I’m trying to migrate CUDA code to Dpc++ using Intel DPCT tool. Here are my details:
OS: Ubuntu 20.04
Application: BlackScholes (Finance app. Nvidia CUDA toolkit samples)
During migration, lot of warnings have been generated by the DPCT tool:
/home/administrator/NVIDIA_CUDA-11.1_Samples/4_Finance/BlackScholes/BlackScholes.cu:116:21: warning: DPCT1003:0: Migrated API does not return error code. (, 0) is inserted. You may need to rewrite this code.
checkCudaErrors(cudaMalloc((void **)&d_CallResult, OPT_SZ));
^
/home/administrator/NVIDIA_CUDA-11.1_Samples/4_Finance/BlackScholes/BlackScholes.cu:117:21: warning: DPCT1003:1: Migrated API does not return error code. (, 0) is inserted. You may need to rewrite this code.
checkCudaErrors(cudaMalloc((void **)&d_PutResult, OPT_SZ));
^
/home/administrator/NVIDIA_CUDA-11.1_Samples/4_Finance/BlackScholes/BlackScholes.cu:118:21: warning: DPCT1003:2: Migrated API does not return error code. (, 0) is inserted. You may need to rewrite this code.
checkCudaErrors(cudaMalloc((void **)&d_StockPrice, OPT_SZ));
^
/home/administrator/NVIDIA_CUDA-11.1_Samples/4_Finance/BlackScholes/BlackScholes.cu:119:21: warning: DPCT1003:3: Migrated API does not return error code. (, 0) is inserted. You may need to rewrite this code.
checkCudaErrors(cudaMalloc((void **)&d_OptionStrike, OPT_SZ));
Can someone tell me how to resolve these warnings?
Steps to reproduce:
intercept-build make
dpct -p compile_commands.json
You can eliminate the warnings related to cudaCheckErrors by just rewriting the code.
For Example:
"checkCudaErrors(cudaMalloc((void **)&d_CallResult, OPT_SZ));" can be rewritten as "cudaMalloc((void **)&d_CallResult, OPT_SZ);" i.e explicitly remove "checkCudaErrors" in cuda sorce code. Same follows with the other cudaCheckErrors warnings.
Later try migrating the code. I hope this might help you
Thanks, Shivani
Related
I'm getting this error "invalid version 0 on git_proxy_options" from libgit2. This did not happen until very recently (I have recompiled my code so it may be a version issue). I'm not using a proxy. Does anyone know anything about this? What causes this message? Is there anything I can do to stop the failure? Either a fix or or a work-around will do.
Specifically, I am using the Rust crate git2 version 0.13.23. I am doing a fetch that triggers the error. This is on Fedora 34 on x86_64. I have searched on this but there does not seem to be any matches that help.
git2 = {version="0.13.22", features = ["vendored-libgit2"]}
in Cargo.toml fixed the issue.
Interesting...
I had upgraded libgit2 to 1.3.0 from 1.1.1. That broke my existing cargo install, so I unpacked the previous libgit2.so.1.1 with a symlink to libgit2.so.1. That's when this error condition showed up. After thinking a bit I decided to symlink libgit2.so.1.3 to libgit2.so.1.1 and it built my new cargo version which linked to the new libgit2.so.1.3. Then I deleted my libgit2.so.1.1 stuff. Kinda weird, and usually you expect things not to work when it has been linked to a previous .so version.
I download openflow. It successfully been built. However, only scenario_Small is correctly working, when try to run other scenarios error runtime appears such as that:
Cannot add statistic 'numOutOfOrderArrivals' to module MultiController.Vancouver.client[0].pingApp[0] (NED type: openflow.apps.PingAppRandom): Error in source=numOutOfOrderArrivals: Signal 'numOutOfOrderArrivals' is not declared on type 'openflow.apps.PingAppRandom' (you can turn off this check by adding checkSignals=false to the #statistic property in the NED file) -- in module (PingAppRandom) MultiController.Vancouver.client[0].pingApp[0] (id=161), during network setup
when trying to run "szenario_Domains_multiController"
I tried to fix the error following the hint mentioned in error, but another errors appeared. How to fix those endless errors.
It seems that the codebase of this openflow project is already old and does not correspond to the new builds of OmNET++ and the INET framework.
I was success to build and run test scenarios from this project on Win 10 and OmNET++ 5.6.2 and INET 3.6.6.
I am using Eclipse Kepler with jdk1.8.0.
I do not face any compilation problems in import statements related to java8 or collect construct. However I get the compilation issue in:- Dish::getName...it says.. Syntax errors on tokens, delete the tokens.
Just trying to run the following code:
List<String> dishNames = menu.stream()
.map(Dish::getName)
.collect(toList());
System.out.println(dishNames);
Dish is a java object with String property name.
