run out of shared memory? using too much shared data
when i compile my project with cuda3.1 on GTX465, i get:
"uses too much shared data (0x4818 bytes + 0x10 bytes system, 0x4000 max)"
but the shared memory available per block on GTX465 is 48KB, i just used 16KB of that. Is there any switch i should open in order to use more shared memory?
when i compile my project with cuda3.1 on GTX465, i get:

"uses too much shared data (0x4818 bytes + 0x10 bytes system, 0x4000 max)"

but the shared memory available per block on GTX465 is 48KB, i just used 16KB of that. Is there any switch i should open in order to use more shared memory?

#1
Posted 05/18/2011 03:14 AM   
Compile with [font="Courier New"]-arch sm_20[/font].
Compile with -arch sm_20.

Always check return codes of CUDA calls for errors. Do not use __syncthreads() in conditional code unless the condition is guaranteed to evaluate identically for all threads of each block. Run your program under cuda-memcheck to detect stray memory accesses. If your kernel dies for larger problem sizes, it might exceed the runtime limit and trigger the watchdog timer.

#2
Posted 05/18/2011 06:50 PM   
[quote name='tera' date='19 May 2011 - 02:50 AM' timestamp='1305744656' post='1239122']
Compile with [font="Courier New"]-arch sm_20[/font].
[/quote]

thanks for the useful information :-)
[quote name='tera' date='19 May 2011 - 02:50 AM' timestamp='1305744656' post='1239122']

Compile with -arch sm_20.





thanks for the useful information :-)

#3
Posted 08/30/2011 05:13 AM   
Scroll To Top