Hi Yanping, Yes it is, the first testsuit for durable native computing is under developing, this testsuit would be very simple i.e. to ask a computing service to print out all values of a Xdims durable Java linked list. however, it is just able to verify some basic/core aspects of this module, there are many issues needed to be discussed further e.g. native computing isolation, handling computing exception, pass complex results and etc. I think it would not be one for all solutions for early phase adoption but we should make it flexible as much as possible for developer to extend. In addition, some general native algorithms code e.g. sorting, map, reduce might be customized to generalize the data type to process and fit into native computing service's conventions. but I think, essentially, the specific native computing services might not be in the scope of Apache Mnemonic. Thanks.
B.R, +Gary. On 5/30/2016 3:27 PM, Yanping Wang wrote: > Hi,Gary > > While it is good to have an overview description of native computing layer. > It would be better to have a list of working items for these features, so > we can discuss and engage more developers. > > Thanks > yanping > > On Fri, May 27, 2016 at 10:07 AM, Gary <ga...@apache.org> wrote: > >> Hi Yanping, >> >> Got it, essentially we already have a durable JVM based object model >> for use to use, we certainly can use Java language to implement application >> logics with this model to leverage large capacity and persistent features >> of hybrid memory stores, furthermore, durable computing abstracts some cost >> operations on massive object graph as services that could provide >> developers a new approach to utilize different computing units such as CPU, >> GPU, FPGA to process massive objectized/blockized data directly on native >> layer. in this way, it could explicitly separate those different concerns >> for high performance analytics, Thanks. >> >> B.R, >> +Gary. >> >> On 5/26/2016 11:33 PM, Yanping Wang wrote: >>> Hi, Gary >>> >>> Regarding the following, I'd suggest you to write down clearly content >>> on what's the plan for this new feature of durable computing so others >> can >>> understand the new feature and how it works, before we are able to >> provide >>> advice, guidance, and joining the develop efforts. >>> >>> In addition, we are developing the new feature about durable >> computing, >>> I think it is relative new concept and probably will cause huge works if >>> not well designed at beginning so the advice and guidance from you and >> our >>> mentoring, community are very crucial >>> >>> >>> On Thursday, May 26, 2016, Gary <ga...@apache.org> wrote: >>> >>>> Hi Patrick, >>>> >>>> Yes please, I uploaded artifacts to Apache dist repo ( >>>> https://dist.apache.org/repos/dist/release/incubator/mnemonic) after >>>> sending out the vote results in general@ list, but I have no idea how >> to >>>> write the release about the first release. >>>> >>>> In addition, we are developing the new feature about durable >> computing, >>>> I think it is relative new concept and probably will cause huge works if >>>> not well designed at beginning so the advice and guidance from you and >> our >>>> mentoring, community are very crucial for next release of Mnemonic >>>> developing, Thanks. >>>> >>>> B.R, >>>> +Gary >>>> >>>> On 5/26/2016 11:09 AM, Patrick Hunt wrote: >>>>> Kudos on getting your first release finalized. Anything remaining that >>>> you >>>>> need help with? wrt the release I mean. >>>>> >>>>> Patrick >>>>> >>>> >> >>
signature.asc
Description: OpenPGP digital signature
