Abstract: An initializable repair circuit and method are provided to facilitate, when enabled, selective replacement of a table-driven output value provided by a lookup structure. The initializable repair circuit includes a compare circuit to identify a cell of the lookup structure based, at least in part, on a first input value and a second input value. The table-driven output value is ascertained, at least in part, using a cell value of the identified cell. The initializable repair circuit further includes a repair enable register and a logic circuit. The repair enable register contains an enable repair indicator to be set when at least one cell value is known to be incorrect, and the logic circuit replaces the incorrect table-driven output value provided by the lookup structure with an initialized replacement value based, at least in part, on the enable repair indicator being set in the repair enable register.

Abstract: Arithmetic logic circuitry is provided for performing a floating point arithmetic add/subtract operation on first and second floating point numbers. The method includes: generating a guard digit for the first or second number by transforming the first and second numbers by a compressing function; determining a result depending on the arithmetic operation, a sum of the transformed floating point numbers, and first and second differences of the transformed floating point numbers, and determining a corresponding result plus one by additionally adding a value of one to the result; generating injection values for rounding the final result; generating injection carry values based on the transformed first and second numbers and the injection values; and selecting the final result from the result, the result plus one, and a least significant digit, based on the injection carry values and the end around carry signals.

Abstract: An arithmetic operation in a data processing unit, preferably by iterative digit accumulations, is proposed. An approximate result of the arithmetic operation is computed iteratively. Concurrently at least two supplementary values of the approximate result of the arithmetic operation are computed, and the final result selected from one of the values of the approximate result and the at least two supplementary values of the arithmetic operation depending on the results of the last iteration step.

Abstract: A tool for supporting vector operations in a scalar data path. The tool determines a location for a split in the scalar mode configuration to enable vector mode operations. The tool determines the number of coarse shift multiplexers in conflict at a bit position receiving data from both the left half and the right half of the vector mode configuration. The tool duplicates a coarse shift multiplexer in conflict at the bit position receiving data from both the left half and the right half of the vector mode configuration. The tool duplicates an intermediate data signal in conflict at a signal position receiving data from both the left half and the right half of the vector mode configuration. The tool receives a control signal to split the scalar mode configuration and shift the leading zeros across the left half and the right half of the vector mode configuration.

Abstract: A condition code can depend upon a numerical output of a floating point operation for a processing pipeline. A classification can be determined for the floating point operation of a received instruction. In response to the classification and using condition determination logic, a value can be calculated for the condition code by inferring from data that is available from the processing pipeline before the numerical output is available. The value for the condition code can be provided to branch decision logic of the processing pipeline.

Abstract: A condition code can depend upon a numerical output of a floating point operation for a processing pipeline. A classification can be determined for the floating point operation of a received instruction. In response to the classification and using condition determination logic, a value can be calculated for the condition code by inferring from data that is available from the processing pipeline before the numerical output is available. The value for the condition code can be provided to branch decision logic of the processing pipeline.

Abstract: A system and method for controlling post-silicon configurable instruction behavior are provided. For example, the method includes receiving data related to a compute circuit. The method also includes detecting a data pattern in the data. The method further includes determining that the data pattern is a special case that the compute circuit may handle improperly. The method also includes selecting a value from a post-silicon configurable data set based on the detected data. Further, the method includes changing a behavior of the compute circuit to produce a different output result based on the value selected from the post-silicon configurable data set.

Abstract: A residue-based error checking mechanism is provided for checking for error in a shift operation of a shifter. The checking includes: partitioning an input vector into the shifter into one or more bit groups of bit width W; generating a predicted residue on the input vector being shifted, the generating including masking out any bit group of bit width W fully shifted out by the shift operation from contributing to the predicted residue, and the generating accounting for any bits of a bit group of the input vector partially shifted out by the shift operation; generating a result residue on a result vector of the shift operation; and comparing the result residue with the predicted residue to check for an error in the result vector of the shift operation.

Abstract: A method for hiding implicit bit corrections in a partial product adder array in a binary and hexadecimal floating-point multiplier such that no additional adder stages are needed for the implicit bit corrections. Two leading-one correction terms are generated for the fraction in the multiplier floating-point number and two leading-one correction terms are generated for the fraction in the multiplicand floating-point number. The floating-point numbers may be single-precision or double-precision. Each leading-one correction term for the single-precision case is appended to the left of an intermediate partial product sum in the adder array that is an input to an adder so as to not to extend the bits in the input further to the left than the bits in another input to the adder. Each leading-one correction term for the double-precision case replaces an adder input that is unused when base-2 floating-point numbers are multiplied.

Abstract: A method for delocalizing an error checking on a data in a pipelined processor from the data checked. A first check-data is generated at a first location on a first data. A second location receives the first data and the first check-data. A second check-data is generated on the first data and the first check-data is compared with the second check-data at the second location. A second data is generated from the first data and a third check-data is generated on the second data at the second location. A third check-data is generated on the second data at the second location and the second data is transferred to a third location. The third check-data is transferred to a fourth location. A fourth check-data is generated on the second data and is transferred to the fourth location. The fourth check-data and the third check-data are compared at the fourth location.

