Comments (3)
ouankou
commented on June 20, 2024
GCC doesn't support 2D array, we may have to rewrite the code to use 1D array.
backprop.c: In function ‘bpnn_adjust_weights’:
backprop.c:296:14: error: array section is not contiguous in ‘map’ clause
296 | present(w[0:nly][0:ndelta],oldw[0:nly][0:ndelta])In the corresponding OpenMP GPU version, the compilation by LLVM is fine, but the execution leads to illegal memory access on the device. It's probably the same reason.
from neorodinia-old.
ouankou
commented on June 20, 2024
For the corresponding OpenMP GPU version, with commit 45a787f, the program can be compiled by clang now, but with the following warnings.
imagenet.c:18:32: warning: implicit conversion from 'int' to 'float' changes value from 2147483647 to 2147483648 [-Wimplicit-const-int-float-conversion]
units[k] = (float)rand() / RAND_MAX;
~ ^~~~~~~~
/usr/include/stdlib.h:86:18: note: expanded from macro 'RAND_MAX'
#define RAND_MAX 2147483647
^~~~~~~~~~
1 warning generated.
imagenet.c:18:32: warning: implicit conversion from 'int' to 'float' changes value from 2147483647 to 2147483648 [-Wimplicit-const-int-float-conversion]
units[k] = (float)rand() / RAND_MAX;
~ ^~~~~~~~
/usr/include/stdlib.h:86:18: note: expanded from macro 'RAND_MAX'
#define RAND_MAX 2147483647
^~~~~~~~~~
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:43:66: note: used here
__DEVICE__ void __attribute__((overloadable)) __brkpt(int __a) { __brkpt(); }
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1189:10: note: used here
return __bool2mask(__vseteq2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1189:22: note: used here
return __bool2mask(__vseteq2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1199:22: note: used here
return __bool2mask(__vseteq4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1209:22: note: used here
return __bool2mask(__vsetges2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1219:22: note: used here
return __bool2mask(__vsetges4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1229:22: note: used here
return __bool2mask(__vsetgeu2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1239:22: note: used here
return __bool2mask(__vsetgeu4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1249:22: note: used here
return __bool2mask(__vsetgts2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1259:22: note: used here
return __bool2mask(__vsetgts4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1269:22: note: used here
return __bool2mask(__vsetgtu2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1279:22: note: used here
return __bool2mask(__vsetgtu4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1289:22: note: used here
return __bool2mask(__vsetles2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1299:22: note: used here
return __bool2mask(__vsetles4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1309:22: note: used here
return __bool2mask(__vsetleu2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1319:22: note: used here
return __bool2mask(__vsetleu4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1329:22: note: used here
return __bool2mask(__vsetlts2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1339:22: note: used here
return __bool2mask(__vsetlts4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1349:22: note: used here
return __bool2mask(__vsetltu2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1359:22: note: used here
return __bool2mask(__vsetltu4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1369:22: note: used here
return __bool2mask(__vsetne2(__a, __b), 16);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1379:22: note: used here
return __bool2mask(__vsetne4(__a, __b), 8);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1399:21: note: used here
unsigned mask = __vcmpgts2(__a, __b);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1493:60: note: used here
__DEVICE__ unsigned int __vneg2(unsigned int __a) { return __vsub2(0, __a); }
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1502:60: note: used here
__DEVICE__ unsigned int __vneg4(unsigned int __a) { return __vsub4(0, __a); }
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1511:10: note: used here
return __vsubss2(0, __a);
^
/opt/llvm/llvm-14.x-install/lib/clang/14.0.0/include/__clang_cuda_device_functions.h:1521:10: note: used here
return __vsubss4(0, __a);
^
1 warning generated.The warning related to those headers seems to be a known bug. It was mentioned here.
https://www.mail-archive.com/[email protected]/msg53641.html
LLVM 15 shows the same warning. If we remove the header stdlib.h in imagenet.c, the header warnings will be gone.
from neorodinia-old.
ouankou
commented on June 20, 2024
The OpenMP GPU offloading version is created from scratch based on the official OpenMP version. Then the OpenACC version is created based on the OpenMP GPU offloading version. The warnings above don't affect compilation or execution, they will be revisited later.
The original OpenACC version doesn't work at all, so it's abandoned.
from neorodinia-old.
Related Issues (7)
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from neorodinia-old.