Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.forth > #44319 > unrolled thread

Xmas

Back to article view | Back to comp.lang.forth

Contents

  Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-23 09:01 -0800
    Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-23 09:44 -0800
      Re: Xmas Ron Aaron <rambamist@gmail.com> - 2015-12-23 20:55 +0200
        Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-23 11:03 -0800
    Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-23 14:10 -0500
    Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-23 11:36 -0800
      Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-23 14:54 -0500
      Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-23 16:47 -0800
    Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-23 13:01 -0800
      Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-23 13:15 -0800
        Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-23 21:20 -0800
          Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-23 22:11 -0800
            Re: Xmas Albert van der Horst <albert@spenarnc.xs4all.nl> - 2015-12-24 11:52 +0000
      Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-23 16:26 -0500
        Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-23 18:20 -0800
          Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-23 21:45 -0500
            Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-23 19:29 -0800
      Re: Xmas hughaguilar96@gmail.com - 2015-12-23 19:45 -0800
        Re: Xmas Raimond Dragomir <raimond.dragomir@gmail.com> - 2015-12-23 22:46 -0800
          Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-23 23:22 -0800
          Re: Xmas hughaguilar96@gmail.com - 2015-12-26 16:41 -0800
          Re: Xmas Waldek Hebisch <hebisch@antispam.uni.wroc.pl> - 2015-12-30 01:55 +0000
            Re: Xmas hughaguilar96@gmail.com - 2015-12-30 19:27 -0800
              Re: Xmas Waldek Hebisch <hebisch@antispam.uni.wroc.pl> - 2016-01-01 18:13 +0000
        Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-24 11:13 -0800
    Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-23 21:33 -0800
      Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-23 22:14 -0800
    Xmas Mikael Nordman <oh2aun@gmail.com> - 2015-12-24 04:25 -0800
    Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-26 21:32 -0800
      Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-27 01:07 -0500
      Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-26 22:10 -0800
    Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-26 23:43 -0800
      Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-27 10:56 -0500
        Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-27 09:01 -0800
          Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-27 12:31 -0500
            Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-27 09:59 -0800
    Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-27 15:56 -0800
      Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-27 16:14 -0800
        Re: Xmas Chris Curl <ccurl609@gmail.com> - 2015-12-27 17:58 -0800
      Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-27 19:58 -0500
        Re: Xmas Paul Rubin <no.email@nospam.invalid> - 2015-12-27 18:07 -0800
        Re: Xmas jim.brakefield@ieee.org - 2015-12-27 18:59 -0800
          Re: Xmas rickman <gnuarm@gmail.com> - 2015-12-27 22:26 -0500

Page 2 of 3 — ← Prev page 1 [2] 3  Next page →

#44366

On Wednesday, December 23, 2015 at 11:46:34 PM UTC-7, Raimond Dragomir wrote:
> > One thing about the AVR is that the code is very bulky due to having 16-bit opcodes operating on 8-bit data so you need more than one instruction for most 16-bit data operations --- so there is a limit to how much of the system can be written in assembly-language due to the non-volatile memory running out of room --- also, the AVR lacks a short CALL instruction, so you can't put common functions in low memory and save space in the CALL (fixing this problem was one of my motivations in designing the Stundurd processor).
> > 
> AVR do has a short call, it's the "rcall" instruction. It takes only two bytes
> (one word) and has a +/-4K words range (you can go anywhere in a 8K bytes vicinity).

I'm aware of the RCALL instruction. I meant that there is no CALL for functions in low memory, so you can't put your common functions (such as + etc.) in low memory and get short calls to them from your program. Fixing this problem was actually the original motivation of the Stundurd design --- a short CALL that uses absolute rather than relative addressing.

One nice thing about the RCALL is that it can be set to wrap-around, so if you have a 4KW non-volatile memory you can use RCALL everywhere. If you have more than 4KW though, then you have to use a mixture of CALL and RCALL. The assembler that comes with Atmel Studio is pretty crude. It doesn't figure out if your call is in reach of RCALL or not, but rather it requires you to explicitly use CALL or RCALL --- if you use RCALL when it is out of range, the assembler aborts --- so, as a practical matter, it is easiest to just use CALL throughout and ignore the wasted time and memory problem. I was told that the GNU AVR assembler is smarter and will compile CALL or RCALL as appropriate, but I haven't tried it. Maybe I should --- another problem I found with Atmel Studio is that it is ponderously slow (I heard it is written in Visual Basic) --- all in all, I was not impressed with Atmel Studio.

