• Re: 6 GHz stack machine

    From Wayne morellini@21:1/5 to All on Tue Jul 5 07:38:42 2022
    It's been over a year. How are things doing?

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jurgen Pitaske@21:1/5 to Stephen Pelc on Tue Jul 5 08:37:09 2022
    On Tuesday, 5 July 2022 at 16:22:25 UTC+1, Stephen Pelc wrote:
    On 5 Jul 2022 at 16:38:42 CEST, "Wayne morellini" <waynemo...@gmail.com> wrote:
    It's been over a year. How are things doing?
    Fine, thanks.

    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612, +34 649 662 974 http://www.mpeforth.com - free VFX Forth downloads

    Nice to hear that things are doing fine.

    I assume the question was,
    if there are any news regarding the 6GHz processor

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Stephen Pelc@21:1/5 to All on Tue Jul 5 15:22:23 2022
    On 5 Jul 2022 at 16:38:42 CEST, "Wayne morellini" <waynemorellini@gmail.com> wrote:

    It's been over a year. How are things doing?

    Fine, thanks.

    Stephen
    --
    Stephen Pelc, stephen@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612, +34 649 662 974 http://www.mpeforth.com - free VFX Forth downloads

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Wayne morellini@21:1/5 to jpit...@gmail.com on Tue Jul 5 10:38:56 2022
    On Wednesday, July 6, 2022 at 1:37:11 AM UTC+10, jpit...@gmail.com wrote:
    On Tuesday, 5 July 2022 at 16:22:25 UTC+1, Stephen Pelc wrote:
    On 5 Jul 2022 at 16:38:42 CEST, "Wayne morellini" <waynemo...@gmail.com> wrote:
    It's been over a year. How are things doing?
    Fine, thanks.

    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612, +34 649 662 974 http://www.mpeforth.com - free VFX Forth downloads
    Nice to hear that things are doing fine.

    I assume the question was,
    if there are any news regarding the 6GHz processor

    I think that's meant to be fine. But will I be alive by the time this comes out, chips can take so long? :)

    If it took me more than a month of work to draft the core of a misc Pico processor, with the right stable tools, I would be disappointed. This might be a thousand times more complex, requiring a lot more effort by a team of people.

    Who was that guy that was designing multiple chip Forth processor in England while Chick was designing the original Novix? Was that you Stephen?

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Wayne morellini@21:1/5 to jpit...@gmail.com on Tue Jul 5 10:55:58 2022
    On Wednesday, July 6, 2022 at 1:37:11 AM UTC+10, jpit...@gmail.com wrote:
    On Tuesday, 5 July 2022 at 16:22:25 UTC+1, Stephen Pelc wrote:
    On 5 Jul 2022 at 16:38:42 CEST, "Wayne morellini" <waynemo...@gmail.com> wrote:
    It's been over a year. How are things doing?
    Fine, thanks.

    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612, +34 649 662 974 http://www.mpeforth.com - free VFX Forth downloads
    Nice to hear that things are doing fine.

    I assume the question was,
    if there are any news regarding the 6GHz processor


    I want to design a 60ghz forth processor. I've been trying to figure out the principles of the mechanics of it. Trying to figure out a full speed memory system. But even with a doctorate in these things, it's a load of it's about what principles will
    even work. Confidence about lower speeds are much higher, but the principle is meant to be overclocked insane amounts. But even so. The transactional cost might be so low, you could perform many low energy circuits in software. The lowest energy
    circuits, obviously not.

    Anyway, there is some YouTube video about people inventing matter from protons. Well, somebody has done that noe, it was another things I was looking at an application for.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jurgen Pitaske@21:1/5 to Stephen Pelc on Sun Apr 30 10:16:51 2023
    On Tuesday, 5 July 2022 at 16:22:25 UTC+1, Stephen Pelc wrote:
    On 5 Jul 2022 at 16:38:42 CEST, "Wayne morellini" <waynemo...@gmail.com> wrote:
    It's been over a year. How are things doing?
    Fine, thanks.

    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612, +34 649 662 974 http://www.mpeforth.com - free VFX Forth downloads

    I just wonder if there is any news regarding this project
    or if people should rather not ask anyway?

