Data decompression with extra precision

• US 8,078,656 B1
• Filed: 11/16/2004
• Issued: 12/13/2011
• Est. Priority Date: 11/16/2004
• Status: Active Grant

• Current Assignee: NVIDIA Corporation
• Original Assignee: NVIDIA Corporation
• Inventors: Rogers, Douglas H., King, Gary C., Donovan, Walter E.
• Primary Examiner(s): Do, Chat C
• Application Number:
  US10/990,884
• Time in Patent Office: 2,583 Days
• Field of Search: 708200-209
• US Class Current: 708/200
• CPC Class Codes: G06T 9/00 Image coding bandwidth or redundancy reduction for static pictures H04N1/41; coding or decoding of static colour picture signals H04N1/64; methods or arrangements for coding, decoding, compressing or decompressing digital video signals H04N19/00

Abstract

Methods and systems for decompressing data are described. The relative magnitudes of a first value and a second value are compared. The first value and the second value represent respective endpoints of a range of values. The first value and the second value each have N bits of precision. Either the first or second value is selected, based on the result of the comparison. The selected value is scaled to produce a third value having N+1 bits of precision. A specified bit value is appended as the least significant bit of the other (non-selected) value to produce a fourth value having N+1 bits of precision.

Claims

1. A computer system comprising:

a processor; and

a memory unit coupled to said processor, said memory unit having computer-readable program instructions stored therein that when executed cause said computer system to implement a method for processing data, said method comprising said computer system;

comparing the magnitude of a first value and the magnitude of a second value in a compression block resident in said memory unit, wherein said first and second values are stored in a particular order relative to one another in said compression block, said first value and said second value representing respective endpoints of a range of values, said first value and said second value each having N bits, wherein said first and second values are either;

signed values normalized to the range of [−

1, 1] coded in eight (8) bits each, said 8 bits having a value in the range of [−

127, 127], or unsigned values coded in 8 bits each, said 8 bits having a value in the range of [0, 255];

selecting one of said first and second values according to the result of said comparing, thereby yielding a selected value that is one of said first and second values and an unselected value that is the other of said first and second values, wherein said selected value has a larger magnitude than said unselected value;

scaling said selected value to produce a third value representing a decompressed value of said selected value and having N+1 bits of precision;

using said particular order in which said first and second values are stored in said compression block to derive a specified one-bit value, wherein said specified one-bit value has one binary value if said first value is stored before said second value in said compression block and otherwise said specified one-bit value has the other binary value, wherein said specified one-bit value is appended to said unselected value as the least significant bit of said unselected value to produce a fourth value representing a decompressed value of said unselected value and having N+1 bits of precision; and

interpolating values that have N+1 bits of precision using said third and fourth values.

2. The computer system of claim 1 wherein, if said first and second values are signed values, then said comparing of said method uses the absolute values of said first and second values.

3. The computer system of claim 1 wherein said data comprise data for a normal map used in graphics processing.

4. A non-transitory computer-readable storage medium having program instructions stored therein for causing a computer system to perform a method comprising:

comparing the magnitude of a first value and the magnitude of a second value in a memory-resident compression block, wherein said first and second values are stored in a particular order relative to one another in said compression block, said first value and said second value representing respective endpoints of a range of values, said first value and said second value each having N bits, wherein said first and second values are either;

signed values normalized to the range of [−

1, 1] coded in eight (8) bits each, said 8 bits having a value in the range of [−

127, 127], or unsigned values coded in 8 bits each, said 8 bits having a value in the range of [0, 255];

selecting one of said first and second values according to the result of said comparing, thereby yielding a selected value that is one of said first and second values and an unselected value that is the other of said first and second values, wherein said selected value has a larger magnitude than said unselected value;

scaling said selected value to produce a third value representing a decompressed value of said selected value and having N+1 bits of precision; and

using said particular order in which said first and second values are stored in said compression block to derive a specified single bit value, wherein said specified single bit value has one binary value if said first value is stored in a position in said compression block that is read before said second value and otherwise said single bit value has the other binary value, wherein said single bit value is appended to said unselected value as the least significant bit of said unselected value to produce a fourth value representing a decompressed value of said unselected value and having N+1 bits of precision.

5. The non-transitory computer-readable storage medium of claim 4 wherein, if said first and second values are signed values, then said comparing uses the absolute values of said first and second values.

6. The non-transitory computer-readable storage medium of claim 4 wherein said range of values comprise data for a normal map used in graphics processing.