http://www.grancelang.org
http://www.virtualworldframework.org http://www.julialang.org http://www.insteon.net http://www.armdevices.net http://www.riscoscode.com/Pages/Item0003.html The meeting was attended by: Bert, Tim, Casmiro, Chris M., Chris C., Colin, Yoshiki, Craig, and Colin. Colin wondered about reifying interfaces and a tool for revealing protocols as they are actually used. Bert went to OOPSLA in Tuscon, Ariz. where there was a lot of talk about JavaScript, Potsdam Squeakers had two presentations, Dave Smith presented his Virtual World Framework, and Rob Pike did a presentation on Go. Andrew Black is working a new teaching language called Grace. Colin mentioned a new language called Julia. Chris M. asked what the holy grail of programming was and why Croquet wasn't a wider success. OpenCobalt sounds stalled as Matthew Fulmer graduated and got a job. One of the topics discussed around Croquet was market position: what is this tool for? It was discussed that a simple, clear tool with a specific niche, an obvious problem could out-market a general tool, a ball of potential. Colin said that DabbleDB had some of that problem. It was great technology, but didn't connect as soundly with given domains (i.e. planning a wedding) Tim talked about how he's networked his house: his lights have switches similar (if not actually from) Insteon. Discussion of the Raspberry Pi computer and its low price point reemerged. Tim pointed out that the RiscOS now runs on the R. Pi computer. Chris |
On Fri, Nov 02, 2012 at 03:15:34PM -0400, Chris Cunnington wrote:
> http://www.grancelang.org > http://www.virtualworldframework.org > http://www.julialang.org > http://www.insteon.net > http://www.armdevices.net > http://www.riscoscode.com/Pages/Item0003.html > > The meeting was attended by: Bert, Tim, Casmiro, Chris M., Chris C., > Colin, Yoshiki, Craig, and Colin. > > Colin wondered about reifying interfaces and a tool for revealing > protocols as they are actually used. > Bert went to OOPSLA in Tuscon, Ariz. where there was a lot of talk about > JavaScript, Potsdam Squeakers had two presentations, Dave Smith > presented his Virtual World Framework, and Rob Pike did a presentation > on Go. > Andrew Black is working a new teaching language called Grace. > Colin mentioned a new language called Julia. > Chris M. asked what the holy grail of programming was and why Croquet > wasn't a wider success. > OpenCobalt sounds stalled as Matthew Fulmer graduated and got a job. > One of the topics discussed around Croquet was market position: what is > this tool for? It was discussed that a simple, clear tool with a > specific niche, an obvious problem could out-market a general tool, a > ball of potential. > Colin said that DabbleDB had some of that problem. It was great > technology, but didn't connect as soundly with given domains (i.e. > planning a wedding) > Tim talked about how he's networked his house: his lights have switches > similar (if not actually from) Insteon. > Discussion of the Raspberry Pi computer and its low price point > reemerged. Tim pointed out that the RiscOS now runs on the R. Pi computer. RiscOS on the Raspberry Pi?!? Woohoo! Clearly this calls for an update to the RiscOS VM on squeakvm.org, otherwise people will not be able to have proper block closure support on their Pi. Tim, I'll promise to buy you a brand new Raspberry Pi if you compile a fresh RiscOS VM for it ;) > > Chris > > |
Hi folks;
On 02-11-2012, at 4:09 PM, David T. Lewis <[hidden email]> wrote: >> >> Tim talked about how he's networked his house: his lights have switches >> similar (if not actually from) Insteon. Actual Insteon. See www.insteon.net & www.smarthome.com Cool new product - a network enabled LED lightbulb. Details on the house at www.rowledge.org/tim >> Discussion of the Raspberry Pi computer and its low price point >> reemerged. Tim pointed out that the RiscOS now runs on the R. Pi computer. > > RiscOS on the Raspberry Pi?!? Woohoo! Clearly this calls for an update to > the RiscOS VM on squeakvm.org, otherwise people will not be able to have > proper block closure support on their Pi. Tim, I'll promise to buy you a > brand new Raspberry Pi if you compile a fresh RiscOS VM for it ;) > I might be able to do something to resurrect RISC OS Squeak. It's interesting to see how people are trying it on RPi and liking the responsiveness over linux etc. Of course, it's still not a 'proper modern OS' since it lack memory protection and pre-emptive process scheduling, but in practice it works really quite well. Some of the other ARM boards currently available are even faster. I'd need a pi, the RISC OS SD card, the NutPi package for the relevant dev tools and some idea of a display to use with it. I'll investigate. Who knows…. