![SSE (Streaming SIMD Extensions)
亠亳 仂亰仄仂亢仆仂亳
1
仂亠仄 128-弍亳仆 亠亞亳仂于 亟仍 4 亳亠仍 仗仂 32 弍亳 ( 仗仍舒于舒ﾑ亠亶
仂从仂亶)
2
丐亠仗亠 2 64-弍亳仆 亳仍舒 于 亠亞亳亠
3
丕亢亠 13 亳仆从亳亶
4.1
47 亳仆从亳亶 (从仂亠仆亳亠 于亳亟亠仂)
4.2
54 亳仆从亳亳 (仂仗亠舒亳亳 仂 仂从舒仄亳)
float a[4] = { 300.0, 4.0, 4.0, 12.0 };
float b[4] = { 1.5, 2.5, 3.5, 4.5 };
__asm {
movups xmm0, a ; // 仗仂仄亠亳 亳亰 a 于 亠亞亳 xmm0
movups xmm1, b ; // 仗仂仄亠亳 亳亰 b 于 亠亞亳 xmm1
mulps xmm0, xmm1 ; // 仗亠亠仄仆仂亢亳 仗舒从亠 仗仍舒于舒ﾑ亳 仂亠从
movups a, xmm0 ; // 于亞亰亳 亠亰仍舒 亳亰 亠亞亳舒 xmm0 仗仂 舒亟亠舒仄 a
};
4](https://image.slidesharecdn.com/20140216parallelprogrammingkalishenkolecture01-140217024550-phpapp01/85/20140216-parallel-programming_kalishenko_lecture01-4-320.jpg)
More Related Content
What's hot (7)
Similar to 20140216 parallel programming_kalishenko_lecture01 (20)
More from Computer Science Club (20)
20140216 parallel programming_kalishenko_lecture01
- 2. 仂亳于舒亳 亠仆亠亳从舒 亳 仗仂亠仂仄亳从舒 仍亳仄舒仂仍仂亞亳 个亳亰亳从舒 于仂从亳 仆亠亞亳亶 仂仆仂仄亳, 弍舒仆从仂于从亳亠 舒仆亰舒亳亳... 2
- 3. 仆仂亞仂仗仂仂仆仂亠 仗仂亞舒仄仄亳仂于舒仆亳亠 仂仂从亳 弌 Java-仗仂仂从亳 Boost-仗仂仂从亳 OpenMP 仆仂亠 仍舒亠仆亠 亳仍亠仆亳 (MPI) 仗亳仄亳亰舒亳亳 于 从仂仄仗亳仍仂舒 丐舒仆亰舒从亳仂仆仆舒 仗舒仄 亳仆仂仆仆亠 于仂亟/于于仂亟 丐亠仂亳 仂仆亠仆 亠弍仍仂从亳ﾑ亳亠 舒仍亞仂亳仄 仍亞仂亳仄 弍亠亰 仂亢亳亟舒仆亳 乂舒弍仍仂仆 || 仗仂亞舒仄仄亳仂于舒仆亳 Intel TBB Java.util.concurrent Fork/join framework Synchronization http://code.google.com/p/hpcource/wiki/Lectures 3
- 4. SSE (Streaming SIMD Extensions) 亠亳 仂亰仄仂亢仆仂亳 1 仂亠仄 128-弍亳仆 亠亞亳仂于 亟仍 4 亳亠仍 仗仂 32 弍亳 ( 仗仍舒于舒ﾑ亠亶 仂从仂亶) 2 丐亠仗亠 2 64-弍亳仆 亳仍舒 于 亠亞亳亠 3 丕亢亠 13 亳仆从亳亶 4.1 47 亳仆从亳亶 (从仂亠仆亳亠 于亳亟亠仂) 4.2 54 亳仆从亳亳 (仂仗亠舒亳亳 仂 仂从舒仄亳) float a[4] = { 300.0, 4.0, 4.0, 12.0 }; float b[4] = { 1.5, 2.5, 3.5, 4.5 }; __asm { movups xmm0, a ; // 仗仂仄亠亳 亳亰 a 于 亠亞亳 xmm0 movups xmm1, b ; // 仗仂仄亠亳 亳亰 b 于 亠亞亳 xmm1 mulps xmm0, xmm1 ; // 仗亠亠仄仆仂亢亳 仗舒从亠 仗仍舒于舒ﾑ亳 仂亠从 movups a, xmm0 ; // 于亞亰亳 亠亰仍舒 亳亰 亠亞亳舒 xmm0 仗仂 舒亟亠舒仄 a }; 4