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lorem Ipsum@21:1/5 to Jurgen Pitaske on Sun Apr 30 10:43:58 2023
    On Sunday, April 30, 2023 at 1:16:53 PM UTC-4, Jurgen Pitaske wrote:
    On Tuesday, 5 July 2022 at 16:22:25 UTC+1, Stephen Pelc wrote:
    On 5 Jul 2022 at 16:38:42 CEST, "Wayne morellini" <waynemo...@gmail.com> wrote:
    It's been over a year. How are things doing?
    Fine, thanks.

    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612, +34 649 662 974 http://www.mpeforth.com - free VFX Forth downloads
    I just wonder if there is any news regarding this project
    or if people should rather not ask anyway?

    LOL By "people", you mean *other* people? Why is asking a question about a technical project inappropriate?

    --

    Rick C.

    --++ Get 1,000 miles of free Supercharging
    --++ Tesla referral code - https://ts.la/richard11209

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Stephen Pelc@21:1/5 to Jurgen Pitaske on Sun Apr 30 21:12:40 2023
    On 30 Apr 2023 at 19:16:51 CEST, "Jurgen Pitaske" <jpitaske@gmail.com> wrote:

    I just wonder if there is any news regarding this project
    or if people should rather not ask anyway?

    When
    a) I have nore news for you, an
    b) I am allowed to tell you
    I will release the news.

    Stephen
    --
    Stephen Pelc, stephen@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612,
    +34 649 662 974
    http://www.mpeforth.com - free VFX Forth downloads

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jurgen Pitaske@21:1/5 to Lorem Ipsum on Sun Apr 30 13:50:04 2023
    On Sunday, 30 April 2023 at 18:44:00 UTC+1, Lorem Ipsum wrote:
    On Sunday, April 30, 2023 at 1:16:53 PM UTC-4, Jurgen Pitaske wrote:
    On Tuesday, 5 July 2022 at 16:22:25 UTC+1, Stephen Pelc wrote:
    On 5 Jul 2022 at 16:38:42 CEST, "Wayne morellini" <waynemo...@gmail.com> wrote:
    It's been over a year. How are things doing?
    Fine, thanks.

    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612, +34 649 662 974 http://www.mpeforth.com - free VFX Forth downloads
    I just wonder if there is any news regarding this project
    or if people should rather not ask anyway?
    LOL By "people", you mean *other* people? Why is asking a question about a technical project inappropriate?

    --

    Rick C.

    --++ Get 1,000 miles of free Supercharging
    --++ Tesla referral code - https://ts.la/richard11209


    YOUR BULLSHIT DAY OR YOU ARE JUST DRUNK?
    Or are you impersonating Stephen Pelc who is most probably the only one who could answer this?

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jurgen Pitaske@21:1/5 to Stephen Pelc on Sun Apr 30 23:04:41 2023
    On Sunday, 30 April 2023 at 22:12:43 UTC+1, Stephen Pelc wrote:
    On 30 Apr 2023 at 19:16:51 CEST, "Jurgen Pitaske" <jpit...@gmail.com> wrote:

    I just wonder if there is any news regarding this project
    or if people should rather not ask anyway?
    When
    a) I have nore news for you, an
    b) I am allowed to tell you
    I will release the news.
    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612,
    +34 649 662 974
    http://www.mpeforth.com - free VFX Forth downloads

    Great, Thank You

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jurgen Pitaske@21:1/5 to Jurgen Pitaske on Mon May 1 00:41:35 2023
    On Monday, 1 May 2023 at 07:04:43 UTC+1, Jurgen Pitaske wrote:
    On Sunday, 30 April 2023 at 22:12:43 UTC+1, Stephen Pelc wrote:
    On 30 Apr 2023 at 19:16:51 CEST, "Jurgen Pitaske" <jpit...@gmail.com> wrote:

    I just wonder if there is any news regarding this project
    or if people should rather not ask anyway?
    When
    a) I have nore news for you, an
    b) I am allowed to tell you
    I will release the news.
    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612,
    +34 649 662 974
    http://www.mpeforth.com - free VFX Forth downloads
    Great, Thank You


    and please have a look at
    www.mpeforth.com - not working here
    https://vfxforth.com/ does

