A parallel Universe : Hardware vs Software parallelism
0 likes592 views
The document discusses hardware and software approaches to parallelism. Hardware parallelism utilizes multicore chips with more than one processor, GPUs with more than 10 processors, and cluster computing across networked systems. Software parallelism employs job scheduling to split a single process across threads, hyper-threading to make a single physical processor appear as two, and simultaneous multi-threading combining hyper-threading with multiple threads running simultaneously. The document also mentions software tools like Pthreads, OpenMP, and OpenCL that enable software parallelism.
A parallel Universe : Hardware vs Software parallelism
- 1. Hardware vs Software Parallelism Shivek Khurana Cluster Innovation Centre University Of Delhi
- 4. Multicore Chips Mo e han 1 on Neumann abs s n e h p (2-8) ex : n e 7 a on n a
- 5. GPU Mo e han 1 on Neumann abs s n e h p (mo e han 10) ex : N d a GeFo e a on n a
- 6. Cluster Computing A ne wo o s s ems wo a s n e pu pose ex : a eboo se e n o e he o
- 7. Grid Computing Remo e us e ompu n
- 9. Job Scheduling/ Pipelining on a single core A s n e p o ess h eads an be b o en n o
- 10. Hyper-threading In e ’s p op e a e hno o p o esso appea as 2 . 1 ph s a
- 11. Simultaneous Multi-threading H pe h ead n h ead n omb ned w h s n e o e
- 12. Software Specific Tools P Open Open L OpenMP