- 5. 舒从仂仆 仄亟舒仍舒 a 亟仂仍 仗仂仍亠亟仂于舒亠仍仆仂亞仂 从仂亟舒 p 亳仍仂 仗仂亠仂仂于 1 S= (1a ) a+ p 5
- 6. 仂亠 亳 仗仂仂从亳 弌从 仗仂仂从舒 main() 弌从 弌从 仗仂仂从舒 仗仂仂从舒 仆舒舒仍亠 于仗仂仍仆亠仆亳 仗仂亠 1 仗仂仂从 仂仂从亳 仄仂亞 仂亰亟舒于舒 仆仂于亠 于 仗亠亟亠仍舒 仂亟仆仂亞仂 仗仂亠舒 亠 仗仂仂从亳 亳仄亠ﾑ 仂弍亳亠 亠亞仄亠仆 从仂亟舒 亳 亟舒仆仆 舒亢亟亶 仗仂仂从 亳仄亠亠 于仂亶 亠从 于仗仂仍仆亠仆亳 弌亠亞仄亠仆 从仂亟舒 亟亳仆亳舒 仗仍舒仆亳仂于舒仆亳 弌 - 仗仂仂从 弌亠亞仄亠仆 亟舒仆仆 6
- 7. 仂仂从亳 弌 仗亠舒亳 Posix 弌仂亰亟舒仆亳亠 pthread_create() 亢亳亟舒仆亳亠 亰舒于亠亠仆亳 pthread_join() 舒于舒 仄ﾑ亠从舒 pthread_mutex_lock() 于仂弍仂亢亟亠仆亳亠 仄ﾑ亠从舒 pthread_mutex_unlock() int pthread_create(pthread_t *thread, const pthread_attr_t *attr, void *(*start_routine)(void*), void *arg); 7
- 8. Java 舒仍亠亟仂于舒仆亳亠 仂 Thread 亠舒仍亳亰舒亳 Runnable 舒仍仆仂亠 仗亳于仆仂: start(), join() 8
- 9. Boost 弍从舒 仆舒亟 仗仂仂从舒仄亳 弌 (Posix 亳仍亳 Win threads) boost::thread_group th_group; for(...i) { boost::thread* th = new boost::thread(boost::bind(&仍舒::<仆从亳>, this, i)); th_group.add_thread(th); } th_group.join_all(); 9
- 10. OpenMP 弌舒仆亟舒 亳仆亠亠亶舒 亟仍 仄仆仂亞仂仗仂仂仆仂亞仂 仗仂亞舒仄仄亳仂于舒仆亳 仆舒亟 仂弍亠亶 仗舒仄 舒弍仂 亠亟于 亟仍 磶从仂于 C/C++ 亳 Fortran: 亳亠从亳于 从仂仄仗亳仍仂舒 (#pragma omp ...) 亳弍仍亳仂亠仆亠 仗仂亟仗仂亞舒仄仄 (get_num_threads()) 亠亠仄亠仆仆亠 仂从亢亠仆亳 (OMP_NUM_THREADS) 10
- 11. 亳仄亠 OpenMP #include <stdio.h> #include <omp.h> int main() { int i; #pragma omp parallel { #pragma omp for for (i=0;i<1000;i++) printf(%d ,i); } return 0; } 11
- 12. Intel TBB 12
- 13. 亳仄亠 Intel TBB 13
- 14. Java.util.concurrent 仍 仗仂仂从仂于: FixedThreadPool CahcedThreadPool SingleThreadPool... 仂仄亳从亳 (AtomicBoolean, AtomicLong...) 仂仂从仂弍亠亰仂仗舒仆亠 从仂仆亠亶仆亠 弌于仂亳 仗亳仄亳亳于 (ReentrantLock...) 亟亠亠 (Future<>) 14