wrote in message news: snipped-for-privacy@googlegroups.com...
Yes, mostly by "gavino" ...
All of the electronics discussion related to computer motherboards on comp.lang.forth and comp.lang.asm.x86 of late are beginning to annoy me. No one here, except me, seems to know anything about electronic manufacturing. However, it's been a while since I worked in electronics industry. So, I added a couple of other newsgroups ...
The OP's complete post here (new and old style Google Groups):
Isn't "wacky" just a euphemism for "hack" ... ?
Yes, ARM microprocessors are becoming faster, but they're also requiring more power to do so. x86 designs are reducing their power consumption at the same time. Once the ARMs reach x86 performance levels, they won't have any advantage.
Myth? Even Gordon Moore predicted his law would slow down. He was wrong. Did something change?
Wrong. By the time it takes to "complete" your project, i.e., many years, all the hardware you just coded for will be obsolete. All hardware requires a huge amount of custom software to make it work correctly. I.e., if you buy a new PC or laptop, you'll have to start all over again in five years, or you'll need a team of people to continually keep the code current.
GHz+ performance level is overkill for those tasks.
A DX4-133Mhz was good enough, except for video. AMD K6-2 500Mhz was good enough with video.
Good luck with that. I.e., highly unlikely.
Such a design with likely use COTS ("common-off-the-shelf") components, such as standard PC chipsets, standard PC GPU's, etc.
*ALL* of that stuff is proprietary and closed. If the project is serious about that, you'll need to find, and likely pay handsomely, an experienced FPGA circuit designer to design custom logic.Good luck with that. I.e., highly unlikely.
If the project uses custom FPGA's, everyone attempting to build it will need either 1) an FPGA programmer or 2) buy programmed FPGA's from the project. If the project uses COTS, e.g., AMD or Intel motherboard chipset, then you'll have to find a way to purchase the components in small quantities.
Delirious?
Well, after seeing his webpage, he's clearly not delirious in terms of "using state of the art technology". But, he's *solidly* delirious in terms of "others of sufficient skill and resources can also build it." He clearly has access to a motherboard manufacturer - needed to get a motherboard manufactured - and is part of an electronics manufacturing firm somewhere - to get high-tech parts in small quantities for prototypes. I.e., he's probably an EE working for a major electronics firm. His motherboard is an **UNREALISTIC** achievement for a hobbyist. That's a modern multi-layer motherboard with current surface mount components. Firstly, "you" - any given hobbyist - don't know enough about electronics to design one yourself. Circuit board design is a highly-specialized trade. Secondly, "you" don't have access to the professional grade design software for required to design one. No, a professional motherboard manufacturer *will not* accept designs from open-source software even if in the correct file formats. Thirdly, "you" can't get the modern components. So, you wouldn't be able to get one of those manufactured. Given you can't do those things, you're going to have to buy a motherboard he's had manufactured and a part set from him. Even if you could buy the motherboard and parts from him, you couldn't assemble the board yourself. Doing so requires access to wave-soldering machines for through-hole components and SMT oven-soldering machines for the SMT components. Also, small electronic manufacturing firms have to submit minimum orders well into the thousands before a board manufacturer will even consider a run of boards.
For small quantities, wirewrapped or point-to-point or grid-style board or perfboard is the way to go. Unfortunately, a modern multi-layer board has way too many connections for any hobbyist method to work correctly. The length of the wire traces affect capacitance, resistance, and inductance. Engineers generally like to keep the traces as short as possible. The multiple layers helps with that.
Everything is feasible. Is it worthy? That's the question you should be asking.
I.e., if the OS is slower, the same speed, or even slightly faster than an a hobbyist x86 OS in assembly or Linux, is there point to using Forth? (No.) It must be _significantly_ better in some way. Cost can be that factor, but that's not the case here. He's used some very recent components.
If the OS must be partially coded in C, is there any point to having any of it coded in Forth? (Not really.)
Rod Pemberton