast) a float to an integer
cdef DTYPE_t_I float_int(np.float_t val):
return <DTYPE_t_I>val
it runs ok
BUT
when I turn off
#cython.wraparound(False)
the code compiles normally and when it runs it gives the following error
CORE GENERATED
It happens compiling in linux with gcc and windows with MGS
What is wrong? Should it be like this?
Because I am trying to gain speed, I would like to know to switch off these flag.
I'm using SWIG to generate wrapper from C++ for Robot Framework as test library. RF issues a warning because it contains no keywords.
The system under test is a Win32 DLL, LibLogin2, created by VS wizard. It exports the function by default:
LIBLOGIN2_API int fnLibLogin2(void);
// This is an example of an exported function.
LIBLOGIN2_API int fnLibLogin2(void)
{
return 42;
}
I add interface file to the project:
/* LibLogin2.i */
%module LibLogin2
%{
extern int fnLibLogin2(void);
%}
extern int fnLibLogin2(void);
LibLogin2.py, LibLogin2_warp.cxx, _LibLogin2.pyd are built successfully with Release|x64.
I have a RF test case as follows:
*** Settings ***
Library LibLogin2.py
*** Test Case ***
Trivial
${value} = fnLibLogin2
Should Be Equal ${value} ${42}
I launch Robot Framework and get THE error:
pybot LoginTests.tsv
[ WARN ] Imported library 'LibLogin2.py' contains no keywords
I can work around this by commenting out the last line of LibLogin.py:
def fnLibLogin2():
return _LibLogin2.fnLibLogin2()
#fnLibLogin2 = _LibLogin2.fnLibLogin2
It's annoying when you have to comment out every key word every time.
Please advice!
Here is my configuration:
Windows 8 64-bit
Visual Studio 2012
Python 2.6.6
Robot Framework 2.7.5
swigwin-2.0.8
I'm trying to use the ISHFT() function to bitshift some 32-bit integers in parallel, using CUDA FORTRAN.
The problem is that I get different answers to ISHFT(-4,-1) and ISHFT(var,-1) even though var = -4. This is the test code I've written:
module testshift
integer :: test
integer, device :: d_test
contains
attributes(global) subroutine testshft ()
integer :: var
var = -4
d_test = ISHFT(var,-1)
end subroutine testshft
end module testshift
program foo
use testshift
integer :: i
call testshft<<<1,1>>>() ! carry out ishft on gpu
test = d_test ! copy device result to host
i = ISHFT(-4,-1) ! carry out ishft on cpu
print *, i, test ! print the results
end program foo
I then compile and execute:
pgf90 testishft.f90 -Mcuda
./a.out
2147483646 -2
Both should be 2147483646 if working correctly. I get the right answer if I replace var with 4.
How do I fix this problem?
Thanks for the help
When I remove the GPU-specific code from the above program I get 2147483646 2147483646 from the g95 compiler, as you expect. Have you tried running a "scalar" version of the program using the pgf90 compiler? If the scalar version works but the GPU version does not, that helps to isolate the problem. If the problem is pgf90/CUDA specific, perhaps the best place to ask your question is
PGI User Forum Forum Index -> Programming and Compiling
http://www.pgroup.com/userforum/viewforum.php?f=4 .
I've found a workaround, which is posted in this forum:
http://www.pgroup.com/userforum/viewtopic.php?t=2455&postdays=0&postorder=asc&start=15
Instead of using ISHFT I use IBITS, which is described here: http://gcc.gnu.org/onlinedocs/gfortran/IBITS.html
Also the problem has since been fixed in version 11.3 of the PGI compiler
http://www.pgroup.com/support/release_tprs_2011.htm