Re: Questions Regarding Mathematica Kernel Supplied
- To: mathgroup at smc.vnet.net
- Subject: [mg132197] Re: Questions Regarding Mathematica Kernel Supplied
- From: Itai Seggev <itais at wolfram.com>
- Date: Mon, 13 Jan 2014 02:25:45 -0500 (EST)
- Delivered-to: l-mathgroup@mail-archive0.wolfram.com
- Delivered-to: l-mathgroup@wolfram.com
- Delivered-to: mathgroup-outx@smc.vnet.net
- Delivered-to: mathgroup-newsendx@smc.vnet.net
- References: <lao7vm$pmh$1@smc.vnet.net>
On Sat, Jan 11, 2014 at 02:36:31AM -0500, pw wrote: > Yes, if you are approaching data processing from that perspective > you are correct. > > However, if I am running massive bulk calculations this would be similar > to the technique of using 'captcha' as a tool for performing > bulk optical character recognition. > > The point is to offload bulk calculations onto unused network cycles > using a background process embedded in a web page as/with advertising. > > The calculations are paid for by ad revenue, the compute cycles > come from the web browsers viewing various random web pages, and the > standardized math libraries come from Mathematica because they are now > available and cross platform. > > I suppose a math library could be programmed for javascript (...yuck) So you seem to be envisioning some sort of commercial use case, let me point out that the usage of the RasPi distribution of M- is only free for non-commercial use. > On 14-01-10 08:30 AM, Richard Fateman wrote: > > Since the Mathematica kernel will likely be quite slow, the > > advantage you might get from running in parallel on n machines > > would be cancelled by the fact that it is n times slower than > > a conventional program on a single machine. (n = 1000 perhaps) > > > > Unless you are using some aspect of Mathematica that is unique to > > it and not available in some primarily numerical code, I suggest you > > look elsewhere. > > > > There are other frameworks for distributed computing via web sites. I > > suggest you look at SETI or folding.stanford.edu/home. > > > > Much as you might like Mathematica, it sucks for some things. > > > > > > > > On 1/9/2014 11:37 PM, pw wrote: > >> Hello, > >> > >> As some may have heard Wolfram company have decided to > >> provide the Mathematica kernel on some embedded devices > >> to achieve as much device coverage as possible. > >> > >> I have a question regarding calling the kernel, > >> from a web browser, using javascript on any operating system. > >> > >> My idea is to be able to queue parallel calculations > >> where they will be calculated by general web object activation > >> via web pages. > >> > >> By embedding calculations into media or web content, > >> distributed computing can happen via global internet idle usage while > >> web pages load. > >> > >> With the Wolfram kernel going onto devices > >> ubiquitously this would make highly distributed > >> parallel processing achievable via a subscription model. > >> > >> Is there a javascript interface for the math kernel > >> that will run calculations on the web browser (client) > >> side? > >> > >> Is there a Mathematica javascript plug-in that could > >> run calculations if the math kernel is detected on the > >> client side? > >> > >> Is there a Mathematica kernel based on javascript (!yuck!)? > >> > >> I realize that javascript is not the best language > >> to use as network middle-ware, but that's all we have > >> since the latest public decline of trust in Java. > >> > >> Thanks, > >> > >> P. > >> > > > > > > -- Itai Seggev Mathematica Algorithms R&D 217-398-0700