Decimal Floating Point Operand Generation via Bitwise Modification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing binary floating point (BFP) format is inefficient for decimal calculations, leading to performance penalties in applications that require decimal data processing, and the generation of decimal floating point (DFP) operands for testing is time-consuming due to the overhead of pseudo-random generation and encoding of encoded DFP operands.
Innovation Solution
A method to modify encoded DFP operands without decoding them to generate additional encoded operands for testing, reducing the effort spent on generating meaningful DFP operands and enhancing operand interdependency, thereby improving test case quality and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If pseudo-random generation is used to generate DFP operands, then test case coverage is improved, but generation time and overhead increase significantly
Solution Approach 1:
The patent pre-generates and stores a table of encoded DFP operands in memory before the actual testing begins. During testing, these pre-computed operands are retrieved from the table rather than generating them pseudo-randomly in real-time, significantly reducing the time overhead while maintaining comprehensive test coverage through the pre-generated diverse operand set
Solution Approach 2:
The patent creates copies of encoded DFP operands by modifying existing encoded operands through bit manipulation operations (such as XOR, OR, AND) on the encoded representation directly. This allows generation of varied test operands without performing time-consuming decoding and re-encoding operations, thus reducing generation time while maintaining test case diversity
2Productivity
If encoded DFP operands are generated and stored in floating point registers, then testing can proceed, but substantial time is spent on encoding operations
Solution Approach 1:
The patent performs encoding operations in advance during the test case generation phase, storing the resulting encoded DFP operands in a lookup table. During actual instruction testing, the pre-encoded operands are simply retrieved from memory rather than being encoded in real-time, eliminating the time penalty associated with repeated encoding operations and significantly improving testing throughput
Solution Approach 2:
The patent generates new encoded DFP operands by copying and modifying existing encoded operands through logical bit operations. This approach avoids the need to perform the full encoding process from scratch for each test case, reducing encoding time while still producing valid and diverse encoded operands for testing
3Ease of operation
If decoded DFP operands are used for testing, then operand manipulation is simplified, but decoding and re-encoding overhead increases
Solution Approach 1:
The patent operates directly on the encoded representation of DFP operands using bit manipulation operations such as XOR, OR, and AND on the encoded bits. This approach simplifies operand manipulation by avoiding the need to decode to binary format and then re-encode, eliminating the round-trip overhead while maintaining ease of operation through direct bitwise manipulation of the encoded structure
Data Source
AI summary
A method, system and program product are provided for establishing one or more decimal floating point (DFP) operand for facilitating testing of a decimal floating point instruction. The method includes obtaining an encoded DFP operand previously generated for testing the decimal floating point instruction, and logically modifying at least one bit of the encoded DFP operand without decoding the encoded DFP operand to obtain an additional encoded DFP operand. In one embodiment, m sequential bits of the encoded DFP operand, n randomly generated bits (wherein n=m), and a logical operation (such as an AND, OR, XOR or SHIFT) are employed in modifying the previously generated, encoded DFP operand.