Method and system for fast 90 degree rotation of arrays

• US 7,511,722 B1
• Filed: 08/27/2004
• Issued: 03/31/2009
• Est. Priority Date: 08/27/2004
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Apple Inc.
• Inventors: Ollmann, Ian R.
• Primary Examiner(s): Tung; Kee M
• Assistant Examiner(s): Amini; Javid A
• Application Number:
  US10/928,450
• Time in Patent Office: 1,677 Days
• Field of Search: 345/649, 345/568, 382/296
• US Class Current: 345/649
• CPC Class Codes: G06T 3/606 by memory addressing or mapping
• CPC Class Codes: G09G 2340/0492 Change of orientation of the displayed image, e.g. upside-down, mirrored
• CPC Class Codes: G09G 2360/123 using interleaving
• CPC Class Codes: G09G 5/393 Arrangements for updating the contents of the bit-mapped memory
• CPC Class Codes: G09G 5/395 Arrangements specially adapted for transferring the contents of the bit-mapped memory to the screen G09G5/399 takes precedence

Abstract

A method and system to rotate a 2^N by 2^N array are described. Consistent with one embodiment of the present invention, the 2^N elements of the 2^N rows of a 2^N by 2^N array are loaded from memory into the vector registers of a processor'"'"'s single instruction multiple data (SIMD) unit. Next, the elements of the rows in the top half of the array are interleaved with corresponding elements from a corresponding row in the bottom half of the array. The loading and interleaving operations are repeated N times before the results, stored in the vector registers, are written back to memory.

Claims

1. A computer-implemented method to rotate a 2^N by 2^N array in the counter-clockwise direction, the computer-implemented method comprising:

loading 2^N elements of each row of the array into vector registers of a SIMD unit;

interleaving the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array;

storing the results of the interleaving operation in vector registers;

repeating said loading and interleaving operations a total of N times, wherein a result of said loading and said interleaving is stored for further processing; and

after N loading and N interleaving operations, writing the vector registers storing the results of the N loading and N interleaving operations to memory in order so that the elements representing the top row of the rotated array precede the elements representing each successive row of the array, the last 2^N elements written to memory representing the bottom row of the rotated array.

2. The method of claim 1, wherein said interleaving the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array is performed in parallel.

3. The method of claim 1, wherein the loading and interleaving operations are achieved via the execution of a single vector merge instruction that operates on data in multiple vector registers.

4. The method of claim 1, wherein the 2^N by 2^N array represents a subsection of a digital image and the loading and interleaving operations are performed iteratively on a plurality of subsections to rotate an entire digital image.

5. A computer-implemented method to rotate a 2^N by 2^N array in the clockwise direction, the computer-implemented method comprising:

loading 2^N elements of each row of the array into vector registers of a SIMD unit;

interleaving the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array follows an element from a corresponding row in the bottom half of the array;

storing the results of the interleaving operation in vector registers;

repeating said loading and interleaving operations a total of N times, wherein a result of said loading and said interleaving is stored for further processing; and

after N loading and N interleaving operations, writing the vector registers storing the results of the N loading and N interleaving operations to memory in order so that the elements representing the top row of the rotated array precede the elements representing each successive row of the array, the last 2^N elements written to memory representing the bottom row of the rotated array.

6. The method of claim 5, wherein said interleaving the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array is performed in parallel.

7. The method of claim 5, wherein the loading and interleaving operations are achieved via the execution of a single vector merge instruction that operates on data in multiple vector registers.

8. The method of claim 5, wherein the 2^N by 2^N array represents a subsection of a digital image and the loading and interleaving operations are performed iteratively on a plurality of subsections to rotate an entire digital image.

9. A computer-readable medium storing instructions, which, when executed, cause a computer to perform a method to rotate a 2^N by 2^N array in the counter-clockwise direction, the method comprising:

loading 2^N elements of each row of the array into vector registers of a SIMD unit;

interleaving the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array;

storing the results of the interleaving operation in vector registers;

repeating said loading and interleaving operations a total of N times, wherein a result of said loading and said interleaving is stored for further processing; and

after N loading and N interleaving operations, writing the vector registers storing the results of the N loading and N interleaving operations to memory in order so that the elements representing the top row of the rotated array precede the elements representing each successive row of the array, the last 2^N elements written to memory representing the bottom row of the rotated array.

10. A computer-readable medium as in claim 9, wherein said interleaving the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array is performed in parallel.

11. A computer-readable medium as in claim 9, wherein the loading and interleaving operations are achieved via the execution of a single vector merge instruction that operates on data in multiple vector registers.

12. A computer-readable medium as in claim 9, wherein the 2^N by 2^N array represents a subsection of a digital image and said loading and interleaving operations are performed iteratively on a plurality of subsections to rotate an entire digital image.

13. A computer-readable medium storing instructions, which, when executed, cause a computer to perform a method to rotate a 2^N by 2^N array in the clockwise direction, the method comprising:

loading 2^N elements of each row of the array into vector registers of a SIMD unit;

interleaving the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array follow an element from a corresponding row in the bottom half of the array;

storing the results of the interleaving operation in vector registers;

repeating said loading and interleaving operations a total of N times, wherein a result of said loading and said interleaving is stored for further processing; and

after N loading and N interleaving operations, writing the vector registers storing the results of the N loading and N interleaving operations to memory in order so that the elements representing the top row of the rotated array precede the elements representing each successive row of the array, the last 2^N elements written to memory representing the bottom row of the rotated array.

