David May is very approachable, he even contacted one user who had a problem on Xmas day! He's probably the only FRS who rides a fast motorbike. I've never dealt with him personally, but I know one of his sons.
I would not criticise the professionalism of David May or Chip Gracey. They both have a passion for what they do that shines out like a beacon. Different styles, different approaches for sure but well, they are different people.
As for approachable, David was quite happy to take the time over the Christmas break to chat with my humble self on their forum and we all know how receptive Chip is to user input here.
Maybe I was a bit harsh, begging for pardon! It is just a different business model, resulting from different histories. Maybe, one wanted the possible and reached the impossible, the other one wanted the impossible and reached the possible. ;-)
They are similar companies in many ways. They both have about 50 employees, and are run by real enthusiasts who believe in what they are doing. XMOS gives $250 worth of hardware to any university lecturer who requests it. A friend of mine took advantage of their offer, and has several students doing projects with the boards he got.
With the major difference that XMOS is a start up kicked off with tens of millions of venture capital. Presumably the investors will want to see a return on their money at some point. They have all their eggs in one basket and if the xcores are not a hit the whole thing vaporises.
On the other hand Parallax is privately owned (as far as I can tell) and has been grown up from nothing with a diversity of products. The Prop need not turn in a huge profit for it to continue indefinitely.
Interestingly XMOS recieved $16 million in venture funding from Foundation Capital, Amadeus Capital Partners and DFJ Esprit. Amadeus Capital is Hermann Hauser a co-founder of Acorn Computers and hence the ARM chip. So I'm confident XMOS will be around for a while.
This discussion is around the propeller and so xmos is just a satellite ;-) Think, the failure of the transputer was not David's, but I also think, they wanted to much with the T9000. So, we should not discuss, why the propeller is not in the consumer market. We are consumers! But we do not throw away what we used one time and what is now old fashioned. We could try to spread this virus we all love. Maybe, Leon want to have the propellers for himself and therefor tells us about the bright shining sun outside. Didn't we have something like this thousands of years ago? But we have no need to go to the desert to find the Promised Land. We just have to wait and be ready when the second empire arrives
They are similar companies in many ways. They both have about 50 employees, and are run by real enthusiasts who believe in what they are doing. XMOS gives $250 worth of hardware to any university lecturer who requests it. A friend of mine took advantage of their offer, and has several students doing projects with the boards he got.
I feel it is reasonable to ask for a significant reduction of the general XMOS promotion on a Parallax forum. Leon, today I respectfully request a reduction and maybe even a complete termination [depending on future posts] of the feature comparisons, cross-marketing (or whatever it's called), and overall attention given to XMOS. We concentrate on what we do best at Parallax, and I'm sure that XMOS does the same.
We really want Propeller newbies to be well-received by the community. And I'm truly impressed with how supportive the community is when I simply posed the reminder in another thread.
What to post is a fine line, Leon, and a difficult one for me to describe. I don't have the time to describe the variety of posts that I might consider inappropriate. In fact, since this is the first time I've ever addressed this kind of issue I'll just leave it up to you to exercise more judgment. The only other moderation I've done on these forums is spam removal, avatar removal, and a few thread relocations. Nothing has been censored. I have a tremendous tolerance but I've finally been pushed off balance.
We respect all companies in this business, and will always maintain a positive marketing position. You will never find any Parallax marketing material that makes feature comparisons. For example, we wouldn't offer any programs that allow you to send in an Arduino for a BASIC Stamp. We focus on what we do best and what our customers want us to do. This internal philosophy should be applied to our forums to some extent.
Leon, you're part of our team and I encourage you to contribute. I read your posts on other forums regularly and recognize that you've got plenty of valuable experience to share with others.
Pity that our two choices are 1) a list of commands and 2) a 300 page tutorial that starts with the creation of the World...but leaves out all the good stuff.
An apology is in order. I downloaded PM 1.1 today, flipped immediately to the Assembly Language Reference, and was pleasantly surprised with what I found there. A bunch of new stuff had been added, and a simple template was included that would get a feller off the ground immediately. The pap and palaver was practically non-existent.