If I do resurrect the AVR cross-compiler I will just write my own assembler in Forth --- the AVR is a pretty simple processor, so this shouldn't be difficult --- still though, it is a lot of work for something that will be most likely never be used by anybody except myself. I don't have any money-making project involving the AVR or any other micro-controller --- if I did have a commercial project going, the AVR seems like an unlikely choice as it is the last 8-bit processor still in use, but there are many 16-bit processors available that are much more powerful --- maybe it is super inexpensive, but that would only be an issue if the project involved high volume, which is pretty unlikely.

Those Arduino boards are expensive toys --- they have only 2KB of RAM typically --- they seem to be mostly used by teenagers for controlling animatronic puppets (scare your little brother, or at least your dog).

[toc] | [prev] | [next] | [standalone]

#44427

Raimond Dragomir <raimond.dragomir@gmail.com> wrote:
> > One thing about the AVR is that the code is very bulky due to having 16-bit opcodes operating on 8-bit data so you need more than one instruction for most 16-bit data operations --- so there is a limit to how much of the system can be written in assembly-language due to the non-volatile memory running out of room --- also, the AVR lacks a short CALL instruction, so you can't put common functions in low memory and save space in the CALL (fixing this problem was one of my motivations in designing the Stundurd processor).
> > 
<snip>
> Despite having a 16bit instruction length, the code tends to be very
> compact, usually on par with other arch that have variable length instruction
> in bytes. For example, an AVR will beat 8051, at least for programs above 4K.

8051 is probably the worst.  But compared to other processors AVR
loses badly.  I compared compiled code for FAT filesystem (pure
computations, no actual access to media which is device specific)
and for ATmege328 I got 2.6k, for MSP430 2k and for Cortex-M0 1.4k.
So 32 bit processor was almost 2 times better.  This was from C
compiler (in all cases gcc), code used fair mixture of 8, 16 and
32 bit operations.  Forth is likely to have different statistic
due to small routines and higher proportion of calls.  However,
native Forth on AVR still will have problem due larger size
of computational instructions.

For native Forth probably the worst featue of AVR is that you
can not execute code from RAM.

> (the atmega328). Arduino is expensive by today standards, but are very
> common. 

Well, Chinese Uno clones are at $2.65 with shiping
included.  I got 10 pieces of Chinese clone of Arduino
Pro Mini at $1.8 and recently I saw lower prices (IIRC $1.3).
The only things with comparably low price are Chinese
mini development boards for STM32F103R8 ($2.65).

-- 
                              Waldek Hebisch

[toc] | [prev] | [next] | [standalone]

#44456

On Tuesday, December 29, 2015 at 6:55:37 PM UTC-7, Waldek Hebisch wrote:
> Raimond Dragomir <raimond.dragomir@gmail.com> wrote:
> > Despite having a 16bit instruction length, the code tends to be very
> > compact, usually on par with other arch that have variable length instruction
> > in bytes. For example, an AVR will beat 8051, at least for programs above 4K.
> 
> 8051 is probably the worst.  But compared to other processors AVR
> loses badly.  I compared compiled code for FAT filesystem (pure
> computations, no actual access to media which is device specific)
> and for ATmege328 I got 2.6k, for MSP430 2k and for Cortex-M0 1.4k.
> So 32 bit processor was almost 2 times better.  This was from C
> compiler (in all cases gcc), code used fair mixture of 8, 16 and
> 32 bit operations.  Forth is likely to have different statistic
> due to small routines and higher proportion of calls.  However,
> native Forth on AVR still will have problem due largerha size
> of computational instructions.

I think the 8051 was really meant for small assembly-language programs that use only global variables and all of them in the low memory --- it doesn't support either the C-style or Forth-style parameter stack efficiently --- this doesn't mean that it is a bad processor; it works well within the limited context that it was designed for.

> For native Forth probably the worst featue of AVR is that you
> can not execute code from RAM.

One guy told me that he had a Forth that had threaded code in both RAM and non-volatile --- his NEXT would compare the cfa to a constant to know which kind of memory the code was in --- for STC however, as you say, code can't be in RAM because the AVR is Harvard-Architecture.

The MiniForth was Harvard-Architecture too --- that is why MFX was a cross-compiler.

The 8051 (and 8032, 80c320, etc.) is not Harvard Architecture --- the 8032 has external memory, so it is possible to tie the high-bit together so the program-memory and data-memory are the same thing.