14. A computer-readable medium as in claim 13, wherein said interleaving the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array is performed in parallel.

15. A computer-readable medium as in claim 13, wherein the loading and interleaving operations are achieved via the execution of a single vector merge instruction that operates on data in multiple vector registers.

16. A computer-readable medium as in claim 13, wherein the 2^N by 2^N array represents a subsection of a digital image and said loading and interleaving operations are performed iteratively on a plurality of subsections to rotate an entire digital image.

17. An apparatus for rotating a 2^N by 2^N array in the counter-clockwise direction, the apparatus comprising:

a memory device to store the 2^N by 2^N array;

a processor connected to the memory device, the processor including a SIMD unit having vector registers capable of storing and processing 2^N elements, wherein the processor is to;

load the 2^N elements of each row into the vector registers of the SIMD unit;

interleave the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array;

store the results of the interleaving operation in vector registers;

repeat said loading and interleaving operations a total of N times, wherein a result of said loading and said interleaving is stored for further processing; and

after N loading and N interleaving operations, the processor to write the vector registers storing the results of the N loading and N interleaving operations to memory in order so that the elements representing the top row of the rotated array precede the elements representing each successive row of the array, the last 2^N elements written to memory representing the bottom row of the rotated array.

18. The apparatus of claim 17, wherein said processor to interleave the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array, performs the interleave operation in parallel.

19. The apparatus of claim 17, wherein the SIMD unit of the processor performs the load and interleave operations in response to the execution of a single vector merge instruction.

20. The apparatus of claim 17, wherein the 2^N by 2^N array represents a subsection of a digital image stored in the memory device, and said processor is to rotate the entire digital image by iteratively rotating subsections of the digital image and writing the rotated subsections back to the memory device.

21. An apparatus for rotating a 2^N by 2^N array in the clockwise direction, the apparatus comprising:

a memory device to store the 2^N by 2^N array;

a processor connected to the memory device, the processor including a SIMD unit having vector registers capable of storing and processing 2^N elements, wherein the processor is to;

load the 2^N elements of each row into the vector registers of the SIMD unit;

interleave the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array;

store the results of the interleaving operation in vector registers;

repeat said loading and interleaving operations a total of N times, wherein a result of said loading and said interleaving is stored for further processing; and

after N loading and N interleaving operations, the processor to write the vector registers storing the results of the N loading and N interleaving operations to memory in order so that the elements representing the top row of the rotated array precede the elements representing each successive row of the array, the last 2^N elements written to memory representing the bottom row of the rotated array.

22. The apparatus of claim 21, wherein said processor to interleave the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array, performs the interleave operation in parallel.

23. The apparatus of claim 21, wherein the SIMD unit of the processor performs the load and interleave operations in response to the execution of a single vector merge instruction.

24. The apparatus of claim 21, wherein the 2^N by 2^N array represents a subsection of a digital image stored in the memory device, and said processor is to rotate the entire digital image by iteratively rotating subsections of the digital image and writing the rotated subsections back to the memory device.

25. An apparatus for rotating a 2^N by 2^N array in the counter-clockwise direction, the apparatus comprising:

storage means to store the 2^N by 2^N array; and

means to load the 2^N by 2^N array from said storage means into means to interleave the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array, wherein said means to interleave is to iteratively perform an interleave operation a total of N times, and wherein a result of said loading and said interleaving is stored for further processing; and

after N loading and N interleaving operations, means to write the vector registers storing the results of the N loading and N interleaving operations to memory in order so that the elements representing the top row of the rotated array precede the elements representing each successive row of the array, the last 2^N elements written to memory representing the bottom row of the rotated array.

26. The apparatus of claim 25, wherein said means to interleave is to interleave the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array, wherein said interleave operation is performed in parallel.

27. The apparatus of claim 25, wherein the 2^N by 2^N array represents a subsection of a digital image stored in said storage means, and said means to interleave is to rotate the entire digital image by iteratively rotating subsections of the digital image and writing the rotated subsections back to said storage means.

28. An apparatus for rotating a 2^N by 2^N array in the clockwise direction, the apparatus comprising:

storage means to store the 2^N by 2^N array; and

means to load the 2^N by 2^N array from said storage means into means to interleave the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array follows an element from a corresponding row in the bottom half of the array, wherein said means to interleave is to iteratively perform an interleave operation a total of N times, wherein a result of said loading and said interleaving is stored in a temporary storage means for further processing; and

after N loading and N interleaving operations, means to write the vector registers storing the results of the N loading and N interleaving operations to memory in order so that the elements representing the top row of the rotated array precede the elements representing each successive row of the array, the last 2^N elements written to memory representing the bottom row of the rotated array.

29. The apparatus of claim 28, wherein said means to interleave is to interleave the 2^N elements of each row in the top half of the array with the 2^N elements of a corresponding row in the bottom half of the array so that each element from the top half of the array precedes an element from a corresponding row in the bottom half of the array, wherein said interleave operation is performed in parallel.

30. The apparatus of claim 28, wherein the 2^N by 2^N array represents a subsection of a digital image stored in said storage means, and said means to interleave is to rotate the entire digital image by iteratively rotating subsections of the digital image and writing the rotated subsections back to said storage means.