So I apologize for being a humbug.
I was minutes away from ordering a whole new round of ARM chips for my ersatz parallel supercomputer when I thought I'd give the Prop a few more minutes of consideration...the prospect of six or seven cogs per chip cranking away on my "Nine Billion Names of God" project was enticing. But the bar seemed too high, and the support documentation too low. Turns out that is ancient history. So it's all good!
Note: The Asimov story seemed an apt metaphor for what I'm doing - even though what I'm doing is entirely secular. But with the Mayan calendar ending soon, how can anyone be sure?! Asimov may have been right. Maybe my little supercomputer is bringing about the End! Parallax may inadvertently be hastening the End by providing me Propellers. Quick! Ignite the thermite!
An apology is in order. I downloaded PM 1.1 today, flipped immediately to the Assembly Language Reference, and was pleasantly surprised with what I found there. A bunch of new stuff had been added, and a simple template was included that would get a feller off the ground immediately. The pap and palaver was practically non-existent.
So I apologize for being a humbug.
I was minutes away from ordering a whole new round of ARM chips for my ersatz parallel supercomputer when I thought I'd give the Prop a few more minutes of consideration...the prospect of six or seven cogs per chip cranking away on my "Nine Billion Names of God" project was enticing. But the bar seemed too high, and the support documentation too low. But that turned out to be ancient history. So it's all good!
Note: The Asimov story seemed to be an apt metaphor for what I'm really doing - even though what I'm really doing is entirely secular.
Actually, K2, your request is a very common one. We are still short on ASM resources for the Propeller. Many inside Parallax feel the same - several efforts are underway to improve the programming resource base.
I think threads like this one are better for the "Sandbox"...
Is that because it has nothing to do with the Propeller?
@ Ken: Excellent! BTW, does Parallax award $2.56 for each error found in their documents? Or are the underscores missing from the "Toggle" ASM listing intentional, perhaps for the purpose of giving noobs a sense of mastery once they conquer that hill?
@Rayman, Re: "all"
I don't think so. The question is whether the discussion furthers the Propeller or an understanding of the Propeller, how it works, what it's good for and what it's not good for. It's not appropriate to just say "the Prop can't do X or has poor performance at Y, but look at this other product and how wonderful it is". It's also not appropriate to not mention the Propeller at all and just suggest that someone should look at a competing product.
The Propeller is not good at everything and it's not comparable to some other microcontrollers on the market with lots and lots of on-board flash and RAM. It's really good at some things, like instrumentation-like applications as has been mentioned here and applications where time to development is short, partly because of its flexibility in providing peripheral functions. It's a robust chip with easy control of power consumption. Spin and assembly is a good mix for application development covering a wide range of speed and space needs. Etc., etc., ...
There's a place here for discussions about the Propeller and whether it's a good fit for a particular application, but these need to be discussions that include the whys, what feature may be crucial for a particular application, how it might be implemented with a Propeller, where it might fall short, what external hardware might be needed. No particular microcontroller is good at everything or even most combinations of things, not XMOS, not ARM, not PIC, not AVR. One of the important characteristics of a good engineer is to know or be able to recognize what tool or tools are appropriate for the job, which tool features are important and which are arbitrary and suitable to be picked based on convenience or previous knowledge or various costs.
Like Mike said - the prop (or any other ucontroller) isn't perfect for everything.
For commercial uses, I've heard code protection and second-source objections. It doesn't have a lot of RAM, but it depends on what your use case is. As more engineers hear about the prop & as the Parallax brand / image begins to really expand beyond hobby+education, it will be used more often.
But for me, an average hobbyist, it's a dream - inexpensive, DIP package, easy programming language, OOP style code re-use, and even video output. I've played with AVR & PIC's & they're a pain compared to how easy the Prop is. The Arduino & PICaxe come close, but the Prop's faster performance, video capability & easy multitasking really set it apart.
In short - I'm sure as more engineers become aware of the Prop, commercial use will expand.
