International Business Machines Corporation

Tsinghua University

I-Hsin Chung

Hui-Fang Wen

Wei Tan

Huan Hu

David Joseph Klepacki

Guojing Cong

Simone Sbaraglia

Jinjun Xiong

Kattamuri Ekanadham

Seetharami R Seelam

Zaid Qureshi

Hubertus Franke

Bhuvana Ramabhadran

Chia-Yu Chen

Yoonho Park

Pin-Yu Chen

Yasushi Negishi

Sijia Liu

Hao Yu

Paul Gregory Crumley

Eun Kyung Lee

Robert E Walkup

Hiroki Murata

Wen-mei William Hwu

Che-Rung Lee

Weichung Wang

Huifang Wen

Carl Pearson

Tsung-Yi Ho

Yun-Yun Tsai

A temperature-aware task scheduling method, system, and computer program product for facilitating a task in a multi-graphical processing unit (GPU) environment, includes executing the task to a GPU in the multi-GPU environment based on a thermal characteristic of the GPU as compared to the other GPUs in the multi-GPU environment.

System, method and recording medium for temperature-aware task scheduling

Transfer learning in machine learning can include receiving a machine learning model. Target domain training data for reprogramming the machine learning model using transfer learning can be received. The target domain training data can be transformed by performing a transformation function on the target domain training data. Output labels of the machine learning model can be mapped to target labels associated with the target domain training data. The transformation function can be trained by optimizing a parameter of the transformation function. The machine learning model can be reprogrammed based on input data transformed by the transformation function and a mapping of the output labels to target labels.

Transfer learning with machine learning systems

A temperature-aware task scheduling method, system, and computer program product, includes learning a condition to accept a task to the GPU based on a prior execution of the task on the GPU according to a varying thermal characteristic of the GPU.

Embodiments for providing critical path analysis of active trace files in a cloud computing environment. A critical path may be identified using a trace of time spans and activities of a plurality of applications, wherein the critical path is a set of activities having time spans free of overlap with other activities.

Critical path analysis of activity trace files

Embodiments for providing direct access to non-volatile memory by a processor. One or more accelerators may be provided, via an application programming interface ('API'), direct access to non-volatile storage independent of a host central processing unit ('CPU') on a control path or data path to perform a read operation and write operation of data.

Providing direct data access between accelerators and storage in a computing environment, wherein the direct data access is independent of host CPU and the host CPU transfers object map identifying object of the data

A temperature-aware task scheduling method, system, and computer program product, includes obtaining a temperature of the GPU and accepting and executing the task to the GPU, in response to the obtaining a temperature of the GPU.

A method and an apparatus of allocating available resources in a cluster system with learning models and tuning methods are provided. The learning model may be trained from historic performance data of previously executed jobs and used to project a suggested amount of resources for execution of a job. The tuning process may suggest a configuration for the projected amount of resources in the cluster system for an optimal operating point. An optimization may be performed with respect to a set of objective functions to improve resource utilization and system performance while suggesting the configuration. Through many executions and job characterization, the learning/tuning process for suggesting the configuration for the projected amount of resources may be improved by understanding correlations of historic data and the objective functions.

Cognitive scheduler

Embodiments for crash recoverability for graphics processing units (GPUs) by a processor. GPU application data and kernel execution state of one or more GPUs may be checkpointed. The checkpointed GPU application data and the kernel execution state may be recovered. The checkpointed GPU application data and the kernel execution state may be persisted on non-volatile memory.

Crash recoverability for graphics processing units (GPU) in a computing environment

A data-centric reduction method, system, and computer program product include configuring a similarity threshold and a correlation threshold for an entire data set from at least two back-end nodes, reducing the entire data set to a reduced data set from the at least two back-end nodes sent to a front-end node by removing data based on the similarity threshold and the correlation threshold, and after the front-end receives the reduced data set, reconstructing the entire data set from the reduced data set using the similarity threshold and correlation threshold.

Data-centric reduction network for cluster monitoring

A temperature-aware task scheduling method, system, and computer program product, include determining a change in an operation intensity factor of the GPU from a previous state and modifying the operation intensity factor, in response to the determining the change in the operation intensity factor from the previous state.

System, method and computer program product for temperature-aware task scheduling

A temperature-aware task scheduling method, system, and computer program product, includes the GPU, receiving a request to execute the task, collecting task information including an intensiveness factor of a computation by an arithmetic logic unit (ALU) and a memory usage of a dynamic random-access memory (DRAM) for the task, obtaining a temperature of the ALU and a temperature of the DRAM, and accepting the task to the GPU based on the intensiveness factor, the ALU temperature, and the DRAM temperature.

Mapping tasks to physical processors in parallel computing system may include partitioning tasks in the parallel computing system into groups of tasks, the tasks being grouped according to their communication characteristics (e.g., pattern and frequency); mapping, by a processor, the groups of tasks to groups of physical processors, respectively; and fine tuning, by the processor, the mapping within each of the groups.

Hierarchical task mapping

A profiling tool identifies a code region with a false sharing potential. A static analysis tool classifies variables and arrays in the identified code region. A mapping detection library correlates memory access instructions in the identified code region with variables and arrays in the identified code region while a processor is running the identified code region. The mapping detection library identifies one or more instructions at risk, in the identified code region, which are subject to an analysis by a false sharing detection library. A false sharing detection library performs a run-time analysis of the one or more instructions at risk while the processor is re-running the identified code region. The false sharing detection library determines, based on the performed run-time analysis, whether two different portions of the cache memory line are accessed by the generated binary code.

Methodology for fast detection of false sharing in threaded scientific codes

A method for application performance data collection includes steps or acts of: customizing a performance tool for collecting application performance data of an application; modifying the application by inserting the performance tool while the application does not need to be rebuilt from the source; executing the application; and collecting the application execution performance data such that only interesting data is collected. Customizing the performance tool proceeds by implementing at least one configurable tracing function that can be programmed by the user; compiling the function(s) into an object file; and inserting the object file into the performance tool using binary instrumentation.

Binary programmable method for application performance data collection

During runtime of a binary program file, streams of instructions are executed and memory references, generated by instrumentation applied to given ones of the instructions that refer to memory locations, are collected. A transformation is performed, based on the executed streams of instructions and the collected memory references, to obtain a table. The table lists memory events of interest for active data structures for each function in the program file. The transformation is performed to translate memory addresses for given ones of the instructions and given ones of the data structures into locations and variable names in a source file corresponding to the binary file. At least the memory events of interest are displayed, and the display is organized so as to correlate the memory events of interest with corresponding ones of the data structures.

Growing community of inventors

Chappaqua, NY, United States of America

I-Hsin Chung