> > (the atmega328). Arduino is expensive by today standards, but are very
> > common. 
> 
> Well, Chinese Uno clones are at $2.65 with shiping
> included.  I got 10 pieces of Chinese clone of Arduino
> Pro Mini at $1.8 and recently I saw lower prices (IIRC $1.3).
> The only things with comparably low price are Chinese
> mini development boards for STM32F103R8 ($2.65).

This is most likely pirated --- I very much doubt that anybody in Beijing purchased the right to do this --- so, if you buy from China, you can't be too surprised when the Chinese pirate your own designs later on.

To a large extent, the reason why the MiniForth was based on a PLD (now an FPGA) was so that it couldn't be pirated --- I originally suggested doing the motion-control board using an MC6812 and programming in C rather than Forth, but this idea was quickly killed --- a motion-control board based on the MC6812 would very soon get pirated, so all that you have accomplished is to do somebody else's R&D for them for free.

[toc] | [prev] | [next] | [standalone]

#44501

hughaguilar96@gmail.com wrote:
> On Tuesday, December 29, 2015 at 6:55:37 PM UTC-7, Waldek Hebisch wrote:
> > > (the atmega328). Arduino is expensive by today standards, but are very
> > > common. 
> > 
> > Well, Chinese Uno clones are at $2.65 with shiping
> > included.  I got 10 pieces of Chinese clone of Arduino
> > Pro Mini at $1.8 and recently I saw lower prices (IIRC $1.3).
> > The only things with comparably low price are Chinese
> > mini development boards for STM32F103R8 ($2.65).
> 
> This is most likely pirated --- I very much doubt that anybody in Beijing purchased the right to do this --- so, if you buy from China, you can't be too surprised when the Chinese pirate your own designs later on.

AFAICS board layouts are original.  And the original authors
agreed to reuse.  So this is not pirated.  I can not exclude
pirated CPU chips, but given insane pricing of chips the
Chinese maker could just get good deal when procuring chips.
And, as US DOD reports shows expensive brand name board
also may contain pirated chips...


-- 
                              Waldek Hebisch

[toc] | [prev] | [next] | [standalone]

#44347

On Wednesday, December 23, 2015 at 10:45:55 PM UTC-5, hughag...@gmail.com wrote:
> Arduinos are inexpensive? They seemed rather pricey to me --- you are mostly paying for their development system, so it doesn't make sense to pay that much money and then use a third-party development system such as Forth --- there are plenty of experimenter boards around such as the LaunchPad that are less expensive but don't provide much of a development system (the MicroStick-II actually has a pretty good development system, and it is not much more expensive).

Yes ... if it costs under $20, then I consider it inexpensive.

[toc] | [prev] | [next] | [standalone]

#44339

So I just ordered this one:

http://smile.amazon.com/gp/product/B012E9J65Y?psc=1&redirect=true&ref_=od_aui_detailpages00

$19.95 with free 2 day shipping

[toc] | [prev] | [next] | [standalone]

#44341

Chris Curl <ccurl609@gmail.com> writes:
> http://smile.amazon.com/gp/product/B012E9J65Y

That's the MSP432 board which uses a super low power ARM Cortex M4F.  It
might be seen as a successor to the Stellaris/Tiva board (which is still
available), but if you want to run Mecrisp on it, you might have to do
some porting work.  It's nice that the Amazon package includes the
breadboard and wires.

[toc] | [prev] | [next] | [standalone]

#44346

Flashforth for the arduino. Subroutine threaded with some inling. There is a hex file in the distro for the arduino uno and arduino mega.
I can also mail preprogrammed 328 chips.
Those are for the Arduin Uno.
http://flashforth.com

Merry Cristmas

Mike

[toc] | [prev] | [next] | [standalone]

#44367

I am returning the Raspberry Pi as I got this instead. It is 1/2 the price of the rPi.
http://www.amazon.com/gp/product/B014Q5EQOK?psc=1&redirect=true&ref_=oh_aui_detailpage_o00_s00

[toc] | [prev] | [next] | [standalone]

#44369

On 12/27/2015 12:32 AM, Chris Curl wrote:
> I am returning the Raspberry Pi as I got this instead. It is 1/2 the price of the rPi.
> http://www.amazon.com/gp/product/B014Q5EQOK?psc=1&redirect=true&ref_=oh_aui_detailpage_o00_s00

How could that at $30 be half the price of a $35 rPi?