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jurgen Pitaske@21:1/5 to All on Tue May 2 06:28:11 2023
    Great, Thank You

    and please have a look at
    www.mpeforth.com - not working here
    https://vfxforth.com/ does

    and now both websites are alive again today.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jurgen Pitaske@21:1/5 to Stephen Pelc on Fri Oct 20 01:29:28 2023
    On Sunday, 30 April 2023 at 22:12:43 UTC+1, Stephen Pelc wrote:
    On 30 Apr 2023 at 19:16:51 CEST, "Jurgen Pitaske" <jpit...@gmail.com> wrote:

    I just wonder if there is any news regarding this project
    or if people should rather not ask anyway?
    When
    a) I have nore news for you, an
    b) I am allowed to tell you
    I will release the news.
    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)23 8063 1441, +44 (0)78 0390 3612,
    +34 649 662 974
    http://www.mpeforth.com - free VFX Forth downloads

    I wonder, if you can give us an update about the new chip now?

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From S@21:1/5 to Stephen Pelc on Fri Oct 20 08:11:53 2023
    On Fri, 20 Oct 2023, 18:29 Jurgen Pitaske, <jpitaske@gmail.com> wrote:
    On Sunday, 30 April 2023 at 22:12:43 UTC+1, Stephen Pelc wrote:
    On 30 Apr 2023 at 19:16:51 CEST, "Jurgen Pitaske" <jpit...@gmail.com> wrote:
    ..

    I wonder, if you can give us an update about the new chip now?

    Oh Please! I could have done with a 6Ghz Forth chip 20 years ago, when it was possible. Even if it's based on Ting's old design, with 32+ processors. Especially if it could run Jeff's software and JavaScript (they also have figured out how to run linux
    on JavaScript now).

    Anyway, the truth is controversial. Back in the tu.e that Intel were trying to push towards 4 and 5Ghz chips, it would have been possible to get higher speeds on a Forth chip. Rather than aim at FPGA, they could have made an Forth MCU with DSP.
    Basically NASA would have lined up, the military, and a whole bunch of high end applications. You could even charge $10,000 for such a chip, if done right. The rest of us could make do with a stripped down 2ghz version with 1/10 of the processing units,
    for $100, paid for out of profit on the top end part. Something even Intel or IBM, could invest in. Now, we need the lower end part for $1, and higher end parts for $10-$100. Amazing what teo decades of development will do.

    The truth, is it possible for Green Arrays to make such a chip at 20-50Ghz with internal localised memory on each node treated as a contiguous address space, to feed the nodes fast enough. That would definitely generate a lot of interest. The fact it
    may consume 1KW, is not so relevant to these people (I guess it might still be below 100 watts). There are some very high speed applications requiring fast software radio processing. Definitely would save the us chip industry. We are definitely
    talking about the CPU side being more like an Ting or shboom instruction set, so they can load and ececute their C code more easily, or my own suggestions. Even a simple version of chip like this can use a node per hardware function in software. An IO,
    wireless, media, processor MCU is four nodes. Even one. You are talking about approaching companies let Intel, and saying "We can push your upcoming process node near the maximum" simply because the more compact simpler processor can complete things
    more simply with less heat, and be cooled better. Basically, you are talking about running it 2-5x faster than an overclocked i series chip at less energy.

    The whole mini transputer thing was a bit of a dead end, but DSP application demand still goes on.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Jurgen Pitaske@21:1/5 to All on Fri Oct 20 08:24:15 2023
    On Friday, 20 October 2023 at 16:11:56 UTC+1, S wrote:
    On Fri, 20 Oct 2023, 18:29 Jurgen Pitaske, <jpit...@gmail.com> wrote:
    On Sunday, 30 April 2023 at 22:12:43 UTC+1, Stephen Pelc wrote:
    On 30 Apr 2023 at 19:16:51 CEST, "Jurgen Pitaske" <jpit...@gmail.com> wrote:
    ..
    I wonder, if you can give us an update about the new chip now?
    Oh Please! I could have done with a 6Ghz Forth chip 20 years ago, when it was possible. Even if it's based on Ting's old design, with 32+ processors. Especially if it could run Jeff's software and JavaScript (they also have figured out how to run linux
    on JavaScript now).