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Liability: a valuable political skill |
On 2012-11-06 12:42 PM, tim Rowledge wrote:
> Hi folks; > > On 02-11-2012, at 4:09 PM, David T. Lewis <[hidden email]> wrote: >>> Tim talked about how he's networked his house: his lights have switches >>> similar (if not actually from) Insteon. > Actual Insteon. See www.insteon.net & www.smarthome.com Cool new product - a network enabled LED lightbulb. Details on the house at www.rowledge.org/tim > >>> Discussion of the Raspberry Pi computer and its low price point >>> reemerged. Tim pointed out that the RiscOS now runs on the R. Pi computer. >> RiscOS on the Raspberry Pi?!? Woohoo! Clearly this calls for an update to >> the RiscOS VM on squeakvm.org, otherwise people will not be able to have >> proper block closure support on their Pi. Tim, I'll promise to buy you a >> brand new Raspberry Pi if you compile a fresh RiscOS VM for it ;) >> > > I might be able to do something to resurrect RISC OS Squeak. It's interesting to see how people are trying it on RPi and liking the responsiveness over linux etc. Of course, it's still not a 'proper modern OS' since it lack memory protection and pre-emptive process scheduling, but in practice it works really quite well. Some of the other ARM boards currently available are even faster. > > I'd need a pi, the RISC OS SD card, the NutPi package for the relevant dev tools and some idea of a display to use with it. I'll investigate. Who knows…. > > > > tim > -- > tim Rowledge; [hidden email]; http://www.rowledge.org/tim > Liability: a valuable political skill > > > David) a RPi or related tool if you needed it. I think this would be a fun way to poke around how the vm works. Chris |
On 2012-11-06, at 19:02, Chris Cunnington <[hidden email]> wrote: > On 2012-11-06 12:42 PM, tim Rowledge wrote: >> Hi folks; >> >> On 02-11-2012, at 4:09 PM, David T. Lewis <[hidden email]> wrote: >>>> Tim talked about how he's networked his house: his lights have switches >>>> similar (if not actually from) Insteon. >> Actual Insteon. See www.insteon.net & www.smarthome.com Cool new product - a network enabled LED lightbulb. Details on the house at www.rowledge.org/tim >> >>>> Discussion of the Raspberry Pi computer and its low price point >>>> reemerged. Tim pointed out that the RiscOS now runs on the R. Pi computer. >>> RiscOS on the Raspberry Pi?!? Woohoo! Clearly this calls for an update to >>> the RiscOS VM on squeakvm.org, otherwise people will not be able to have >>> proper block closure support on their Pi. Tim, I'll promise to buy you a >>> brand new Raspberry Pi if you compile a fresh RiscOS VM for it ;) >>> >> >> I might be able to do something to resurrect RISC OS Squeak. It's interesting to see how people are trying it on RPi and liking the responsiveness over linux etc. Of course, it's still not a 'proper modern OS' since it lack memory protection and pre-emptive process scheduling, but in practice it works really quite well. Some of the other ARM boards currently available are even faster. >> >> I'd need a pi, the RISC OS SD card, the NutPi package for the relevant dev tools and some idea of a display to use with it. I'll investigate. Who knows…. >> >> >> >> tim >> -- >> tim Rowledge; [hidden email]; http://www.rowledge.org/tim >> Liability: a valuable political skill >> >> > I'm interested in this too. I also would be willing to provide you (like David) a RPi or related tool if you needed it. I think this would be a fun way to poke around how the vm works. > > Chris I got a Raspberry Pi sent to Tim. He still needs an SD card with RISC OS, the Nut Pi tools card, and a paying job. Nut Pi tools (£35 plus P&P): https://www.riscosopen.org/content/sales/nutpi RISC OS (free, just need an SD card): http://www.raspberrypi.org/downloads - Bert - |