-- 

Rick

[toc] | [prev] | [next] | [standalone]

#44370

Chris Curl <ccurl609@gmail.com> writes:
> I am returning the Raspberry Pi as I got this instead. It is 1/2 the
> price of the rPi.
> http://www.amazon.com/gp/product/B014Q5EQOK

The BBB is a real nice board but that offer is a bit suspicious:

1) $29 is a very low price for a 4gb Beaglebone Black (BBB).  They are
$50+ from regular distributors.  The older 2gb version (note this is the
amount of on-board flash; they all have 512MB ram) was $40+.

2) it claims the board is qualified for industrial temp range, but I
don't think Circuitco ever supported that.  There was some talk of an
alternate version but I don't know if it happened.  So I wonder if
you're getting some weird clone.  Have you already received it and tried
it out?  Did it come in Circuitco packaging?

Also: the RPI2 is $35 from the official distributor (Element14.com)
though you have to add shipping and an SD card.

The BBB is in many ways a better and more serious board than the RPi's,
but it's older than the current Pi's and is way behind the Pi2 in cpu
performance because of the Pi2's quad core cpu.  The Pi2 also has 2x as
much ram.  I don't know how much you care about this.

There is a new, faster Beagleboard in the works, but it will be much
fancier and more expensive than the BBB.

The BBB's main ARM processor doesn't seem that interesting as a Forth
host, since it's just another Linux board.  However, the chip has a pair
of realtime coprocessors (PRU's) that would be cool to program in Forth.
Right now the tool situation for them is pretty weak, from what I understand.

[toc] | [prev] | [next] | [standalone]

#44371

The rPi I got cost $69 at MicroCenter. It was this one:

http://www.microcenter.com/product/458478/Raspberry_Pi_2_Model_B_Media_Center

[toc] | [prev] | [next] | [standalone]

#44372

On 12/27/2015 2:43 AM, Chris Curl wrote:
> The rPi I got cost $69 at MicroCenter. It was this one:
>
> http://www.microcenter.com/product/458478/Raspberry_Pi_2_Model_B_Media_Center

That's because you paid for this...

Raspberry Pi 2 Model B Board, 8GB Operating System microSD Card, 
Multicomp Black Pi 2 Model B Case, Raspberry Pi Power Supply, Wi-Pi 
Wireless Adapter, 3' HDMi Cable, 7' Ethernet Cable, 36 Page Raspberry Pi 
User Guide

When you buy the BBB for $30 you will get... a BBB.

The rPi by itself can be bought for $35.

Still, I don't know how they are selling the BBB for $30.  I know a year 
ago or so they announced that they would be selling customized versions 
of the BBB.  If the quantity were high enough the price could be lower 
than the going rate.  But I don't really think this is where your BBB 
comes from.

-- 

Rick

[toc] | [prev] | [next] | [standalone]

#44373

Yeah ... I see that now. That was the only rPi thing they had in the 
store when i was there, and it was so close to xmas and I was short 
on time, so I grabbed it. I am returning it, and ordered this:

http://smile.amazon.com/gp/product/B00T2U7R7I?psc=1&redirect=true&ref_=oh_aui_detailpage_o00_s00

I guess it's a few dollars more than the $35 because of the 2-day 
"free" delivery for Amazon Prime. I'm OK with that.

But the thing I'm most interested in is the LaunchPad that came today.
It is the Texas Instruments MSP432P401R LaunchPad. It has 256KB 
flash memory, 64KB SRAM, and 32KB ROM. According to the mecrisp 
page, is it listed in the Mecrisp-Stellaris so I think I can 
get off the ground pretty quickly with this thing.

No, I will not be writing my own Forth for it. Well, maybe later ... :)

[toc] | [prev] | [next] | [standalone]

#44374

On 12/27/2015 12:01 PM, Chris Curl wrote:
> Yeah ... I see that now. That was the only rPi thing they had in the
> store when i was there, and it was so close to xmas and I was short
> on time, so I grabbed it. I am returning it, and ordered this:
>
> http://smile.amazon.com/gp/product/B00T2U7R7I?psc=1&redirect=true&ref_=oh_aui_detailpage_o00_s00
>
> I guess it's a few dollars more than the $35 because of the 2-day
> "free" delivery for Amazon Prime. I'm OK with that.
>
> But the thing I'm most interested in is the LaunchPad that came today.
> It is the Texas Instruments MSP432P401R LaunchPad. It has 256KB
> flash memory, 64KB SRAM, and 32KB ROM. According to the mecrisp
> page, is it listed in the Mecrisp-Stellaris so I think I can
> get off the ground pretty quickly with this thing.
>
> No, I will not be writing my own Forth for it. Well, maybe later ... :)

I don't get the part numbering on the MSP432.  MSP430 has always been 
their 16 bit proprietary processor.  So why pick the very confusing part 
number MSP432 for an ARM CM4 processor?  I guess with the low power 
emphasis of these parts they are focusing on target markets rather than 
construction.  Either way they would be a good platform for Mecrisp.  It 
should be very simple to port.  I think the only mods needed are for the 
serial port and memory size.

