XMOS announces $99 dev kit!
Leon
Posts: 7,620
XMOS has just announced a $99 dev kit for their XC-1 chip:
http://forums.parallaxinc.com/products.xmos.com/xc1
1600 MIPS for $99 can't be bad.
I've put my name down for one, I've always been a sucker for interesting hardware.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
http://forums.parallaxinc.com/products.xmos.com/xc1
1600 MIPS for $99 can't be bad.
I've put my name down for one, I've always been a sucker for interesting hardware.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Comments
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
J
On the other hand for some time I have been pondering the idea of putting a Prop and an FPGA on a little board but perhaps a Prop with XS1 would be neat also:
1) Provides some external memory for the Prop.
2) Implement USB and ethernet etc I/O
3) Loads of extra I/O
4) Don't have to get into Verilog or VHDL
5) Potentially cheaper than FPGA
But why keep the Prop then? Hmmm....
Still at the moment the 20 euro I spent building my home made Prop demo board is a lot less than 99 dollars.
Must check this out some more.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
I've just spoken to my contact at XMOS. He was about to catch a plane to the USA, so it was quite brief. They intend to put the chips in QFP or QFN packages, which is better for the hobbyist. He told me that they have already had some big design wins - including one in Japan in a completely unexpected application. There is also a lot of interest by companies making high-end audio systems. I see it as being very useful in software-defined radio systems.
It'll be interesting to see how this affects the development of the Prop II. It looks as though the XMOS devices will be cheaper as well as offering much more performance and better tools.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
However, it would be a lot easier to use a CPLD like one of the Altera Mach II chips to start with. They actually have an FPGA architecture but are available with only 44 pins. Development would be very cheap (I could even make a PCB for a prototype at home).
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
That's exactly what we have been thinking too and came to the same conclusion, why keep the Prop.....
So we abandoned the idea cos we love the Prop so much.....
This chip is worth a look at though.....
Regards,
Coley
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PropGFX Forums - The home of the Hybrid Development System and PropGFX Lite
I doubt it will. I'm sure Chip & Co will look at what it offers, contrast and compare with the Propeller II but I think the market each is aiming for is significantly distinct for both to co-exist.
That may be, but then one has to also weigh-up ease of use and simplicity of programming.
The bottom line question is, "is the XC-1 a Propeller II killer" ? I don't think so.
It seems human nature to get excited by "faster" and 1600 MIPS looks phenomenally good, but do people really need that ? Some people undoubtedly do, and some people want that, and it will probably do well for those sectors. Cost plays its part as well, so do tools, but I personally don't see a compelling reason to choose the XC-1 over the Propeller.
I wish the XC-1 every success, but for me the Propeller family is a much cleaner design with amazing ease of use.
I know which one I'll by prefering.
not forgetting the price, the dev board is $99, on that you don't get much, the chip + what? 12leds + 4 buttons?
I bet the propII proto board will be better and cheaper even with paying extra for the vga and ps2 addon, I bet it'll still be cheaper.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
http://www.propgfx.co.uk/forum/·home of the PropGFX Lite
·
I still want one though! (Might see you on 23rd Leon!)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Cheers,
Simon
www.norfolkhelicopterclub.com
You'll always have as many take-offs as landings, the trick is to be sure you can take-off again
BTW: I type as I'm thinking, so please don't take any offence at my writing style
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (trodoss) : 9/30/2008 2:25:19 PM GMT
It looks to me that "XC" is an enhancement of Occam which has been moulded to fit within a C-style framework - not surprising as XMOS is the legacy of INMOS and its Transputer chip. The emphasis being on running tasks / threads in parallel and passing data through channels rather then the Propeller model which is more about how to get eight MCU's with a shared memory co-operating. The idea of channels ( "links" ) rather than shared memory is perhaps the most significant concept difference.
Where XS wins over Propeller is it's I/O capability, which looks to be 48 per core ( but needs a BGA package ), its speed of 400MIPS per core against Prop II's 160MIPS, total of 1600MIPS (XS1-G4), 1280MIPS for the Prop II. The XS1 has 64KB per core, the Prop 2KB but I wouldn't know how to compare the two in practical terms.