    Anyway, the truth is controversial. Back in the tu.e that Intel were trying to push towards 4 and 5Ghz chips, it would have been possible to get higher speeds on a Forth chip. Rather than aim at FPGA, they could have made an Forth MCU with DSP.
    Basically NASA would have lined up, the military, and a whole bunch of high end applications. You could even charge $10,000 for such a chip, if done right. The rest of us could make do with a stripped down 2ghz version with 1/10 of the processing units,
    for $100, paid for out of profit on the top end part. Something even Intel or IBM, could invest in. Now, we need the lower end part for $1, and higher end parts for $10-$100. Amazing what teo decades of development will do.

    The truth, is it possible for Green Arrays to make such a chip at 20-50Ghz with internal localised memory on each node treated as a contiguous address space, to feed the nodes fast enough. That would definitely generate a lot of interest. The fact it
    may consume 1KW, is not so relevant to these people (I guess it might still be below 100 watts). There are some very high speed applications requiring fast software radio processing. Definitely would save the us chip industry. We are definitely talking
    about the CPU side being more like an Ting or shboom instruction set, so they can load and ececute their C code more easily, or my own suggestions. Even a simple version of chip like this can use a node per hardware function in software. An IO, wireless,
    media, processor MCU is four nodes. Even one. You are talking about approaching companies let Intel, and saying "We can push your upcoming process node near the maximum" simply because the more compact simpler processor can complete things more simply
    with less heat, and be cooled better. Basically, you are talking about running it 2-5x faster than an overclocked i series chip at less energy.

    The whole mini transputer thing was a bit of a dead end, but DSP application demand still goes on.

    When I asked Stepen Pelc for an update - if possible - I did not expect such a useless post as you have placed it.

    20 years ago you would not have been able to get such a chip - you would have had to save your pocket money for quite some years or tens of years .

    Ting got close with having his EP32 made as ASIC.

    Greenarrays has not done any new designs for about 10 years now.

    And it seems you have no clue what the Transputer was about.

    I wonder what the S stands for - is it short for Silly comments ???

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From S@21:1/5 to Jurgen Pitaske on Sun Oct 22 06:46:01 2023
    On Saturday, October 21, 2023 at 1:24:17 AM UTC+10, Jurgen Pitaske wrote:
    On Friday, 20 October 2023 at 16:11:56 UTC+1, S wrote:
    On Fri, 20 Oct 2023, 18:29 Jurgen Pitaske, <jpit...@gmail.com> wrote:
    On Sunday, 30 April 2023 at 22:12:43 UTC+1, Stephen Pelc wrote:
    On 30 Apr 2023 at 19:16:51 CEST, "Jurgen Pitaske" <jpit...@gmail.com> wrote:
    ..
    I wonder, if you can give us an update about the new chip now?
    Oh Please! I could have done with a 6Ghz Forth chip 20 years ago, when it was possible. Even if it's based on Ting's old design, with 32+ processors. Especially if it could run Jeff's software and JavaScript (they also have figured out how to run
    linux on JavaScript now).

    Anyway, the truth is controversial. Back in the tu.e that Intel were trying to push towards 4 and 5Ghz chips, it would have been possible to get higher speeds on a Forth chip. Rather than aim at FPGA, they could have made an Forth MCU with DSP.
    Basically NASA would have lined up, the military, and a whole bunch of high end applications. You could even charge $10,000 for such a chip, if done right. The rest of us could make do with a stripped down 2ghz version with 1/10 of the processing units,
    for $100, paid for out of profit on the top end part. Something even Intel or IBM, could invest in. Now, we need the lower end part for $1, and higher end parts for $10-$100. Amazing what teo decades of development will do.