Chip made a lot of choices that I like a great deal. Executing out of RAM is one of them.
The choice ARM made to execute out of FLASH means that they end up with memory that is three times slower than their core. That can be mitigated to a degree by using a very wide memory bus and prefetching code. But branching comes with stiff penalties when you do that.
Another problem with FLASH is that the FLASH wears out. LPC ARM chips are good for only 10,000 erase/write cycles. That doesn't seem like a very big number to me anymore. I realized the other day that one of my chips would literally wear out in a few more months if I didn't change how I was using it. I hate worrying about Smile like that.
The Prop can be reloaded a billion times, or until the End comes...
K2: "The choice ARM made to execute out of FLASH means that they end up with memory that is three times slower than their core."
Depends which ARM you mean. An ATMEL AT91SAM7X256 has 64K of single cycle access RAM.
"This new learning amazes me!"
Are you saying that the CPU executes instructions stored in SRAM? What provisions does the chip provide for transferring code from FLASH to SRAM, and then transferring execution to it? Or is this an automatic thing on boot-up?
Or are you saying that this particular chip does not have a Harvard architecture?
Perhaps, Baggers, there is a difference between discussing CPU architectures and advocating for a competitor's product and company.
Personally, it seems appropriate to mention that the ARM7 is an old, slower, and less efficient design, and that most new ARM designs are Harvard. I'm not advocating ARM by saying this. In fact I have a benchmark I'm working on, comparing a specific task on a 125 MIPS ARM Cortex M3 (Harvard...) and a Propeller. Preliminary results indicate that the Prop solution is 30% cheaper than the ARM, and that the execution speed is 2.5 - 3 times faster.
It's not a comprehensive benchmark, rather it's a specific routine that my ersatz supercomputer has executed trillions and even quadrillions of times, and so small performance differences are of interest to me.
I'll also admit it's not an apples-to-apples comparison because of the language differences and my limited programming skills. I'm an engineer, not a programmer. (A real programmer would know that ADA didn't exist at the time of the Apollo program!)
But it is worth noting that the reason the ARM takes such a hit on this particular routine is the very fact that every time execution necessarily branches, the prefetch queue has to be dumped, and it has to be dumped three times each loop. The 125 MIPS rating of the ARM is strictly for inline execution. Leon knows this perfectly well. A 125 MIPS rating is also more than double Leon's LPC21XX chips or Heater's AT91SAM7X256.
Agreed. Unfortunately there is no practical way to unroll this particular loop inasmuch as exit conditions have to be examined each time. Ignoring exit conditions and then post-processing to find the correct result would penalize performance more than testing for it every loop.
I should also mention that there might be a way to speed execution of the ARM code if I were a crack ARM assembly programmer. I doubt I'll ever be that. So, in my tests, the ARM will always suffer accordingly.
On the other hand, expense-wise, I'm also not factoring in the fact that I can't hand wire LQFP100 packages but I can hand wire 40-pin DIPs. So, in practical terms, the ARM is actually costing me nearly four times more than the Prop.
Comments
I would not criticise the professionalism of David May or Chip Gracey. They both have a passion for what they do that shines out like a beacon. Different styles, different approaches for sure but well, they are different people.
As for approachable, David was quite happy to take the time over the Christmas break to chat with my humble self on their forum and we all know how receptive Chip is to user input here.
They are similar companies in many ways. They both have about 50 employees, and are run by real enthusiasts who believe in what they are doing. XMOS gives $250 worth of hardware to any university lecturer who requests it. A friend of mine took advantage of their offer, and has several students doing projects with the boards he got.
With the major difference that XMOS is a start up kicked off with tens of millions of venture capital. Presumably the investors will want to see a return on their money at some point. They have all their eggs in one basket and if the xcores are not a hit the whole thing vaporises.
On the other hand Parallax is privately owned (as far as I can tell) and has been grown up from nothing with a diversity of products. The Prop need not turn in a huge profit for it to continue indefinitely.