To me it looks like the XS1 is best suited to software-based parallel processing, its large I/O capability primarily for interfacing to memory or maybe LCD, the Prop II more suited to traditional microcontroller products with bit-banging I/O. I can imagine the XS1 dedicating a whole core to bus memory interfacing, taking 25% (XS1-G4) or 50% (XS1-G2) of its cores to do that. The Prop II would have more cores free after dedicating one to bus memory interfacing but that would quite significantly reduce its directly accessible I/O.
I can also visualise the XS1 as a single core processor with soft-peripherals in the other cores, similar to how the Prop II could be used. With memory access through one core I'd see a reasonable target market of PDA, phone, UMPC for the XS1, the Propeller holding the more 'electronic interfacing' market.
No pricing details available beyond the XS1 kit. Five times mark-up would put an XS1-G4 at near $20 per chip but that's a guess. Hard to tell if XS1 is priced high to recoup development costs or priced low to attract interest. Chips with less cores one would expect to be cheaper. We don't have a solid price for the Prop II either, but the estimate there has been twice the cost; around $25.
One factor which is impossible to estimate is how well a British product will do in foreign markets, especially the US. A number of quite capable products from the UK and Europe have failed to have any real impact in the US in the past. That may come down to marketing and the good ol' trick of passing a foreign product off as an American one.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Chris Savage
Parallax Engineering
They are bringing out a kit specifically for hobbyists early next year.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 10/2/2008 9:38:14 PM GMT
I've had a play with their on-line development software, it looks pretty good. Here is a simple UART transmit function written in XC:
/**************************************************************** * * UartTransmit.xc * * Uart transmit * * */ #include <xs1_reva.h> #define BIT_TIME 1000 #include <stdio.h> out port txd = XS1_PORT_1A; void uart_transmit(out port txd, timer tmr, char bytes[noparse][[/noparse]], int numBytes) { int i; unsigned time; tmr :> time; for (i = 0; i < numBytes; i += 1) { int j; int byte = bytes[noparse][[/noparse] i ]; // Start bit time += BIT_TIME; txd @ time <: 0; // Data bits for (j = 0; j < 8; j += 1) { time += BIT_TIME; txd @ time <: >> byte; } // Stop bit time += BIT_TIME; txd @ time <: 1; } } int main() { timer tmr; char bytes[noparse][[/noparse]] = { 1, 2, 3 }; uart_transmit(txd, tmr, bytes, 3); puts("Done!"); return 0; }I added the puts("Done!") output to check the simulator.
Can anyone see why the font changes to italics in the middle of the code?
They have a good support forum:
www.xlinkers.org/
I've posted a couple of queries and got quick responses from XMOS people.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 10/3/2008 5:40:26 PM GMT
Thanks. I didn't notice the missing [noparse][[/noparse] i ]. I've added the spaces.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Interesting the product is seen, I expect that its dev kit have a price reduced in the future.
A question Leon: Functions like the Propeller, that is to say with "Objects"?
Thanks Leon for all this info.
No "objects" as such, the XC language is very low-level. They will be providing libraries with lots of useful stuff in them. C++ could be implemented, I suppose.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
On the final point, XMOS's long term future; the failure of INMOS and Transputer had a lot to do with the way the industry in the UK was run and financed back then plus its being somewhat ahead of its time while XMOS are in a very different position now. XMOS may fail, and that could be for many reasons, but I wouldn't predict that simply on the back of previous INMOS woes.
What is interesting to note is that the Transputer was programmed in an indentation important language, Occam, much as the Propeller is with Spin. In the latest incarnation, an extended version of C has been chosen. The same complaints levelled at Occam not being an industry standard nor familiar language are levelled at Spin now from some quarters ( and also at SEAforth product ). How providing development in C helps or doesn't is something which could provide lessons for the future development of similar multi-core chips.
There also seems to be an innuendo in your comment of "Given the nature of British technological progress ..." which doesn't really match reality and isn't entirely fair. Britain did of course gave the world ARM technology as just a single example.