    The truth, is it possible for Green Arrays to make such a chip at 20-50Ghz with internal localised memory on each node treated as a contiguous address space, to feed the nodes fast enough. That would definitely generate a lot of interest. The fact it
    may consume 1KW, is not so relevant to these people (I guess it might still be below 100 watts). There are some very high speed applications requiring fast software radio processing. Definitely would save the us chip industry. We are definitely talking
    about the CPU side being more like an Ting or shboom instruction set, so they can load and ececute their C code more easily, or my own suggestions. Even a simple version of chip like this can use a node per hardware function in software. An IO, wireless,
    media, processor MCU is four nodes. Even one. You are talking about approaching companies let Intel, and saying "We can push your upcoming process node near the maximum" simply because the more compact simpler processor can complete things more simply
    with less heat, and be cooled better. Basically, you are talking about running it 2-5x faster than an overclocked i series chip at less energy.

    The whole mini transputer thing was a bit of a dead end, but DSP application demand still goes on.
    When I asked Stepen Pelc for an update - if possible - I did not expect such a useless post as you have placed it.

    20 years ago you would not have been able to get such a chip - you would have had to save your pocket money for quite some years or tens of years .

    That's just crazy miscomprehension. It's about designing a miniature overclocked CPU, using something like the process node that Pentium 4 used. As it is simplified with less transistors, it produces less heat, allowing more overclocking. You should
    know this, rather than throw false allegations around!
    ..
    Ting got close with having his EP32 made as ASIC

    My point. He was in Europe shopping it, the chip Stephen is testing, is in Europe, and just enough time for such a design to be at the stage the 6 ghz one is . So, maybe it's related.

    Greenarrays has not done any new designs for about 10 years now.

    It completely does not matter, and is not equivalent to what can be done.


    And it seems you have no clue what the Transputer was about.

    Non Equivalent again Saying that dumping the mass parallel low energy transputer matrix for
    a reduced core high demand DSP product makes more sense.

    I wonder what the S stands for - is it short for Silly comments ???

    Well, the second letter is not U, that's your reply

    Next time, be morre serious, please.

    It is often intellectually petty useless people who make such replies around here. Wake up, you misread much of it. You were being silly and petty, I was being serious and realistic.

    I wish you a good day sir.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Anton Ertl@21:1/5 to waynemorellini@gmail.com on Sun Oct 22 14:17:04 2023
    S <waynemorellini@gmail.com> writes:
    On Saturday, October 21, 2023 at 1:24:17=E2=80=AFAM UTC+10, Jurgen Pitaske = >wrote:
    20 years ago you would not have been able to get such a chip - you would = >have had to save your pocket money for quite some years or tens of years .= >=20

    That's just crazy miscomprehension. It's about designing a miniature overc= >locked CPU, using something like the process node that Pentium 4 used. As = >it is simplified with less transistors, it produces less heat, allowing mor= >e overclocking.

    Actually, Intel released the Pentium 4 HT 3.0 on April 14, 2003 (3.06
    on November 14, 2002, 3.2 on June 23, 2003). These are all
    Northwood-based Pentium 4s, i.e., they have ALUs running at double
    clock speed, i.e., 6GHz for the 3.0 and 6.4GHz for the 3.2. The
    Northwood (and before it, the Willamette) actually were able to
    perform two dependent adds per clock cycle (I have measured this on a Northwood).

    So, in 20 years old technology it is possible to create a 6GHz stack
    machine. Whether it is any good in application performance is another
    question (the Pentium 4 was somewhat disappointing).

    - anton
    --
    M. Anton Ertl http://www.complang.tuwien.ac.at/anton/home.html
    comp.lang.forth FAQs: http://www.complang.tuwien.ac.at/forth/faq/toc.html
    New standard: https://forth-standard.org/
    EuroForth 2023: https://euro.theforth.net/2023

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lorem Ipsum@21:1/5 to All on Sun Oct 22 07:17:37 2023
    On Sunday, October 22, 2023 at 9:46:04 AM UTC-4, S wrote:
    On Saturday, October 21, 2023 at 1:24:17 AM UTC+10, Jurgen Pitaske wrote:
    On Friday, 20 October 2023 at 16:11:56 UTC+1, S wrote:
    On Fri, 20 Oct 2023, 18:29 Jurgen Pitaske, <jpit...@gmail.com> wrote:
    On Sunday, 30 April 2023 at 22:12:43 UTC+1, Stephen Pelc wrote:
    On 30 Apr 2023 at 19:16:51 CEST, "Jurgen Pitaske" <jpit...@gmail.com> wrote:
    ..
    I wonder, if you can give us an update about the new chip now?
    Oh Please! I could have done with a 6Ghz Forth chip 20 years ago, when it was possible. Even if it's based on Ting's old design, with 32+ processors. Especially if it could run Jeff's software and JavaScript (they also have figured out how to run
    linux on JavaScript now).