Interestingly XMOS recieved $16 million in venture funding from Foundation Capital, Amadeus Capital Partners and DFJ Esprit. Amadeus Capital is Hermann Hauser a co-founder of Acorn Computers and hence the ARM chip. So I'm confident XMOS will be around for a while.
I feel it is reasonable to ask for a significant reduction of the general XMOS promotion on a Parallax forum. Leon, today I respectfully request a reduction and maybe even a complete termination [depending on future posts] of the feature comparisons, cross-marketing (or whatever it's called), and overall attention given to XMOS. We concentrate on what we do best at Parallax, and I'm sure that XMOS does the same.
We really want Propeller newbies to be well-received by the community. And I'm truly impressed with how supportive the community is when I simply posed the reminder in another thread.
What to post is a fine line, Leon, and a difficult one for me to describe. I don't have the time to describe the variety of posts that I might consider inappropriate. In fact, since this is the first time I've ever addressed this kind of issue I'll just leave it up to you to exercise more judgment. The only other moderation I've done on these forums is spam removal, avatar removal, and a few thread relocations. Nothing has been censored. I have a tremendous tolerance but I've finally been pushed off balance.
We respect all companies in this business, and will always maintain a positive marketing position. You will never find any Parallax marketing material that makes feature comparisons. For example, we wouldn't offer any programs that allow you to send in an Arduino for a BASIC Stamp. We focus on what we do best and what our customers want us to do. This internal philosophy should be applied to our forums to some extent.
Leon, you're part of our team and I encourage you to contribute. I read your posts on other forums regularly and recognize that you've got plenty of valuable experience to share with others.
An apology is in order. I downloaded PM 1.1 today, flipped immediately to the Assembly Language Reference, and was pleasantly surprised with what I found there. A bunch of new stuff had been added, and a simple template was included that would get a feller off the ground immediately. The pap and palaver was practically non-existent.
So I apologize for being a humbug.
I was minutes away from ordering a whole new round of ARM chips for my ersatz parallel supercomputer when I thought I'd give the Prop a few more minutes of consideration...the prospect of six or seven cogs per chip cranking away on my "Nine Billion Names of God" project was enticing. But the bar seemed too high, and the support documentation too low. Turns out that is ancient history. So it's all good!
Note: The Asimov story seemed an apt metaphor for what I'm doing - even though what I'm doing is entirely secular. But with the Mayan calendar ending soon, how can anyone be sure?! Asimov may have been right. Maybe my little supercomputer is bringing about the End! Parallax may inadvertently be hastening the End by providing me Propellers. Quick! Ignite the thermite!
Actually, K2, your request is a very common one. We are still short on ASM resources for the Propeller. Many inside Parallax feel the same - several efforts are underway to improve the programming resource base.
(although I rarely go there...)
Is that because it has nothing to do with the Propeller?
@ Ken: Excellent! BTW, does Parallax award $2.56 for each error found in their documents? Or are the underscores missing from the "Toggle" ASM listing intentional, perhaps for the purpose of giving noobs a sense of mastery once they conquer that hill?
It's not about using a Propeller.
Propeller Chip Forum for all Propeller related discussions.
But, I think Ken is saying not "all", particularly extended discussions on exactly how features of the other chips compare to the Propeller...
I don't think so. The question is whether the discussion furthers the Propeller or an understanding of the Propeller, how it works, what it's good for and what it's not good for. It's not appropriate to just say "the Prop can't do X or has poor performance at Y, but look at this other product and how wonderful it is". It's also not appropriate to not mention the Propeller at all and just suggest that someone should look at a competing product.
The Propeller is not good at everything and it's not comparable to some other microcontrollers on the market with lots and lots of on-board flash and RAM. It's really good at some things, like instrumentation-like applications as has been mentioned here and applications where time to development is short, partly because of its flexibility in providing peripheral functions. It's a robust chip with easy control of power consumption. Spin and assembly is a good mix for application development covering a wide range of speed and space needs. Etc., etc., ...
