cudpp compile error

Accelerated Computing CUDA CUDA Programming and Performance
1

I’m trying to use CUDPP. However, I met a error when compiling it on
G8800GTX+CUDA1.1:
nvcc -o obj/release/segmented_scan_app.cu_o -c
src/app/segmented_scan_app.cu --host-compilation=C --compiler-options
-fno-strict-aliasing -I./ -I./include/ -Isrc/ -Isrc/app/
-Isrc/kernel/ -Isrc/cta/ -I. -I/usr/local/cuda/include
-I./…/common/inc -DUNIX -O3
nvcc error : ‘ptxas’ died due to signal 11 (Invalid memory reference)
make: *** [obj/release/segmented_scan_app.cu_o] Error 255

How to fix it?

2

Let me guess - 64bit system?

Got the same problem - this usually happens, when some identifiers get too long. I found it earlier by some browsing.

In this case i replaced in every cudpp file the string “SegementedScanTraits” with “SST”. The resulting generated internal identifiers were shorter and everything worked.

3

It works! Thank you so much!
Could you tell me how you found out this way?