Goal Flashcards
What is the goal of parallel processors?
Connecting multiple computers to get higher performance.
What is a parallel processing program?
Single program runs in multiple computers.
What is multicore microprocessors?
Chips with multiple processors.
2 type of hardware. Name them:
Serial and parallel
Two tyoes of software. Name them:
Sequential and concurrent
Sequential runs on 1 whereas concurrent on 2
1 serial hardware
2 parallel
SISD stands for:
Single instruction stream single data stream uniprocessor
MIMD stands for:
Multiple instruction streams multiple data streamsmultiprocessor
SPMD stands for:
Single program multiple data streams a single program runs accross multiple processors
SIMD stands for:
Single instruction stream multiple data streams. The same instruction is applied to many data streams.
Load/store in vector instructions:
lv/sv
Add vectors of double:
addv.d
Add scalar to each element of vector double:
addvs.d
Vector processors reduce:
Instruction fetch bandwidth
What are the advantages of vector architectures and compilers?
. Simplify data parallel programming
. Absence of loop
. Regular access patterns benefit from interleaved and burst memory
. Avoid control hazards
Using a vector instruction indicates that the computation of each result in the vector is independent of the computation of other results.
True
Vector architectures are better than MIMD multiprocessors
True
In vector hardware only checks for data hazards between 2 instructions once per vector operand not once for every element
True
The cost of the latency to main memory is seen only once for the entire vector
True
SIMD other benefits:
All processors execute the same instruction at the same time, simplifies synchronization
Difference between vector and multimedia extensions:
Vector have a variable vector width while multimedia have fixed width
What are the various multithreading techniques?
Fine grain multithreading
Coarse grain multithreading and simultaneous multithreading
Fine grain multithreading:
. Switch threads after each cycle
. Interleave instruction execution
. If one thread stalls others are executed
Coarse grain multithreading:
Only switch in long stall and simplifies hardware