International Business Machines Corporation

Donald R DeSota

Robert Joersz

Robert H Bell, Jr

Bruce Michael Gilbert

Rajendra D Panda

Maged M Michael

Thomas D Lovett

Luigi Brochard

Eric N Lais

Venkat Rajeev Indukuru

Carl E Love

Russell M Clapp

Davis A Miller

Sameh Sherif Sharkawi

Francois Thomas

Jaeheon Jeong

Wayne Alan Downer

Joseph H Robichaux

Rob Joersz

Robert J Safranek

Jason Frederick Cantin

Michael Grassi

Adrian C Moga

Steven Paul Hartman

Kenneth Frank Dove

William Durr

According to one aspect of the present disclosure a method and technique for managing memory access is disclosed. The method includes setting a memory databus utilization threshold for each of a plurality of processors of a data processing system to maintain memory databus utilization of the data processing system at or below a system threshold. The method also includes monitoring memory databus utilization for the plurality of processors and, in response to determining that memory databus utilization for at least one of the processors is below its threshold, reallocating at least a portion of unused databus utilization from the at least one processor to at least one of the other processors.

Reallocating unused memory databus utilization to another processor when utilization is below a threshold

A performance projection system includes a test IHS and a currently existing IHS. The performance projection system includes surrogate programs and user application software. The test IHS employs a memory that includes a virtual future IHS, currently existing IHS, surrogate programs, and user application software for determination of runtime and HW counter performance data. The user application software and surrogate programs execute on the currently existing MS to provide designers with runtime data and HW counter or microarchitecture dependent data. Designers execute surrogate programs on the future IHS to provide runtime and HW counter data. Designers normalize and weight the runtime and HW counter data to provide a representative surrogate program for comparison to user application software performance on the future IHS. Using a scaling factor, designers may generate a projection of runtime performance for the user application software executing on the future IHS.

Workload performance projection for future information handling systems using microarchitecture dependent data

A job scheduler can select a processor core operating frequency for a node in a cluster to perform a job based on energy usage and performance data. After a job request is received, an energy aware job scheduler accesses data that specifies energy usage and job performance metrics that correspond to the requested job and a plurality of processor core operating frequencies. A first of the plurality of processor core operating frequencies is selected that satisfies an energy usage criterion for performing the job based, at least in part, on the data that specifies energy usage and job performance metrics that correspond to the job. The job is assigned to be performed by a node in the cluster at the selected first of the plurality of processor core operating frequencies.

Energy-aware job scheduling for cluster environments

Queuing of received transactions that have a resource conflict is disclosed. A first node receives a first transaction from a second node, where the first transaction relates to a resource of the first node. The transaction may be a request relating to a memory line of the first node, for instance. It is determined that a second transaction that relates to this resource of the first node is already being processed by the first node. Therefore, the first transaction is enqueued in a conflict queue within the first node. The queuing may be a linked list, a priority queue, or another type of queue. Once the second transaction has been processed, the first transaction is restarted for processing by the first node. The first transaction is then processed by the first node.

Queuing of conflicted remotely received transactions

Partitioning a node of a multi-node system into more than one partition is disclosed. First resources of the node are physically partitioned into more than one partition. The first resources physically partitioned to each partition are directly inaccessible by other partitions of the node. Second resources of the node are then internally logically partitioned into the more than one partition. Each second resource internally separates transactions of one partition from transactions of other partitions. Furthermore, the node can be dynamically repartitioned into other partitions, such as a single partition, without having to take the multi-node system down. Operating system (OS) instances of the partitions may have assumptions provided to allow for dynamic partitioning, such as quiescing the processors and/or the input/output components being reconfigured, purging remote cache entries across the entire OS, etc. Each partition may have associated therewith registers that are protected from access by other partitions.

Partitioning of node into more than one partition

A performance projection system includes a test IHS and multiple currently existing IHSs. The performance projection system includes user application software and surrogate programs that execute on currently existing IHSs. The performance projection system measures user application software and surrogate program performance during execution on currently existing IHSs. The performance projection systems measures runtime program performance during execution of surrogate programs on a future semiconductor die IC design model or virtualized future system. Designers normalize and compare surrogate program runtime performance data with user application software performance data. Designers un-normalize the normalized runtime performance data to generate a projection of runtime performance on the future system.

Workload performance projection via surrogate program analysis for future information handling systems

The projected power consumption for an application program running on a future IHS is determined by generating a power proxy for the application program, and power proxies for a plurality of standard benchmarks. An algorithm correlates the power proxy for the application program, and the power proxies for the standard benchmarks to generate weighted surrogates that are combined with power proxies derived from running the benchmark surrogates on the future IHS to generate the projected power consumption.

Workload power consumption projection on information handling system

A memory controller includes ports and corresponding tables. Each port is receptive to one or more memory modules. Each table includes entries mapping memory addresses to the memory modules. Each entry corresponds to no more than one of the memory modules. The tables support asymmetric population of the memory modules within the ports; each port is capable of having a different number of memory modules relative to the other ports. The tables impose no restrictions on where the memory modules are to be inserted within the ports, both number-wise and position-wise. The tables are independently configurable; the configuration of each table is modifiable independently of the configurations of the other tables. Each table is dynamically configurable. The entries of a table are modifiable to reflect changes in the number and type of the memory modules connected, without restarting or temporarily halting the computer system containing the memory controller.

Memory controller having tables mapping memory addresses to memory modules

Selecting bits in a string-based genetic algorithm is provided. A type of genetic operation to perform is determined. Responsive to a determination to perform a crossover operation, an input comprising a pair of strings is received. The strings in the pair of strings are compared to identify a set of non-matching points. A set of points from the set of non-matching points is randomly selected, forming a set of randomly selected non-matching points. A new string for the pair of strings is generated using the set of randomly selected non-matching points.

Bit-selection for string-based genetic algorithms

An implementation-efficient, multiple-counter value hardware performance counter is disclosed. A hardware counter of one embodiment includes a memory array and a hardware incrementer. The array stores counter values that are indexable by an index constructed based at least on the number of events to which the counter values correspond. The index may be constructed as a concatenation of a number of bits binarily representing the number of events, and a number of bits binarily representing the number of qualifiers to the events. The incrementer reads the counter values from the array, increments the counter values, and writes the resulting counter values back into the array. The array may be divided into banks over which the counter values are stored, where each bank has a separate instance of the incrementer. Each bank may have a separate instance of the index that indexes only those counters stored in the bank.

Implementation-efficient multiple-counter value hardware performance counter

Non-inline transaction error correction is disclosed. Where a transaction being processed in a pipeline is determined to include a correctable error, it is output, or drained, from the pipeline into an error queue. The pipeline is switched from a normal mode of operation to a correction mode of operation. In the correction mode, a correction command is inserted into and processed within the pipeline to correct the error within the transaction. The pipeline is switched from the correction mode of operation to a restart mode of operation. In the restart mode, the transaction is reprocessed within the pipeline. The pipeline is then switched from the restart mode of operation back to the normal mode of operation.

Non-inline transaction error correction

A system, method and article of manufacture for reducing latencies associated with cache coherence directory misses on external caches in a shared distributed memory data processing system. A cache coherence directory is evaluated for possible prefetching of a directory entry into a directory cache. A prefetch miss indicator is set if the prefetch evaluated results in a directory miss. The prefetch miss indicator is consulted during subsequent processing of a memory block request corresponding to the directory entry. An accelerated snoop response action is taken if the prefetch miss indicator is set. The latency of a second lookup into the cache coherence directory, which would otherwise be required, is thereby avoided.

Growing community of inventors

Portland, OR, United States of America

Donald R DeSota