Abstract: A condition code can depend upon a numerical output of a floating point operation for a processing pipeline. A classification can be determined for the floating point operation of a received instruction. In response to the classification and using condition determination logic, a value can be calculated for the condition code by inferring from data that is available from the processing pipeline before the numerical output is available. The value for the condition code can be provided to branch decision logic of the processing pipeline.

Abstract: A condition code can depend upon a numerical output of a floating point operation for a processing pipeline. A classification can be determined for the floating point operation of a received instruction. In response to the classification and using condition determination logic, a value can be calculated for the condition code by inferring from data that is available from the processing pipeline before the numerical output is available. The value for the condition code can be provided to branch decision logic of the processing pipeline.

Abstract: Arithmetic logic circuitry is provided for performing a floating point arithmetic add/subtract operation on first and second floating point numbers. The method includes: generating a guard digit for the first or second number by transforming the first and second numbers by a compressing function; determining a result depending on the arithmetic operation, a sum of the transformed floating point numbers, and first and second differences of the transformed floating point numbers, and determining a corresponding result plus one by additionally adding a value of one to the result; generating injection values for rounding the final result; generating injection carry values based on the transformed first and second numbers and the injection values; and selecting the final result from the result, the result plus one, and a least significant digit, based on the injection carry values and the end around carry signals.

Abstract: Embodiments include a method for temporary pipeline marking for processor error workarounds. The method includes monitoring an execution unit pipeline of a processor for an event associated with a programmable instruction operational code that is predetermined to cause a stuck state resulting in an errant instruction execution. The execution unit pipeline is marked for a workaround action based on detecting the event. A clearing action is triggered based on the marking, where the triggering is conditionally triggered by a next instruction in the execution unit pipeline having a same instruction type as the programmable instruction operational code. The marking of the pipeline is cleared based on the triggering of the clearing action, where the clearing action is a subsequent pipeline flush event based on the next instruction having the same instruction type reaching a same pipeline stage that results in a stuck state prior to completion of the next instruction.

Abstract: Arithmetic logic circuitry is provided for performing a floating point arithmetic add/subtract operation on first and second floating point numbers. The method includes: generating a guard digit for the first or second number by transforming the first and second numbers by a compressing function; determining a result depending on the arithmetic operation, a sum of the transformed floating point numbers, and first and second differences of the transformed floating point numbers, and determining a corresponding result plus one by additionally adding a value of one to the result; generating injection values for rounding the final result; generating injection carry values based on the transformed first and second numbers and the injection values; and selecting the final result from the result, the result plus one, and a least significant digit, based on the injection carry values and the end around carry signals.

Abstract: Embodiments include a method for temporary pipeline marking for processor error workarounds. The method includes monitoring a pipeline of a processor for an event that is predetermined to place the processor in a stuck state that results in an errant instruction execution result due to the stuck state or repeated resource contention causing performance degradation. The pipeline is marked for a workaround action based on detecting the event. A clearing action is triggered based on the marking of the pipeline. The marking of the pipeline is cleared based on the triggering of the clearing action.

Abstract: Embodiments include a computer system for temporary pipeline marking for processor error workarounds, the computer system having a processor configured to perform a method. The method includes monitoring a pipeline of the processor for an event that is predetermined to place the processor in a stuck state that results in an errant instruction execution result due to the stuck state or repeated resource contention causing performance degradation. The pipeline is marked for a workaround action based on detecting the event. A clearing action is triggered based on the marking of the pipeline. The marking of the pipeline is cleared based on the triggering of the clearing action.

Abstract: A method for hiding implicit bit corrections in a partial product adder array in a binary and hexadecimal floating-point multiplier such that no additional adder stages are needed for the implicit bit corrections. Two leading-one correction terms are generated for the fraction in the multiplier floating-point number and two leading-one correction terms are generated for the fraction in the multiplicand floating-point number. The floating-point numbers may be single-precision or double-precision. Each leading-one correction term for the single-precision case is appended to the left of an intermediate partial product sum in the adder array that is an input to an adder so as to not to extend the bits in the input further to the left than the bits in another input to the adder. Each leading-one correction term for the double-precision case replaces an adder input that is unused when base-2 floating-point numbers are multiplied.

Abstract: Software that combines parity bits with error correcting codes (ECC) such that a subset of ECC bits is also used for parity purposes, by performing the following steps: (i) providing a first set of redundant bit(s) in a data block, where the first set of redundant bit(s) is adapted to detect and/or correct errors in the data block; (ii) providing, within the first set of redundant bit(s), a first sub-set of parity bit(s), where the first sub-set of parity bit(s) is adapted to provide single bit error detection for the data block; and (iii) determining, based, at least in part, on a first set of data read requirements, whether to use the first set of redundant bit(s) and/or the first sub-set of parity bit(s) to detect and/or correct potential errors while reading data on the data block.

Abstract: Embodiments include a computer system for temporary pipeline marking for processor error workarounds, the computer system having a processor configured to perform a method. The method includes monitoring a pipeline of the processor for an event that is predetermined to place the processor in a stuck state that results in an errant instruction execution result due to the stuck state or repeated resource contention causing performance degradation. The pipeline is marked for a workaround action based on detecting the event. A clearing action is triggered based on the marking of the pipeline. The marking of the pipeline is cleared based on the triggering of the clearing action.