I have installed JDK8, added the latest JRE in eclipse Java Build Path.
Java compiler compliance.
Any hints what something silly I am missing?
I think you missed installing support of java8 in eclipse kepler.
You can follow this link for detailed instructions.
https://wiki.eclipse.org/JDT/Eclipse_Java_8_Support_For_Kepler
This is a very annoying problem while generating release build files. Here is the steps I did while building:
Select Project > Export Release Build
Select Signed Native Installer
Import certificate
Click Next.
I always got the following error:
https://forums.adobe.com/servlet/JiveServlet/showImage/2-6524012-651585/build_error.png
Platform conversion failed. Process exited with error 5. Output was
"[ErrorEvent type="error" bubbles=false cancelable=false eventPhase=2
text="Unhandled exception Error: EndUpdateResource: Windows Error #5"
errorID=0]
I have the following setup:
OS: Windows 8.1
System type: 64-bit
Flash Builder Version: 4.6 Premium
Installed FB: FlashBuilder_4_6_LS10.exe
Please help. This is giving me a lot of headaches for a couple of days now.
Additionally, here's the stack trace:
java.io.IOException: Platform conversion failed. Process exited with
error 5. Output was "[ErrorEvent type="error" bubbles=false
cancelable=false eventPhase=2 text="Unhandled exception Error:
EndUpdateResource: Windows Error #5" errorID=0] " at
com.adobe.air.nai.NativePackager.conversionFail(NativePackager.java:222)
at
com.adobe.air.nai.NativePackager.invokePlatformConversion(NativePackager.java:203)
at
com.adobe.air.nai.NativePackager.createPackage(NativePackager.java:92)
at
com.adobe.flexbuilder.multisdk.apollo.export.AIRNativeInstallerPackager.create(AIRNativeI
nstallerPackager.java:129) at
com.adobe.flexbuilder.exportimport.releaseversion.ExportReleaseVersionManager.doExport(Ex
portReleaseVersionManager.java:586) at
com.adobe.flexbuilder.exportimport.releaseversion.ui.ExportReleaseVersionWizard$1.run(Exp
ortReleaseVersionWizard.java:208)
I also suspect this is because of the following settings:
Eclipse.ini
https://forums.adobe.com/servlet/JiveServlet/showImage/2-6525217-651726/eclipse.png
...
-startup plugins/org.eclipse.equinox.launcher_1.2.0.v20110502.jar
--launcher_library plugins/org.eclipse.equinox.launcher.win32.win32.x86_1.1.100.v20110502.jar
...
Notice that the plugin used is not intended for x64 processors.
#ActionScript3030 is correct, for anyone finding this later on, if you re-link your Library Paths (even if they look correct) this should rectify your issue.
The issue is caused by the links not being able to resolve to their GUID counterparts in your AppData/Local folder. This happens quite commonly when you create a new user profile and try to copy over your old user profiles information.
I already figured this out. I created a virtual machine with windows 7 32bit operating system and I'm happy now :)
If you initially created your project on a different system then try to compile it in another system, the class source and swc paths may not match. These paths could be there but not being implemented in code, so you will not get any run time errors while debugging, but it will give you the error 5 during compilation.
Solution:
Remove the idle paths not directly in use in your code. You can identify them by simply checking the path.
Found the same issue when trying to use Microsoft Essential Security on my Windows 7 x32. When i uinstall it - problem gone. So try to check your firewall settings if stucking the same problem.
Just to add an alternative solution (I'm on an x64 Windows and want to keep build exported for x64), in my case the error popped up when building the project with the AIR SDK defined in the app.xml (an older one, 3.9) and went away when I explicitly used a newer SDK (19).
(Disclaimer: I used the command line adt tool from the SDK to build, not Flash Builder).
A little late, but might possibly help someone.
I had the same error recently with FlashDevelop. Figured out that the my path was too long. When I renamed some long folder and file names, the error disappeared.
Try to disable UAC on Windows 8. Change it to "Never notify". The problem
I am not getting TBB to work. I am following the steps in the "Getting started" document.
I am doing the following steps:
downloading the linux files + the sources files.
extracting them in 1 directory
calling make
going to tbb.../bin calling source tbbvars.sh intel64
going to examples/Getting_started/sub_string_finder
calling make
I then get the error:
sub_string_finder.cpp:32:30: fatal error: tbb/parallel_for.h: No such file or directory
I really googled a lot but can't find any related stuff.
I did also try to add some -I statement but it didnt help
I assume it is kind of a including/linking problem but I dont know how to fix.
This is all done on fedora 16 64bit. (kernel 3.1.4) // TBB version 4.0
The solution was to install tbb-devel package.