Programming with CUDA WS 08/09 Lecture 12 Tue, 02 Dec, 2008.

Programming with Programming with CUDACUDAWS 08/09WS 08/09

Lecture 12Lecture 12Tue, 02 Dec, 2008Tue, 02 Dec, 2008

PreviouslyPreviously

Optimization example: parallel reductionOptimization example: parallel reduction

TodayToday

Graded/ungraded course?Graded/ungraded course? Revisiting shared memory bank conflictsRevisiting shared memory bank conflicts Final projectsFinal projects

Graded/ungradedGraded/ungraded

All settled?All settled?

Shared MemoryShared Memory

Devices of compute capability 1.x Devices of compute capability 1.x have 16 bankshave 16 banks– 16K shared memory in 16 banks, 1K 16K shared memory in 16 banks, 1K

eacheach– Successive 32-bit words are stored in Successive 32-bit words are stored in

successive bankssuccessive banks

Final ProjectsFinal Projects

Time-lineTime-line– Thu, 20 Nov:Thu, 20 Nov:

Float write-ups on ideas of Jens & WaqarFloat write-ups on ideas of Jens & Waqar

– Tue, 25 Nov:Tue, 25 Nov: Suggest groups and topicsSuggest groups and topics

– Thu, 27 Nov:Thu, 27 Nov: Groups and topics assignedGroups and topics assigned

– Tue, 2 Dec (today):Tue, 2 Dec (today): Last chance to change groups/topicsLast chance to change groups/topics Groups and topics finalizedGroups and topics finalized

Final ProjectsFinal Projects

There will be no lectures in the There will be no lectures in the second halfsecond half– Meetings with groupsMeetings with groups– Schedule will be put onlineSchedule will be put online

Final ProjectsFinal Projects

General tipsGeneral tips– Optimize your codeOptimize your code– Document your codeDocument your code– Make your code platform Make your code platform

independentindependent– don't mix C and C++don't mix C and C++

Final ExamFinal Exam

When?When?

On to exercises!On to exercises!