Path: csiph.com!aioe.org!.POSTED!not-for-mail From: Rod Pemberton Newsgroups: comp.lang.forth Subject: Re: Are stack machines inherently faster? Date: Sat, 3 Dec 2016 01:58:22 -0500 Organization: Aioe.org NNTP Server Lines: 51 Message-ID: <20161203015822.3ebc730e@_> References: <40b9c778-3a98-450b-b65b-6cbaee444921@googlegroups.com> <4b916947-a764-4954-ae96-b64e4bf8c046@googlegroups.com> <0733ea56-878c-4049-8d34-1b650b97e596@googlegroups.com> <570ff98e-04eb-468a-aa4c-06c31d093a8b@googlegroups.com> <8da27e1a-b79b-4c9e-a02f-2d806f3fcf65@googlegroups.com> <2016Dec2.095019@mips.complang.tuwien.ac.at> <2016Dec2.114421@mips.complang.tuwien.ac.at> NNTP-Posting-Host: n4wpt9zq8xR26Ttf9mo2BA.user.gioia.aioe.org Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Complaints-To: abuse@aioe.org X-Notice: Filtered by postfilter v. 0.8.2 X-Newsreader: Claws Mail 3.13.1 (GTK+ 2.24.13; x86_64-unknown-linux-gnu) Xref: csiph.com comp.lang.forth:53057 On Fri, 02 Dec 2016 10:44:21 GMT anton@mips.complang.tuwien.ac.at (Anton Ertl) wrote: > Concerning the software, there is the myth and dream of a pure Forth > software base that consumes much less resources. Even if that Forth dream of fewer resources was true, which I doubt, transformative technologies can't simply be equivalent, or marginally superior, or be a substitute good. They must be radically superior in order to displace the "good enough" solution that's currently in use. Also, in the time that is needed to implement such a system, existing microprocessors will double in speed, perhaps multiple times. So, the real goal becomes designing a Forth processor and computer that is perhaps 3200% faster than what exists today just to match the growth of existing processors 5 years in the future. For this computer to be transformative, it would need to be 10 years ahead of it's time, and be developed in under five years. Such rapid advancement is usually a nearly insurmountable goal. > A problem with having your own OS kernel is that there is a huge > amount of hardware that you would need to have drivers for; even Linux > has this problem to some extent for the hardware in various smartphone > SoCs. Yes, the "simple" computing hardware of the 8-bit microprocessor computer era has been over for a long time now. Forth as an OS is doable and has been done, but is not realistic for modern computer users simply due to the lack of Forth software. Situations which have dedicated Forth applications are fine, e.g., embedded. > Anyway, using a Linux kernel seems to be a good way to > circumvent this problem; it costs some memory, but the speed is ok in > my experience. > > So, the next step is to have the user space in Forth. ... > Then one would have to write all the applications in Forth. What's wrong with translation of C code? e.g., C2Forth etc. It would require far less time to convert C code to Forth than to write Forth code from scratch. Many translated programs wouldn't need any further work, i.e., low-use or no need for major improvements. Only a few programs would likely need to be coded in Forth from nothing. This would likely either be for efficiency or code maintainability. Rod Pemberton