    Anyway, the truth is controversial. Back in the tu.e that Intel were trying to push towards 4 and 5Ghz chips, it would have been possible to get higher speeds on a Forth chip. Rather than aim at FPGA, they could have made an Forth MCU with DSP.
    Basically NASA would have lined up, the military, and a whole bunch of high end applications. You could even charge $10,000 for such a chip, if done right. The rest of us could make do with a stripped down 2ghz version with 1/10 of the processing units,
    for $100, paid for out of profit on the top end part. Something even Intel or IBM, could invest in. Now, we need the lower end part for $1, and higher end parts for $10-$100. Amazing what teo decades of development will do.

    The truth, is it possible for Green Arrays to make such a chip at 20-50Ghz with internal localised memory on each node treated as a contiguous address space, to feed the nodes fast enough. That would definitely generate a lot of interest. The fact
    it may consume 1KW, is not so relevant to these people (I guess it might still be below 100 watts). There are some very high speed applications requiring fast software radio processing. Definitely would save the us chip industry. We are definitely
    talking about the CPU side being more like an Ting or shboom instruction set, so they can load and ececute their C code more easily, or my own suggestions. Even a simple version of chip like this can use a node per hardware function in software. An IO,
    wireless, media, processor MCU is four nodes. Even one. You are talking about approaching companies let Intel, and saying "We can push your upcoming process node near the maximum" simply because the more compact simpler processor can complete things more
    simply with less heat, and be cooled better. Basically, you are talking about running it 2-5x faster than an overclocked i series chip at less energy.

    The whole mini transputer thing was a bit of a dead end, but DSP application demand still goes on.
    When I asked Stepen Pelc for an update - if possible - I did not expect such a useless post as you have placed it.

    20 years ago you would not have been able to get such a chip - you would have had to save your pocket money for quite some years or tens of years .
    That's just crazy miscomprehension. It's about designing a miniature overclocked CPU, using something like the process node that Pentium 4 used. As it is simplified with less transistors, it produces less heat, allowing more overclocking. You should
    know this, rather than throw false allegations around!

    Do you know what overclocking is??? It doesn't mean how you are using it.

    --

    Rick C.

    -+-- Get 1,000 miles of free Supercharging
    -+-- Tesla referral code - https://ts.la/richard11209

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Lorem Ipsum@21:1/5 to Anton Ertl on Sun Oct 22 10:22:01 2023
    On Sunday, October 22, 2023 at 10:32:54 AM UTC-4, Anton Ertl wrote:
    S <waynemo...@gmail.com> writes:
    On Saturday, October 21, 2023 at 1:24:17=E2=80=AFAM UTC+10, Jurgen Pitaske =
    wrote:
    20 years ago you would not have been able to get such a chip - you would =
    have had to save your pocket money for quite some years or tens of years .= >=20

    That's just crazy miscomprehension. It's about designing a miniature overc= >locked CPU, using something like the process node that Pentium 4 used. As = >it is simplified with less transistors, it produces less heat, allowing mor=
    e overclocking.

    Actually, Intel released the Pentium 4 HT 3.0 on April 14, 2003 (3.06
    on November 14, 2002, 3.2 on June 23, 2003). These are all
    Northwood-based Pentium 4s, i.e., they have ALUs running at double
    clock speed, i.e., 6GHz for the 3.0 and 6.4GHz for the 3.2. The
    Northwood (and before it, the Willamette) actually were able to
    perform two dependent adds per clock cycle (I have measured this on a Northwood).

    So, in 20 years old technology it is possible to create a 6GHz stack machine. Whether it is any good in application performance is another question (the Pentium 4 was somewhat disappointing).

