摘要:
Techniques for eradicating operating system jitter-induced slowdown are provided. The techniques include allocating one or more computing resources to one or more logical partitions of one or more parallel programs in proportion of one or more cycles consumed by one or more sources of operating system jitter in each compute phase in each of the one or more logical partitions.
摘要:
Techniques for enabling high-performance computing are provided. The techniques include resizing a logical partition in a non-dedicated compute cluster server to enable high-performance computing, wherein a high performance computing application is executed such that the high performance computing application is configured to complete execution of each of one or more application threads at a similar instance as a slowest thread in the cluster, and wherein the non-dedicated compute cluster comprises one or more servers and the logical partition is created by partitioning one or more server resources.
摘要:
A method for tuning performance of an operating system, the method comprising identifying all sources of operating system jitter; measuring the impact of each of the operating system jitter source; and tuning performance of the operating system, preferably by use of different approaches/techniques, which could include removing the sources of operating system jitter and/or delaying their execution and/or smoothening their execution over a longer period of time. Computer program code and systems are also provided.
摘要:
Techniques for eradicating operating system jitter-induced slowdown are provided. The techniques include allocating one or more computing resources to one or more logical partitions of one or more parallel programs in proportion of one or more cycles consumed by one or more sources of operating system jitter in each compute phase in each of the one or more logical partitions.
摘要:
Techniques for emulating operating system jitter on a platform using a given trace are provided. The techniques include calculating a scale factor, wherein the scale factor is equal to a maximum of measured overhead of introducing synthetic jitter on the platform and a resolution of one or more timer calls on the platform, scaling up each of one or more jitter values and each of one or more gaps between each of one or more jitter instances in the trace and an execution period of a benchmark parallel application using the scale factor, introducing synthetic jitter using each of the one or more scaled jitter values and each of the one or more scaled gaps from the trace while running the benchmark parallel application for the scaled execution period to emulate operating system jitter on a platform, and scaling down one or more final time measurements from the benchmark parallel application by the scale factor.
摘要:
Techniques for emulating operating system jitter on a platform using a given trace are provided. The techniques include calculating a scale factor, wherein the scale factor is equal to a maximum of measured overhead of introducing synthetic jitter on the platform and a resolution of one or more timer calls on the platform, scaling up each of one or more jitter values and each of one or more gaps between each of one or more jitter instances in the trace and an execution period of a benchmark parallel application using the scale factor, introducing synthetic jitter using each of the one or more scaled jitter values and each of the one or more scaled gaps from the trace while running the benchmark parallel application for the scaled execution period to emulate operating system jitter on a platform, and scaling down one or more final time measurements from the benchmark parallel application by the scale factor.
摘要:
Techniques for enabling high-performance computing are provided. The techniques include resizing a logical partition in a non-dedicated compute cluster server to enable high-performance computing, wherein a high performance computing application is executed such that the high performance computing application is configured to complete execution of each of one or more application threads at a similar time as a slowest thread in the cluster, and wherein the non-dedicated compute cluster comprises one or more servers and the logical partition is created by partitioning one or more server resources.
摘要:
A method for tuning performance of an operating system, the method comprising identifying all sources of operating system jitter; measuring the impact of each of the operating system jitter source; and tuning performance of the operating system, preferably by use of different approaches/techniques, which could include removing the sources of operating system jitter and/or delaying their execution and/or smoothening their execution over a longer period of time. Computer program code and systems are also provided.
摘要:
Techniques, an apparatus and an article of manufacture for management of personalized advertisements through the mass media. A method includes capturing content of an advertisement that is displayed publically and/or publically accessible to a plurality of individuals through a mass media mechanism by recording an image of the advertisement using a hand-held device, storing the content of the advertisement in a user-dedicated repository, and managing the content of the advertisement in the user-dedicated repository to facilitate distribution of the advertisement through the user to one or more additional individual users.
摘要:
Methods and arrangements for the granular distribution of privileges to contextual communities. A user request for an active capability is assimilated. A code is generated which permits exercise of the active capability by the user and at least one secondary recipient. Execution of the active capability is permitted responsive to an attempt at executing the active capability by at least one taken from the group consisting of: the user and the at least one secondary recipient, and execution of the active capability is denied responsive to an attempt at executing the active capability by a recipient other than the user and the at least one secondary recipient.