On 2012-11-07 12:14 PM, Bert Freudenberg wrote:
> On 2012-11-06, at 19:02, Chris Cunnington <[hidden email]> wrote: > >> On 2012-11-06 12:42 PM, tim Rowledge wrote: >>> Hi folks; >>> >>> On 02-11-2012, at 4:09 PM, David T. Lewis <[hidden email]> wrote: >>>>> Tim talked about how he's networked his house: his lights have switches >>>>> similar (if not actually from) Insteon. >>> Actual Insteon. See www.insteon.net & www.smarthome.com Cool new product - a network enabled LED lightbulb. Details on the house at www.rowledge.org/tim >>> >>>>> Discussion of the Raspberry Pi computer and its low price point >>>>> reemerged. Tim pointed out that the RiscOS now runs on the R. Pi computer. >>>> RiscOS on the Raspberry Pi?!? Woohoo! Clearly this calls for an update to >>>> the RiscOS VM on squeakvm.org, otherwise people will not be able to have >>>> proper block closure support on their Pi. Tim, I'll promise to buy you a >>>> brand new Raspberry Pi if you compile a fresh RiscOS VM for it ;) >>>> >>> I might be able to do something to resurrect RISC OS Squeak. It's interesting to see how people are trying it on RPi and liking the responsiveness over linux etc. Of course, it's still not a 'proper modern OS' since it lack memory protection and pre-emptive process scheduling, but in practice it works really quite well. Some of the other ARM boards currently available are even faster. >>> >>> I'd need a pi, the RISC OS SD card, the NutPi package for the relevant dev tools and some idea of a display to use with it. I'll investigate. Who knows…. >>> >>> >>> >>> tim >>> -- >>> tim Rowledge; [hidden email]; http://www.rowledge.org/tim >>> Liability: a valuable political skill >>> >>> >> I'm interested in this too. I also would be willing to provide you (like David) a RPi or related tool if you needed it. I think this would be a fun way to poke around how the vm works. >> >> Chris > I got a Raspberry Pi sent to Tim. > > He still needs an SD card with RISC OS, the Nut Pi tools card, and a paying job. > > Nut Pi tools (£35 plus P&P): > https://www.riscosopen.org/content/sales/nutpi > > RISC OS (free, just need an SD card): > http://www.raspberrypi.org/downloads > > - Bert - > > > his address? Chris |
On 2012-11-07, at 18:28, Chris Cunnington <[hidden email]> wrote: > On 2012-11-07 12:14 PM, Bert Freudenberg wrote: >> On 2012-11-06, at 19:02, Chris Cunnington <[hidden email]> wrote: >> >>> On 2012-11-06 12:42 PM, tim Rowledge wrote: >>>> Hi folks; >>>> >>>> On 02-11-2012, at 4:09 PM, David T. Lewis <[hidden email]> wrote: >>>>>> Tim talked about how he's networked his house: his lights have switches >>>>>> similar (if not actually from) Insteon. >>>> Actual Insteon. See www.insteon.net & www.smarthome.com Cool new product - a network enabled LED lightbulb. Details on the house at www.rowledge.org/tim >>>> >>>>>> Discussion of the Raspberry Pi computer and its low price point >>>>>> reemerged. Tim pointed out that the RiscOS now runs on the R. Pi computer. >>>>> RiscOS on the Raspberry Pi?!? Woohoo! Clearly this calls for an update to >>>>> the RiscOS VM on squeakvm.org, otherwise people will not be able to have >>>>> proper block closure support on their Pi. Tim, I'll promise to buy you a >>>>> brand new Raspberry Pi if you compile a fresh RiscOS VM for it ;) >>>>> >>>> I might be able to do something to resurrect RISC OS Squeak. It's interesting to see how people are trying it on RPi and liking the responsiveness over linux etc. Of course, it's still not a 'proper modern OS' since it lack memory protection and pre-emptive process scheduling, but in practice it works really quite well. Some of the other ARM boards currently available are even faster. >>>> >>>> I'd need a pi, the RISC OS SD card, the NutPi package for the relevant dev tools and some idea of a display to use with it. I'll investigate. Who knows…. >>>> >>>> >>>> >>>> tim >>>> -- >>>> tim Rowledge; [hidden email]; http://www.rowledge.org/tim >>>> Liability: a valuable political skill >>>> >>>> >>> I'm interested in this too. I also would be willing to provide you (like David) a RPi or related tool if you needed it. I think this would be a fun way to poke around how the vm works. >>> >>> Chris >> I got a Raspberry Pi sent to Tim. >> >> He still needs an SD card with RISC OS, the Nut Pi tools card, and a paying job. >> >> Nut Pi tools (£35 plus P&P): >> https://www.riscosopen.org/content/sales/nutpi >> >> RISC OS (free, just need an SD card): >> http://www.raspberrypi.org/downloads >> >> - Bert - >> >> >> > I could get him the Nut Pi, but how would I send it to him? Do you have his address? > > Chris (address sent privately) Yay! Thanks! - Bert - |