    The Pentium 4 was lacking in speed compared to the clock speed, because they made the high clock speeds possible by increasing the pipeline length, and so faster pipeline stages. The downside of this approach is the longer time to recover from pipeline
    stalls. With a couple of iterations, they were able to further increase the clock speeds. In all versions, the Pentium 4 was faster then the Pentium 3s, just not always by much.

    The Pentium 3s were easy to overclock, sometimes by 2x. Pentium 4s hardly at all.

    --

    Rick C.

    -+-+ Get 1,000 miles of free Supercharging
    -+-+ Tesla referral code - https://ts.la/richard11209

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From S@21:1/5 to Lorem Ipsum on Tue Oct 24 01:22:40 2023
    On Monday, October 23, 2023 at 12:17:39 AM UTC+10, Lorem Ipsum wrote:

    Do you know what overclocking is??? It doesn't mean how you are using it.

    --

    Rick C.

    -+-- Get 1,000 miles of free Supercharging
    -+-- Tesla referral code - https://ts.la/richard11209

    What an incredible idiotic Gas light! Be a bit more sophisticated if you are going say things they are only going fool the naive with very little technical knowledge.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From S@21:1/5 to Anton Ertl on Tue Oct 24 02:09:24 2023
    On Monday, October 23, 2023 at 12:32:54 AM UTC+10, Anton Ertl wrote:
    S <waynemo...@gmail.com> writes:
    On Saturday, October 21, 2023 at 1:24:17=E2=80=AFAM UTC+10, Jurgen Pitaske =
    wrote:
    20 years ago you would not have been able to get such a chip - you would =
    have had to save your pocket money for quite some years or tens of years .= >=20

    That's just crazy miscomprehension. It's about designing a miniature overc= >locked CPU, using something like the process node that Pentium 4 used. As = >it is simplified with less transistors, it produces less heat, allowing mor=
    e overclocking.

    Actually, Intel released the Pentium 4 HT 3.0 on April 14, 2003 (3.06
    on November 14, 2002, 3.2 on June 23, 2003). These are all
    Northwood-based Pentium 4s, i.e., they have ALUs running at double
    clock speed, i.e., 6GHz for the 3.0 and 6.4GHz for the 3.2. The
    Northwood (and before it, the Willamette) actually were able to
    perform two dependent adds per clock cycle (I have measured this on a Northwood).

    So, in 20 years old technology it is possible to create a 6GHz stack machine. Whether it is any good in application performance is another question (the Pentium 4 was somewhat disappointing).

    Thank you Anton. The performance issue, is going to depend on how it's coded and what it is used for. The simplified design allows higher clocking and negates pipelining somewhat. It's just a numbers game to high end buyers. Even if they had to put
    the thing in a Cryo cooler in missile, they wouldn't care, as long as they can get the increased response time (just a hypothetical). I'm purely saying the ultra lowest energy did not pan out. But, a three pronged approach would have been better. The
    Seaforth ultra low energy cores, a higher performance low energy version, which acted more like a regular microcontroller, and a high performance chip that starts out low and clocks up and resources out for very high end work, where the budgets are very
    high. No normal processor contender is going to offer only one chip for a very low profit market (though they did get to two or so chips). If you can sell into three market segments, that is more sahee potential. The catch was, we should have aimed
    at the more high end market with a needed response version, than thinking about it in terms of avoiding the flooded MCU/processor market. That pays for a low end version that can go into those markets, and servers. You then offer a clear advantage
    lower end MCU, and offer the clear lowest energy version, and it doesn't matter if they don't get contracts and sit on the hard drive, as your costs are largely covered by shared development, business admin, costs and profits, of the highest.end parts.
    I just saw today that NVIDIA is going to mop up with an high end Arm product. 20 years ago, the bar was a lot lower, and billions could have been saved now with a low clocked version of the high end part by now. We can't really do the same now, there
    might be some low cost low energy server farm applications, but Arm and RiscV are already onto that. If it was me, I would be saying 15 years ago, who do we sell put to, inorder to keep the business running with investment and distribution of better
    products. You could have sold to google back then, with plans for a cheaper server implementation. That's a major cost for them. Pragmatism. Anyway, recovering from being very sick all day again, so better get a move on

    Thanks again, Anton.