-- 

Rick

[toc] | [prev] | [next] | [standalone]

#44375

rickman <gnuarm@gmail.com> writes:
> So why pick the very confusing part number MSP432 for an ARM CM4
> processor?  I guess with the low power emphasis of these parts they
> are focusing on target markets rather than construction. 

Yes, I think the MSP430 has been losing ground to ARM parts for years,
so they used MSP432 to designate that they now have an ARM in the super
low power space.  But, I think the 430 still has some unique features
like FRAM, use less power at low speeds, seem simpler to program, and
are very cheap in the low end models.  

[toc] | [prev] | [next] | [standalone]

#44377

LOL ... So I was returning the rPi, and they had a rPi Zero there for $5. 
Now I have one of those. I also grabbed another Uno R3 because they were 
on sale for $4.99. The last thing I grabbed was a Mega 2560 r3 (256k 
Flash, 8k RAM, 4k EPROM) for $9.99.

Those rPi Zero things are ridiculous!

XMAS is over. Now to figure out how to use these things ...

[toc] | [prev] | [next] | [standalone]

#44378

Chris Curl <ccurl609@gmail.com> writes:
> LOL ... So I was returning the rPi, and they had a rPi Zero there for $5. 
> Now I have one of those. I also grabbed another Uno R3 because they were 
> on sale for $4.99. The last thing I grabbed was a Mega 2560 r3 (256k 
> Flash, 8k RAM, 4k EPROM) for $9.99.

The R3's must be Chinese clones.

> Those rPi Zero things are ridiculous!

Yeah, it's not real clear what to do with them though.  By the time you
add enough stuff to make them useful, you might as well have bought a
more complete but more expensive board.

> XMAS is over. Now to figure out how to use these things ...

Good luck!  Post your adventures.

[toc] | [prev] | [next] | [standalone]

#44381

On Sunday, December 27, 2015 at 7:14:52 PM UTC-5, Paul Rubin wrote:
> Chris Curl <cc...@gmail.com> writes:
> > LOL ... So I was returning the rPi, and they had a rPi Zero there for $5. 
> > Now I have one of those. I also grabbed another Uno R3 because they were 
> > on sale for $4.99. The last thing I grabbed was a Mega 2560 r3 (256k 
> > Flash, 8k RAM, 4k EPROM) for $9.99.
> 
> The R3's must be Chinese clones.
> 
> > Those rPi Zero things are ridiculous!
> 
> Yeah, it's not real clear what to do with them though.  By the time you
> add enough stuff to make them useful, you might as well have bought a
> more complete but more expensive board.
> 
> > XMAS is over. Now to figure out how to use these things ...
> 
> Good luck!  Post your adventures.
Yep. They are the MicroCenter house brand "Inland". Made in China. The 
quality seems pretty decent though.

One of my goals is to build a CNC router for my woodworking hobby.

[toc] | [prev] | [next] | [standalone]

#44379

On 12/27/2015 6:56 PM, Chris Curl wrote:
> LOL ... So I was returning the rPi, and they had a rPi Zero there for $5.
> Now I have one of those. I also grabbed another Uno R3 because they were
> on sale for $4.99. The last thing I grabbed was a Mega 2560 r3 (256k
> Flash, 8k RAM, 4k EPROM) for $9.99.
>
> Those rPi Zero things are ridiculous!

Yeah, I don't have a Micro Center convenient to me.  The shipping on the 
zero is more than the unit.  Once I have a real need for it I'll pick up 
a couple.  They are pretty amazing at that price.

At one time TI sold an MSP430 board for $4.30.  That's the only thing 
I've seen that was comparable in price and still useful.

-- 

Rick

[toc] | [prev] | [next] | [standalone]

Page 2 of 3 — ← Prev page 1 [2] 3  Next page →

Back to top | Article view | comp.lang.forth

csiph-web