Abstract: A new zSeries floating-point unit has a fused multiply-add dataflow capable of supporting two architectures and fused MULTIPLY and ADD and Multiply and SUBTRACT in both RRF and RXF formats for the fused functions. Both binary and hexadecimal floating-point instructions are supported for a total of 6 formats. The floating-point unit is capable of performing a multiply-add instruction for hexadecimal or binary every cycle with a latency of 5 cycles. This supports two architectures with two internal formats with their own biases. This has eliminated format conversion cycles and has optimized the width of the dataflow. The unit is optimized for both hexadecimal and binary floating-point architecture supporting a multiply-add/subtract per cycle.

Abstract: An apparatus and method for extending digital data transmission link distances provide a mechanism in which data is transmitted over relatively long distances using a repeater to provide data buffering and handle the flow control. The repeater function incorporates large virtual lane receiver buffers in its long distance ports. Enough virtual lanes are implemented so that service level to virtual lane mapping is not required, allowing the repeaters to be made invisible to Subnet Management.

Abstract: A method, apparatus and program product for testing at least one scan chain in an electronic chip in which the scan chain is formed by shift register latches arranged in the chain having a scan path with input pins and output pins. A flush test is executed for the scan chain under test and the flush test diagnostics for the flush test are recorded. A scan test is then executed for the scan chain under test and further test diagnostics are recorded in the event either or both the flush test or the scan test fails. The recorded flush test diagnostics and further test diagnostics are then analyzed to identify a call to one or more probable failed or failing shift register latches in the tested scan chain. The further scan chain diagnostics may include Disturb, Deterministic, ABIST, LBIST and Look-Ahead diagnostics. The tests may also be conducted for different voltage levels to determine the sensitivity of the scan chain being tested to differing voltage levels.

Abstract: A method and system for extensions to earlier patents dealing with the implementation of the InterSystem Channel (ISC) link architecture. First, it describes hardware state machines that handle all valid link messaging sequences without any processor involvement. These state machines also process larger commands and responses that may be divided into multiple frame segments. Finally, the missing frame detection is expanded for the multi frame segment commands and responses.

Abstract: A method for identifying, managing, and signaling uncorrectable errors among a plurality of clusters of symmetric multiprocessors (SMPs) detects, manages and reports data errors. The method allows merging of newly detected errors, including memory, cache, control, address, and interface errors, into existing error status. Also, error status is distributed in several possible formats, including separate status signals, special UE (uncorrectable errors) ECC codewords, encoded data patterns, parity error injection, and response codepoints. The error status is also available for logging and analysis while the machine is operating, allowing for recovery and component failure isolation as soon as the errors are detected without stopping the machine.

Abstract: The present invention relates to the field of computer hardware locic circuits, and in particular to a method for verifying the proper operation of a digital logic circuit, and in particular to symbolic simulation of a gate-level netlist corresponding to said hardware logic circuit. In order to add a useful alternative in the field of functional, exhaustive simulation and of symbolic simulation, it is proposed to perform the steps of: a) analyzing symbolic expressions visible at predetermined locations within said logic; b) determining, which nets in the netlist carry complex symbolic expressions, which comprise more than one symbol; c) replacing said complex expressions with a “crunshed color”, for cutting off said complex symbolic expression from further propagation through the netlist; d) continuing said symbolic simulation including said crunched color information on predetermined nets.

Abstract: CMOS circuitry used to multiplex between data inputs suffers from high sensitivity to power supply noise, resulting in delay variations. By utilizing current controlled inverters in a multiplexer structure, power supply insensitivity can be achieved with either of two multiplexing methods. The first method places switches on the data inputs while the second places the switches on the analog bias voltages inherent to a current controlled inverter.

Abstract: ABIST apparatus with integrated directory compare logic functionality, and ABIST error detection functionality. The apparatus includes two subsystems NOR'ed together. The first subsystem is for bit-wise logically ANDing corresponding array valid bits and tag valid inputs, generating “0” for a match and “1” for a mis-match, and logically ORing the bit-wise result to generate a “1” hit if there are any bit-wise mismatches. The second subsystem further receives ABIST control logic as an input to either: (a). combine array valid bits tag valid inputs to produce valid output, or (b) compare array valid bits with tag valid inputs. The apparatus further includes logical NOR functionality for the outputs of the first and second subsystems.

Abstract: Disclosed is a method and apparatus providing a microprocessor the ability to reuse data cache content fetched during runahead execution. Said data is stored and later retrieved based upon the instruction address of an instruction which is accessing the data cache. The reuse mechanism allows the reduction of address generation interlocking scenarios with the ability to self-correct should the stored values be incorrect due to subtleties in the architected state of memory in multiprocessor systems.