    - anton
    --
    M. Anton Ertl h ttp://www.complang.tuwien.ac.at/anton/home.html comp.lang.forth FAQs: http://www.complang.tuwien.ac.at/forth/faq/toc.html New standard: https://forth-standard.org/
    EuroForth 2023: https://euro.theforth.net/2023

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Stephen Pelc@21:1/5 to chat.to.dot@gmail.com on Mon Jan 22 21:26:15 2024
    On 20 Jan 2024 at 01:53:40 CET, "•" <chat.to.dot@gmail.com> wrote:

    On Friday 2 July 2021 at 9:49:54 pm UTC+10, Stephen Pelc wrote:
    An MPE client is currently designing a new dual-stack machine. The
    predicted performance is 6 GHz (instructions per second). 40 CPUs
    occupy less than 1 sqare mm.

    Do you post updates somewhere else? I'm interested in energy envelope range, and if it will be available in quantity for products, or single units for maler projects? I also suggest: a Royal Break Out card. Flexible thin board that exposes lines to holes for maker crowd, with minimal direct memory + flash, usb3 C 10mb/s+ with video over HDMI x2, and exposure of IO pins. That would put it ahead of most SBC products in Arduino/pi spaces and easily surface mount for everybody else, and able to have linux developed for it eventually using the C compiler, even if in a virtual interpreted machine for security. It would then be suitable for small Chinese Linux nucs with internet
    media player functions..

    It's a real chip (now) with a specific first application. Once the first production
    application devices are running I expect that the developers will look at
    other
    applications and dev tools.

    Until then, you just have to wait.

    Stephen
    --
    Stephen Pelc, stephen@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)78 0390 3612, +34 649 662 974
    http://www.mpeforth.com - free VFX Forth downloads

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Stephen Pelc@21:1/5 to Jurgen Pitaske on Tue Jan 23 15:21:11 2024
    On 23 Jan 2024 at 08:23:16 CET, "Jurgen Pitaske" <jpitaske@gmail.com> wrote:

    On Monday 22 January 2024 at 21:26:18 UTC, Stephen Pelc wrote:
    On 20 Jan 2024 at 01:53:40 CET, "•" <chat....@gmail.com> wrote:

    On Friday 2 July 2021 at 9:49:54 pm UTC+10, Stephen Pelc wrote:
    An MPE client is currently designing a new dual-stack machine. The
    predicted performance is 6 GHz (instructions per second). 40 CPUs
    occupy less than 1 sqare mm.

    Do you post updates somewhere else? I'm interested in energy envelope range,
    and if it will be available in quantity for products, or single units for >>> maler projects? I also suggest: a Royal Break Out card. Flexible thin board >>> that exposes lines to holes for maker crowd, with minimal direct memory + >>> flash, usb3 C 10mb/s+ with video over HDMI x2, and exposure of IO pins. That
    would put it ahead of most SBC products in Arduino/pi spaces and easily
    surface mount for everybody else, and able to have linux developed for it >>> eventually using the C compiler, even if in a virtual interpreted machine for
    security. It would then be suitable for small Chinese Linux nucs with internet
    media player functions..
    It's a real chip (now) with a specific first application. Once the first
    production
    application devices are running I expect that the developers will look at
    other
    applications and dev tools.

    Until then, you just have to wait.
    Stephen
    --
    Stephen Pelc, ste...@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)78 0390 3612, +34 649 662 974
    http://www.mpeforth.com - free VFX Forth downloads

    One point is still unclear to me:
    Is this a chip just for internal projects,
    or will this chip be as well available for the outside world to buy?
    Either as a chip or as a development board?

    ... You just have to wait. ...

    Stephen
    --
    Stephen Pelc, stephen@vfxforth.com
    MicroProcessor Engineering, Ltd. - More Real, Less Time
    133 Hill Lane, Southampton SO15 5AF, England
    tel: +44 (0)78 0390 3612, +34 649 662 974
    http://www.mpeforth.com - free VFX Forth downloads

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)