There's a place here for discussions about the Propeller and whether it's a good fit for a particular application, but these need to be discussions that include the whys, what feature may be crucial for a particular application, how it might be implemented with a Propeller, where it might fall short, what external hardware might be needed. No particular microcontroller is good at everything or even most combinations of things, not XMOS, not ARM, not PIC, not AVR. One of the important characteristics of a good engineer is to know or be able to recognize what tool or tools are appropriate for the job, which tool features are important and which are arbitrary and suitable to be picked based on convenience or previous knowledge or various costs.
I could not agree MORE!
RS_Jim
For commercial uses, I've heard code protection and second-source objections. It doesn't have a lot of RAM, but it depends on what your use case is. As more engineers hear about the prop & as the Parallax brand / image begins to really expand beyond hobby+education, it will be used more often.
But for me, an average hobbyist, it's a dream - inexpensive, DIP package, easy programming language, OOP style code re-use, and even video output. I've played with AVR & PIC's & they're a pain compared to how easy the Prop is. The Arduino & PICaxe come close, but the Prop's faster performance, video capability & easy multitasking really set it apart.
In short - I'm sure as more engineers become aware of the Prop, commercial use will expand.
The choice ARM made to execute out of FLASH means that they end up with memory that is three times slower than their core. That can be mitigated to a degree by using a very wide memory bus and prefetching code. But branching comes with stiff penalties when you do that.
Another problem with FLASH is that the FLASH wears out. LPC ARM chips are good for only 10,000 erase/write cycles. That doesn't seem like a very big number to me anymore. I realized the other day that one of my chips would literally wear out in a few more months if I didn't change how I was using it. I hate worrying about Smile like that.
The Prop can be reloaded a billion times, or until the End comes...
Ok, I think you're right and I didn't mean to try to speak for Ken.
Still, I think these sorts of threads just invite things we don't want...
Depends which ARM you mean. An ATMEL AT91SAM7X256 has 64K of single cycle access RAM.
The NXP ARM chips have a wide 128 bit interface to the on-chip flash memory, which speeds things up considerably.
A careful rereading of my post will reveal that I addressed that very topic.
"This new learning amazes me!"
Are you saying that the CPU executes instructions stored in SRAM? What provisions does the chip provide for transferring code from FLASH to SRAM, and then transferring execution to it? Or is this an automatic thing on boot-up?
Or are you saying that this particular chip does not have a Harvard architecture?
That's odd. Mine don't.
Leon, you've already been warned! and should know better!
Personally, it seems appropriate to mention that the ARM7 is an old, slower, and less efficient design, and that most new ARM designs are Harvard. I'm not advocating ARM by saying this. In fact I have a benchmark I'm working on, comparing a specific task on a 125 MIPS ARM Cortex M3 (Harvard...) and a Propeller. Preliminary results indicate that the Prop solution is 30% cheaper than the ARM, and that the execution speed is 2.5 - 3 times faster.
It's not a comprehensive benchmark, rather it's a specific routine that my ersatz supercomputer has executed trillions and even quadrillions of times, and so small performance differences are of interest to me.
I'll also admit it's not an apples-to-apples comparison because of the language differences and my limited programming skills. I'm an engineer, not a programmer. (A real programmer would know that ADA didn't exist at the time of the Apollo program!)
But it is worth noting that the reason the ARM takes such a hit on this particular routine is the very fact that every time execution necessarily branches, the prefetch queue has to be dumped, and it has to be dumped three times each loop. The 125 MIPS rating of the ARM is strictly for inline execution. Leon knows this perfectly well. A 125 MIPS rating is also more than double Leon's LPC21XX chips or Heater's AT91SAM7X256.
I should also mention that there might be a way to speed execution of the ARM code if I were a crack ARM assembly programmer. I doubt I'll ever be that. So, in my tests, the ARM will always suffer accordingly.
On the other hand, expense-wise, I'm also not factoring in the fact that I can't hand wire LQFP100 packages but I can hand wire 40-pin DIPs. So, in practical terms, the ARM is actually costing me nearly four times more than the Prop.