I'm not sure if we need special API support for GPUs. You can already use GPUs on individual executor nodes to build your own applications. If we want to leverage GPUs out of the box, I don't think the solution is to provide GPU specific APIs. Rather, we should just switch the underlying execution to GPUs when it is more optimal.
Anyway, I don't want to distract this topic, If you want to discuss more about GPUs, please start a new thread. On Mon, Dec 21, 2015 at 11:18 PM, Allen Zhang <allenzhang...@126.com> wrote: > plus dev > > > > > > > 在 2015-12-22 15:15:59，"Allen Zhang" <allenzhang...@126.com> 写道： > > Hi Reynold, > > Any new API support for GPU computing in our 2.0 new version ? > > -Allen > > > > > 在 2015-12-22 14:12:50，"Reynold Xin" <r...@databricks.com> 写道： > > FYI I updated the master branch's Spark version to 2.0.0-SNAPSHOT. > > On Tue, Nov 10, 2015 at 3:10 PM, Reynold Xin <r...@databricks.com> wrote: > >> I’m starting a new thread since the other one got intermixed with feature >> requests. Please refrain from making feature request in this thread. Not >> that we shouldn’t be adding features, but we can always add features in >> 1.7, 2.1, 2.2, ... >> >> First - I want to propose a premise for how to think about Spark 2.0 and >> major releases in Spark, based on discussion with several members of the >> community: a major release should be low overhead and minimally disruptive >> to the Spark community. A major release should not be very different from a >> minor release and should not be gated based on new features. The main >> purpose of a major release is an opportunity to fix things that are broken >> in the current API and remove certain deprecated APIs (examples follow). >> >> For this reason, I would *not* propose doing major releases to break >> substantial API's or perform large re-architecting that prevent users from >> upgrading. Spark has always had a culture of evolving architecture >> incrementally and making changes - and I don't think we want to change this >> model. In fact, we’ve released many architectural changes on the 1.X line. >> >> If the community likes the above model, then to me it seems reasonable to >> do Spark 2.0 either after Spark 1.6 (in lieu of Spark 1.7) or immediately >> after Spark 1.7. It will be 18 or 21 months since Spark 1.0. A cadence of >> major releases every 2 years seems doable within the above model. >> >> Under this model, here is a list of example things I would propose doing >> in Spark 2.0, separated into APIs and Operation/Deployment: >> >> >> APIs >> >> 1. Remove interfaces, configs, and modules (e.g. Bagel) deprecated in >> Spark 1.x. >> >> 2. Remove Akka from Spark’s API dependency (in streaming), so user >> applications can use Akka (SPARK-5293). We have gotten a lot of complaints >> about user applications being unable to use Akka due to Spark’s dependency >> on Akka. >> >> 3. Remove Guava from Spark’s public API (JavaRDD Optional). >> >> 4. Better class package structure for low level developer API’s. In >> particular, we have some DeveloperApi (mostly various listener-related >> classes) added over the years. Some packages include only one or two public >> classes but a lot of private classes. A better structure is to have public >> classes isolated to a few public packages, and these public packages should >> have minimal private classes for low level developer APIs. >> >> 5. Consolidate task metric and accumulator API. Although having some >> subtle differences, these two are very similar but have completely >> different code path. >> >> 6. Possibly making Catalyst, Dataset, and DataFrame more general by >> moving them to other package(s). They are already used beyond SQL, e.g. in >> ML pipelines, and will be used by streaming also. >> >> >> Operation/Deployment >> >> 1. Scala 2.11 as the default build. We should still support Scala 2.10, >> but it has been end-of-life. >> >> 2. Remove Hadoop 1 support. >> >> 3. Assembly-free distribution of Spark: don’t require building an >> enormous assembly jar in order to run Spark. >> >> > > > > > > > >
