IPC분류정보
국가/구분 |
United States(US) Patent
등록
|
국제특허분류(IPC7판) |
|
출원번호 |
US-0233895
(2008-09-19)
|
등록번호 |
US-8266610
(2012-09-11)
|
발명자
/ 주소 |
|
출원인 / 주소 |
- International Business Machines Corporation
|
대리인 / 주소 |
|
인용정보 |
피인용 횟수 :
1 인용 특허 :
38 |
초록
▼
A method, apparatus, and computer instructions for scheduling instructions for execution. Identify a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency. Determine whether the series of instructions is a uniform series of instructions. Schedule execution
A method, apparatus, and computer instructions for scheduling instructions for execution. Identify a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency. Determine whether the series of instructions is a uniform series of instructions. Schedule execution of the uniform series of instructions within the loop to optimize execution of the loop in response to the identified series of instructions being the uniform series of instructions.
대표청구항
▼
1. A method in a data processing system for scheduling instructions for execution, the method comprising: identifying a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency;determining whether the series of instructions is a uniform series of instructions
1. A method in a data processing system for scheduling instructions for execution, the method comprising: identifying a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency;determining whether the series of instructions is a uniform series of instructions;scheduling a spacing of the uniform series of instructions within the loop, wherein scheduling a spacing of the uniform series of instructions within the loop includes adding a number of slack cycles between a first instruction and a second instruction following the first instruction in the uniform series, wherein scheduling a spacing of the uniform series of instructions within the loop includes using the following: spacing between instruction a(i) and instruction b(i+1)=(((i)*SCC slack/M)−((i+1)*SCC slack)/M), wherein i is an index number, a(i) is a first instruction, b(i+1) is a second instruction following the first instruction, strongly connected component slack is a target initiation interval minus a longest cyclic delay for any series of instructions having the cyclic data dependency in the loop, and M is a number of repetitions of a sequence of instructions in the series of instructions and wherein the spacing is in cycles; andresponsive to the identified series of instructions being the uniform series of instructions, scheduling execution of the uniform series of instructions within the loop to optimize execution of the loop. 2. The method of claim 1, wherein the series of instructions is a strongly connected component. 3. The method of claim 1, wherein the execution of the loop is optimized by reducing a number of live general purpose registers present at a beginning of a cycle. 4. The method of claim 1, wherein the cyclic data dependency for the series of instructions is based on a single register used by the series of instructions. 5. The method of claim 1, wherein the scheduling step includes: spacing instructions from the uniform series of instructions evenly within the loop. 6. The method of claim 1, wherein the identifying step, the determining step, and the scheduling step is performed by a compiler. 7. A computer program product in a computer readable, recordable-type medium in a data processing system for scheduling instructions for execution, the computer program product comprising: first instructions for identifying a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency;second instructions for determining whether the series of instructions is a uniform series of instructions;third instructions for scheduling a spacing of the uniform series of instructions within the loop, wherein scheduling a spacing of the uniform series of instructions within the loop includes adding a number of slack cycles between a first instruction and a second instruction following the first instruction in the uniform series, wherein the third instructions include: sub-instructions for scheduling a spacing of the uniform series of instructions within the loop using the following:spacing=(((b)*SCC slack/N)−((a*SCC slack)/N), wherein a is a first instruction, b is a second instruction following the first instruction, strongly connected component (SCC) slack is a target initiation interval minus a longest cyclic delay for any series of instructions having the cyclic data dependency in the loop, and N is a number of instructions and wherein the spacing is in cycles; andfourth instructions, responsive to the identified series of instructions being the uniform series of instructions, for scheduling execution of the uniform series of instructions within the loop to optimize execution of the loop. 8. The computer program product of claim 7, wherein the series of instructions is a strongly connected component. 9. The computer program product of claim 7, wherein the execution of the loop is optimized by reducing a number of live general purpose registers present at a beginning of a cycle. 10. The computer program product of claim 7, wherein the cyclic data dependency for the series of instructions is based on a single register used by the series of instructions. 11. The computer program product of claim 7, wherein the third instructions includes: second sub-instructions for spacing instructions from the uniform series of instructions evenly within the loop. 12. The computer program product of claim 7, wherein the first instructions, the second instructions, and the third instructions arc performed by a compiler. 13. The computer program product of claim 7, wherein the first instructions, second instructions, third instructions, and fourth instructions are implemented for use with swing modulo scheduling. 14. A computer program product in a computer readable, recordable-type medium in a data processing system for scheduling instructions for execution, the computer program product comprising: first instructions for identifying a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency;second instructions for determining whether the series of instructions is a uniform series of instructions;third instructions for scheduling a spacing of the uniform series of instructions within the loop, wherein scheduling a spacing of the uniform series of instructions within the loop includes adding a number of slack cycles between a first instruction and a second instruction following the first instruction in the uniform series, wherein the third instructions include: sub-instructions for scheduling a spacing of the uniform series of instructions within the loop using the following:spacing=(((b)*SCC slack/M)−((a*SCC slack)/M), wherein a is a first instruction, b is a second instruction following the first instruction, strongly connected component (SCC) slack is a target initiation interval minus a longest cyclic delay for any series of instructions having the cyclic data dependency in the loop, and M is a number of repetitions of a sequence of instructions in the series of instructions and wherein the spacing is in cycles; andfourth instructions, responsive to the identified series of instructions being the uniform series of instructions, for scheduling execution of the uniform series of instructions within the loop to optimize execution of the loop.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.