A Macrotask-level Unlimited Speculative Execution on Multiprocessors
(6/19)
The following description is based on the paper presented at
1995 ACM International Conference on Supercomputing head
in Spain on July 3rd to 7th,1995.
If you want to have the copy of the paper, please
let me know.
The next transparency shows problems of conventional speculative
execution and our solutions:
Most of the researches are related to the speculation in a VLIW or
a Superscalar processor. These schemes, however, can not attain high
performance because the speedup is limited by the number of ALU's in
a processor.
On the other hand, Banerjee proposed the speculation on Multiprocessors
for the first time. The scheme, however, supports only a special loop
containing one branch in it.
In 1993, Theobald proposed the speculative execution model on
multiprocessors. However, they restrict the speculation path.