Summary: | [patch] [new port] java/aparapi: data parallel workflows for Java | ||||||
---|---|---|---|---|---|---|---|
Product: | Ports & Packages | Reporter: | Johannes M Dieterich <jmd> | ||||
Component: | Individual Port(s) | Assignee: | Kurt Jaeger <pi> | ||||
Status: | Closed FIXED | ||||||
Severity: | Affects Only Me | CC: | kwm, pi | ||||
Priority: | --- | Keywords: | patch | ||||
Version: | Latest | ||||||
Hardware: | amd64 | ||||||
OS: | Any | ||||||
Attachments: |
|
Testbuilds@work Committed, thanks. I'm sorry for the long delay! A commit references this bug: Author: pi Date: Sun Jun 26 11:51:10 UTC 2016 New revision: 417519 URL: https://svnweb.freebsd.org/changeset/ports/417519 Log: New port: java/aparapi aparapi is an open source API for expressing data parallel workflows in Java. Originally an AMD product, Aparapi was released to open source on September 14, 2011. Aparapi is an API for expressing data parallel workloads in Java and a runtime component capable of converting the Java# bytecode of compatible workloads into OpenCL# so that it can be executed on a variety of GPU devices. WWW: https://github.com/aparapi/aparapi PR: 204024 Submitted by: dieterich@ogolem.org Changes: head/java/Makefile head/java/aparapi/ head/java/aparapi/Makefile head/java/aparapi/distinfo head/java/aparapi/files/ head/java/aparapi/files/patch-build.xml head/java/aparapi/files/patch-com.amd.aparapi.jni_build.xml head/java/aparapi/files/patch-com.amd.aparapi.jni_src_cpp_Common.h head/java/aparapi/files/patch-com.amd.aparapi_src_java_com_amd_aparapi_internal_opencl_OpenCLLoader.java head/java/aparapi/pkg-descr head/java/aparapi/pkg-plist |
Created attachment 162454 [details] port shar archive Please find attached a port of aparapi, an API for expressing data parallel workflows in Java (in our case: through OpenCL). This port is for the 1.0.0-SNAPSHOT release of upstream. Some of the patches in this port have been upstream'd by me to their master. However, their master is not stable/released. I successfully tested this port on 11-CURRENT for an accelerated matrix multiplication using lang/beignet (single precision on IvyBridge). And yes, it is fast! :-) As none of the OpenCL ports are supported for releases <10.1, this port also does not support them. Together with the ongoing OpenCL effort, I believe this is a valuable addition to our ports tree.