David May has plenty of venture capital for XMOS. To produce a new chip architecture and working silicon in under three years with such a small team (Parallax apart) is quite unprecedented; they should do very well.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 10/4/2008 4:22:41 PM GMT
Parallax:
1) Has a small, dedicated, enthusiastic team, most of whom give the impression that they would pay to work there if they had to.
2) Has been around for quite some years already.
3) They have no banks, shareholders, or venture capitalists breathing down their necks (as far as I know)
4) They have a business model that looks to be sustainable for the long haul even without the Propeller.
5) They have made promises as best they can that the Prop will be around for at least ten years, which given the above seems quite reasonable.
XMOS:
1) David May is of course one of my heros as for the rest we don't know.
2) Totally new (Flash in the pan?)
3) Has a huge responsibility to make money of their new gizmo. If not the plug may get pulled rather quickly.
4) Is a one song band. If that song is not a hit and quickly there they go.
5) ????
Sorry about any innuendo in my comment of "Given the nature of British technological progress ..." I phrased my thoughts very badly there. As a Brit myself I am very proud of the technological and scientific contributions of those "Island Monkeys" (See note 1) from Stone Henge through the industrial revolution to Concorde the Transputer and on.
What I was really thinking is that for a long time there seemed to be a lot of good talent working on cutting edge technology and products but that somehow it always got screwed up by lack of financing, lack of management vision, lack of political will, hopeless marketing or what ever.
Examples:
The death of the British motorcycle industry when modern designs pushed by the engineers were ignored by their management.
Concorde - Brilliant and beautiful but not really a commercially viable proposition.
The collapse of big chunks of the defence sector and hence it's technological contribution to everything else when we decided to buy from the Yanks instead. When I was at Marconi Radar Systems we had quite an OK Airborne Early Warning System but it was dropped.
In the early eighties I worked for a large engineering firm that had developed a brilliant automated production line machining system basically robotics 10 or twenty years ahead of it's time. Very flexible, very accurate. They only sold I think 2 systems to Texas instruments then had to drop it as to expensive. When time came to redevelop it with new cheap micros (instead of big expensive PDP11s) no one was interested. Except Texas Instruments who came back to us for the source code 1o years later as they still wanted to enhance it.
The statement by Margaret Thatcher that "Britain need not manufacture anything any more, we can get by on services"
The INMOS product launch for the T9 which I attended was for sure not the style of sales pitch we used to get from such events by Intel.
I could go on but you get the idea. It used to depress me no end. Hope times have changed since I've been away.
Anyway I wish all the best to both Parallax and XMOS. I'm sure they will both keep me fascinated for a long time.
Note 1) Here in Helsinki I have been referred to as an "Island Monkey" on a number of occasions when they find out where I'm from. See here knowledgerush.com/kr/encyclopedia/Informal_terms_for_the_Finnish/
I quite like it actually.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
I'm getting very excited about XMOS, and am even creating a PCB part for the XS1-G4 (BGA512), ready for when I can get my hands on some sample chips. I've put in all the pin names (not too difficult as I was able to copy/paste them, assigned the pin types and am just about to start assigning the pins. It'll need a 6-layer PCB, of course.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 10/6/2008 11:38:08 AM GMT
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Cheers,
Simon
www.norfolkhelicopterclub.com
You'll always have as many take-offs as landings, the trick is to be sure you can take-off again
BTW: I type as I'm thinking, so please don't take any offence at my writing style
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
The official report after the event should have come as no surprise to those who had been following the project. There was so much wrong with that project it was a near inevitable outcome and the damaging consequences in its wake were entirely predictable.
Sadly I think it has got worse, certainly with respect to manufacturing and industry. It seems we've moved to service and financial sectors and even a lot of that has been exported for others to do. We're a country in descent IMO, shored up by smoke and mirrors and recent weeks have shown just how fragile it all is.
Strangely enough my father was a Czechoslovakian, in England since his time in the British army during the second world war, almost never heard him speak Czech and never did learn it. Very sad actually.
Sounds like you have another interesting board on the horizon is this one we can sign up for as well ?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.