On 2012-11-07 12:34 PM, Bert Freudenberg wrote:
> On 2012-11-07, at 18:28, Chris Cunnington <[hidden email]> wrote: > >> On 2012-11-07 12:14 PM, Bert Freudenberg wrote: >>> On 2012-11-06, at 19:02, Chris Cunnington <[hidden email]> wrote: >>> >>>> On 2012-11-06 12:42 PM, tim Rowledge wrote: >>>>> Hi folks; >>>>> >>>>> On 02-11-2012, at 4:09 PM, David T. Lewis <[hidden email]> wrote: >>>>>>> Tim talked about how he's networked his house: his lights have switches >>>>>>> similar (if not actually from) Insteon. >>>>> Actual Insteon. See www.insteon.net & www.smarthome.com Cool new product - a network enabled LED lightbulb. Details on the house at www.rowledge.org/tim >>>>> >>>>>>> Discussion of the Raspberry Pi computer and its low price point >>>>>>> reemerged. Tim pointed out that the RiscOS now runs on the R. Pi computer. >>>>>> RiscOS on the Raspberry Pi?!? Woohoo! Clearly this calls for an update to >>>>>> the RiscOS VM on squeakvm.org, otherwise people will not be able to have >>>>>> proper block closure support on their Pi. Tim, I'll promise to buy you a >>>>>> brand new Raspberry Pi if you compile a fresh RiscOS VM for it ;) >>>>>> >>>>> I might be able to do something to resurrect RISC OS Squeak. It's interesting to see how people are trying it on RPi and liking the responsiveness over linux etc. Of course, it's still not a 'proper modern OS' since it lack memory protection and pre-emptive process scheduling, but in practice it works really quite well. Some of the other ARM boards currently available are even faster. >>>>> >>>>> I'd need a pi, the RISC OS SD card, the NutPi package for the relevant dev tools and some idea of a display to use with it. I'll investigate. Who knows…. >>>>> >>>>> >>>>> >>>>> tim >>>>> -- >>>>> tim Rowledge; [hidden email]; http://www.rowledge.org/tim >>>>> Liability: a valuable political skill >>>>> >>>>> >>>> I'm interested in this too. I also would be willing to provide you (like David) a RPi or related tool if you needed it. I think this would be a fun way to poke around how the vm works. >>>> >>>> Chris >>> I got a Raspberry Pi sent to Tim. >>> >>> He still needs an SD card with RISC OS, the Nut Pi tools card, and a paying job. >>> >>> Nut Pi tools (£35 plus P&P): >>> https://www.riscosopen.org/content/sales/nutpi >>> >>> RISC OS (free, just need an SD card): >>> http://www.raspberrypi.org/downloads >>> >>> - Bert - >>> >>> >>> >> I could get him the Nut Pi, but how would I send it to him? Do you have his address? >> >> Chris > (address sent privately) > > Yay! Thanks! > > - Bert - > > > Chris |
On 07-11-2012, at 9:42 AM, Chris Cunnington <[hidden email]> wrote [snip] >>> I could get him the Nut Pi, but how would I send it to him? Do you have his address? >>> >>> Chris >> (address sent privately) >> >> Yay! Thanks! >> >> - Bert - >> >> >> > OK, that's done. It should find it's way to the right place. > > Chris > tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Oxymorons: Microsoft Works |
In reply to this post by Bert Freudenberg
I just saw where the FreeBSD foundation has awarded a grant to have the OS ported to the Genesi Efika MX Smartbook portable computer. I would like to get Squeak running on this platform. Also looking at their MX53 as the core of a self-made project. I am concerned that the hardware is underpowered, 800MHz Freescale i.MX515 CPU and 512MB RAM. I see where the new OLPC uses 1GB RAM. If we assume that the FreeBSD port will be on par with the available Linux OS in terms of RAM requirements and performance, is this enough juice to provide an acceptable level of performance in Squeak?