Abstract: The invention relates to a method and respective system for accessing a cache memory in a computer system, wherein the cache memory is split up in at least two segments, wherein the cache memory is accessed by a plurality of competing cache memory requests via a number of commonly used input registers, wherein a cache segment model is utilized for reflecting the cache use by said competing requests, wherein cache memory requests are processed by a processing pipe and wherein each cache-request, before entering the processing pipe, is checked whether the segments of the cache memory are available at the cycle it needs, wherein said memory comprises the steps of: a) marking a segment model cell as busy with storing, if a store-request targeting to a cache segment corresponding to said model cell has received pipe access, b) blocking off from pipe access a fetch-request targeting to a segment model cell, which is marked busy with a store operation; and c) blocking off any store-request from pipe access, if at

Abstract: A domino SRAM array restore pulse generation system launches the word decode line by the same local clock as the restore pulse, thus eliminating any race issues with the word line select. This system allows the global bit select (or column select) to have fast activation by releasing the reset signal (with the earliest arriving array clock, ckl), while guaranteeing almost perfect tracking with the bit decode system. This allows for the widest possible write window; earliest release of the pre-charge in the global column select, and resetting only after the bit decode system is deactivated.

Abstract: A method of pre-aligning data for storage during instruction execution improves performance by eliminating the cycles otherwise required for data alignment. The method can convert data between ASCII and Packed Decimal format, and between Unicode Basic Latin and Packed Decimal format. Conversion to Packed Decimal format is needed for decimal hardware in a microprocessor designed to generate decimal results. Converting from Packed Decimal to ASCII and Unicode Basic Latin is necessary to report Decimal Arithmetic results in a required format for the application program. To further improve performance, all available write ports in the fixed point unit (FXU) are utilized to reduce the number of cycles necessary to store results. To prevent data fetching of the unused destination data from slowing down instruction execution, the destination locations are tested for storage access exceptions, but the data for these operands are not actually fetched.

Abstract: An SRAM design evaluation circuit topology has the gates of the SRAM cell pass Gate Field Effect Transistors (FETs) connected to the cross-coupled gates of the inverter pair of the SRAM cell. This evaluation circuit typology is used in a full cell implementation. A series of full cells are interconnected one to another in a loop to form a ring oscillator. The output of the ring is frequency divided and measured to study the read and write behavior of the cell design. Similarly, half-cells, with the gates of their pass gates grounded, are interconnected one to another to form a ring oscillator, the output of which is frequency divided and measured to help isolate pass gate impact on memory function. The modified SRAM cell topology, connected as a ring oscillator in hardware, can be used to fully characterize an SRAM cell design, without the use of peripheral read/write circuitry.

Abstract: Disclosed is a method and apparatus providing the capability to create a dynamic associative branch target buffer (BTB). A dynamically based associative BTB allows for either an increase number of entries and/or a reduction in area over current based static based BTBs while up to retaining the same confidence level of prediction accuracy.

Abstract: A domino SRAM is provided with active pull-up PFET devices that overwhelm “slow to write but very fast to read” cells and allow the cells to recover from timing mismatch situations. This approach allows the traditional “bit select” clamp to actively control the “local select” through “wired-or” PFET pull-up transistors. Separate read and write global “bit line” pairs allow the read and write performance to be optimized independently.

Abstract: A computer architecture that provides the definition of a 20 bit signed displacement value used to form the operand storage address.

Abstract: Local bit line pairs in a domino SRAM include an amplifier to amplify the voltage differential across the bit lines during a read operation if a cell in the local group of cells has been identified as a slow to read cell. The amplifier includes a transistor switch that is turned on by a timing pulse during the read operation, but only if the Array Built In Self-Test (ABIST) has detected a slow to read cell in the local group. If there is no slow cell, the amplifier is not activated, and the domino read operation is carried out. The amplifier can be used globally across the SRAM or selectively in certain sub-arrays.

Abstract: Memory is scrubbed by an improved non-linear method giving scrubbing preference to the central storage region having the characteristic of a high risk read-only memory such as the CPA region to prevent the accumulation of temporary data errors. The chip row on which the CPA resides is scrubbed after each time the scrubbing of a non-CPA chip row in a PMA completed successfully. The next non-CPA least recently scrubbed chip row would be selected for scrubbing after scrubbing completed on the CPA chip row. This in a first case provides non-linear selection methods of scrubbing central storage of computer systems to more frequently select (“select” herein encompasses the meaning of “favor”) scrub regions having the characteristic of a predominately read-only memory making those regions at a higher risk of failure than those regions having lower risk because of frequent write operations.

Abstract: A Digital Design Method which may be automated is for obtaining timing closure in the design of large, complex, high-performance digital integrated circuits. The methodincludes the use of a tuner on random logic macros that adjusts transistor sizes in a continuous domain. To accommodate this tuning, logic gates are mapped to parameterized cells for the tuning and then back to fixed gates after the tuning. Tuning is constrained in such a way as to minimize “binning errors” when the design is mapped back to fixed cells. Further, the critical sections of the circuit are marked in order to make the optimization more effective and to fit within the problem-size constraints of the tuner. A specially formulated objective function is employed during the tuning to promote faster global timing convergence, despite possibly incorrect initial timing budgets. The specially formulated objective function targets all paths that are failing timing, with appropriate weighting, rather than just targeting the most critical path.

Abstract: A computer system enabling dynamic sparing employs a standby component which is identical to three other additional components and which operates like these other three active components while the computer system is running. Any one of these three other active components can be spared out dynamically in the computer system while it is running using a result of voting scheme and connecting of these four components in such a way that the system can dynamically spare while the system is still in operation. Such dynamic sparing gives the system a better reliability and availability when compared to today's computer system.