How difficult would it be to port the existing FreeBSD vm, which is Intel based, to the Freescale CPU? (http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=i.MX515) On Wed, Nov 7, 2012 at 7:14 AM, Bert Freudenberg <[hidden email]> wrote:
-- Gary Dunn Open Slate Project http://openslate.org/ |
On 07-11-2012, at 1:39 PM, Open Slate <[hidden email]> wrote: > I just saw where the FreeBSD foundation has awarded a grant to have the OS ported to the Genesi Efika MX Smartbook portable computer. I would like to get Squeak running on this platform. Also looking at their MX53 as the core of a self-made project. I am concerned that the hardware is underpowered, 800MHz Freescale i.MX515 CPU and 512MB RAM. I see where the new OLPC uses 1GB RAM. If we assume that the FreeBSD port will be on par with the available Linux OS in terms of RAM requirements and performance, is this enough juice to provide an acceptable level of performance in Squeak? > > How difficult would it be to port the existing FreeBSD vm, which is Intel based, to the Freescale CPU? (http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=i.MX515) It wouldn't. Leaving aside the work to make the dynamic translation VM, you load the code, build environment, OS etc on your machine. You `make`with whatever options and stuff is currently required (see the unix vm instructions, probably on squeakvm.org) and wait. Your main problem would be RAM related; so far as I can make out pretty much all current *nix & related OSs have insane memory appetites. Squeak - indeed Smalltalk in general - is now typically less memory hungry than a simple word processor. Hell, a Squeak image can be smaller than a Word document. As an aside, there are literally dozens, possibly hundreds of extremely cheap ARM based tablets and laptops now available. How does a $70 10" screen ARM CortexA9 laptop grab you? Or a $46 7" tablet? Or perhaps a MacBookAir clone 13" screen A9 laptop? tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim "Bother" said Pooh as he said f**k in the wrong conf. |
Actually, compiling from source is the FreeBSD way. Last time I tried, a couple years ago, I could not compile a vm from source due to the old version of gcc available. The FreeBSD folks have decided against bringing GPL 3 code into the base because it conflicts with the BSD license. The vm required a newer version of gcc. I believe that Ian Piumarta builds the FreeBSD binaries -- which work great -- on a Linux box. So, I was wondering if the same technique could be used to create a vm for the Smartbook. Another issue is resources. Does a smartbook have enough RAM and "disk" space to compile a vm? Cross compiling on a full-sized desktop seems like a more practical solution.
As for all those inexpensive tablets and smartbooks, yes, the price is attractive, but a self-made client will give students many opportunities for Project Based Learning. When I was in school we had wood and Home-Ec. Time to replace those with computer maintenance and network management. No better way to learn how a compter works than to build one. On Wed, Nov 7, 2012 at 11:51 AM, tim Rowledge <[hidden email]> wrote:
-- Gary Dunn Open Slate Project http://openslate.org/ |
In reply to this post by openslateproj
On 2012-11-07, at 22:39, Open Slate <[hidden email]> wrote:
> I just saw where the FreeBSD foundation has awarded a grant to have the OS ported to the Genesi Efika MX Smartbook portable computer. I would like to get Squeak running on this platform. Also looking at their MX53 as the core of a self-made project. I am concerned that the hardware is underpowered, 800MHz Freescale i.MX515 CPU and 512MB RAM. I see where the new OLPC uses 1GB RAM. If we assume that the FreeBSD port will be on par with the available Linux OS in terms of RAM requirements and performance, is this enough juice to provide an acceptable level of performance in Squeak? The RPi has only 256 MB of RAM and Squeak runs fine under Linux (they do not use GNOME or KDE but a less memory-hungry environment, LXDE). The main problem is display speed. Since the display has lots of pixels, and Morphic isn't exactly frugal about screen updates, we end up copying a ton of pixel data around. > How difficult would it be to port the existing FreeBSD vm, which is Intel based, to the Freescale CPU? (http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=i.MX515) Should Just Work. - Bert - |
On Thu, Nov 8, 2012 at 2:11 AM, Bert Freudenberg <[hidden email]> wrote:
Yeah, I've given up on Gnome. Currently using Enlightenment on my desktop but that would not be good on a tablet. Sounds like Squeak has not become blotted as hardware specs expanded. That's good. Why would the port "just work?" Does the Freescale chip execute i86 instruction set? (I feel like I am in over my head here.) My thinking is that the ARM chip will require its own vm, then an image will "just work." -- Gary Dunn Open Slate Project http://openslate.org/ |
Compiling the unix VM should just work, as others have pointed out. And since BSD ports are source-based, that should Just Work, too. - Bert - |
I believe the VM is, at base, compiling C code against an OS - not a chip set. So, given C compiler for BSD, it should just compile and run.
The COG VM, where they do machine level JITting, though, that requires knowledge of the chip used.
-Chris On Thu, Nov 8, 2012 at 2:37 PM, Bert Freudenberg <[hidden email]> wrote:
|
On 08.11.2012, at 23:41, Chris Cunningham <[hidden email]> wrote:
That's a bit over-simplified. The interpreter VM is implemented in C, and the same C code should work pretty much on all 32 bit processors (and nowadays 64 bit processors too), perhaps except for the FFI plugin. It still needs to be compiled specifically for that processor architecture though - you can't run an x86 VM on ARM. The Cog VM is implemented in C, too, but has parts that generate machine code directly. Those parts are implemented only for Intel's x86 32 bit architecture yet. An ARM port is underway but far from finished. - Bert -
|
In reply to this post by openslateproj
On 08-11-2012, at 2:29 PM, Open Slate <[hidden email]> wrote: > > Why would the port "just work?" Does the Freescale chip execute i86 instruction set? (I feel like I am in over my head here.) My thinking is that the ARM chip will require its own vm, then an image will "just work." To expand upon Bert's comment, the 'ordinary' VM is a simple (for certain values of 'simple') interpreter and has no particular dependence upon the exact cpu when one is looking from the Smalltalk side. Obviously, the VM program must be compiled for the cpu upon which it will run. And the ARM most definitely does not run x86 instructions. Well, except for the small number that were built that did in fact run x86 as well as ARM. But that was a long time ago and in a galaxy far, far, away. It had nothing to do with them being radiation hardened as well. Really. ARMs up to architecture v7 are 32 bit (OK the A15 is kinda-sorta 40 bit addresses but we'll do our best to ignore that outburst of insanity) and the new and exciting v8 is 64 bit and likely to crush intel. When there are real A57 cpu based units to play with we'll see what happens. At some point I will probably have a go at doing the CogVM stuff needed for ARM cpus. Enormous amounts of money deposited into my bank account would speed up that process, as well as Squeak, Pharo & Scratch. ARM v8 would be a quite separate job since it is an almost completely new ISA. I'd also like to have a chance to re-do some ancient self-compiling bitBLT stuff since it made a huge difference to graphics performance - in '88 an 8MHz ARM2 machine could bitBLT faster than any other Smalltalk implementation of the time. Ah, Abba is on the radio; "money, money, money…." tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Strange OpCodes: HEM: Hide Evidence of Malfunction |
On 2012-11-09, at 03:01, tim Rowledge <[hidden email]> wrote:
> > On 08-11-2012, at 2:29 PM, Open Slate <[hidden email]> wrote: >> >> Why would the port "just work?" Does the Freescale chip execute i86 instruction set? (I feel like I am in over my head here.) My thinking is that the ARM chip will require its own vm, then an image will "just work." > > To expand upon Bert's comment, the 'ordinary' VM is a simple (for certain values of 'simple') interpreter and has no particular dependence upon the exact cpu when one is looking from the Smalltalk side. Obviously, the VM program must be compiled for the cpu upon which it will run. And the ARM most definitely does not run x86 instructions. Well, except for the small number that were built that did in fact run x86 as well as ARM. But that was a long time ago and in a galaxy far, far, away. It had nothing to do with them being radiation hardened as well. Really. > > ARMs up to architecture v7 are 32 bit (OK the A15 is kinda-sorta 40 bit addresses but we'll do our best to ignore that outburst of insanity) and the new and exciting v8 is 64 bit and likely to crush intel. When there are real A57 cpu based units to play with we'll see what happens. > > At some point I will probably have a go at doing the CogVM stuff needed for ARM cpus. Enormous amounts of money deposited into my bank account would speed up that process, as well as Squeak, Pharo & Scratch. ARM v8 would be a quite separate job since it is an almost completely new ISA. > > I'd also like to have a chance to re-do some ancient self-compiling bitBLT stuff since it made a huge difference to graphics performance - in '88 an 8MHz ARM2 machine could bitBLT faster than any other Smalltalk implementation of the time. > > Ah, Abba is on the radio; "money, money, money…." > > tim > -- > tim Rowledge; [hidden email]; http://www.rowledge.org/tim > Strange OpCodes: HEM: Hide Evidence of Malfunction Man, I really missed seeing your quips :) For the state of the ARM Cog VM, see Lars' GSoC report: http://lists.squeakfoundation.org/pipermail/vm-dev/2012-August/011143.html - Bert - |
Free forum by Nabble | Edit this page |