128 long sure is not much memory to write code in but is a cool idea. would allow you to not have to waist an entire cog for small monitoring codes. I assume the logical cogs would run at 1/4 speed or slower to allow the hardware to thread them. Now if we could have 512 longs still for each virtual cog and have 4 pages of ram for a full cog would make things better. I know paging ram would make coding a little harder but would make larger assembly codes possible.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are now in stock. Only $30. PCB available for $5
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
If I understod Chip corect Logical COGs will have same speed as Physical
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Nothing is impossible, there are only different degrees of difficulty. For every stupid question there is at least one intelligent answer. Don't guess - ask instead. If you don't ask you won't know. If your gonna construct something, make it·as simple as·possible yet as versatile as posible.
now how would that work? he has already said the prop2 will execute 1 instruction per cycle so if you can run 4 logical cogs on 1 cog you would be running 4 instructions per cycle. That would be very impressive.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are now in stock. Only $30. PCB available for $5
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
Like posted earlier, I tend to run out of Cog or hub memory before I run out of cogs.
It pretty clear, from posts and third party products like the hydra 512 xtreme, that there is a strong demand for more memory.
There are tons of posts where someone is trying to add more memory without sacrificing too many gpio or speed. That's a pretty good sign of user needs.
Also, it's understandable that there is a desire to get away from windows (get a mac or use Linux). But it's not clear that commiting prop resources to puting the develop environment inside the chip is a good idea.
The whole team I work with would probably run away as far as possible from
the chip that had the vxworks development environment built into it. It would also prevent it from being used in critical systems, not that the prop is in this market, because development tools can't reside in them by many regulations. If it was a good idea we would probably see that more often.
grietz,
There already is a usable development environment that can run standalone on the Propeller I (Sphinx). It could use a better editor and some I/O driver work, but it runs and can be used to maintain itself (compile the compiler and I/O drivers).
There should not be a problem with critical systems and regulations unless the regulations are written very poorly. The fact that a development system is stored in masked ROM on the chip doesn't mean that it's usable without specific attached hardware and the memory that stores the "user" program doesn't have to be modifiable (it can be write protected).
Howard,
It's a bad idea. There was a long discussion earlier in this thread about such a scheme and it would really mess up the determinacy of the current Propeller design which is a deliberate design decision / advantage.
16cogs... each with 100mips
512kb eeprom on the chip to get rid of the external eeprom.
64kb ram for each cog
Maybe one much more capable video generator for all cogs to have access to
instead of or in addition to the ones now in each cog....full HD maybe.
And why can't a fab process be invented to integrate a xtal on chip or some way to get
precision frequency control without an external xtal? xtals on the board seems sorta last century.
There just has to be a way....maybe approximate freq control that is fine tunable like some
AVRs do with the internal oscillator...only more precise.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
CRPs>> Just curious - have the Prop II's specs been frozen yet?
Mike> It's a bad idea. There was a long discussion earlier in this thread about such a scheme
> and it would really mess up the determinacy of the current Propeller design which is a deliberate design decision / advantage.
Mike,
ah, sorry, I may have unwittingly used a silicon-industry jargon word: "frozen" --- I was merely asking if the architectural design of the Prop II was decided and set in stone yet.
Phil> Maybe it takes a wooden stake and a silver bullet to keep this thread buried!
No, Phil, that's not powerful enough ...
it will take a wooden stake, silver bullet, and a Prop II to end this beast ! [noparse]:)[/noparse])
A fab process needs to be developed that gives you a standard dip or sm chip but has a set of connectors on top that
connects to all the interesting leads of the silicon wafer inside. It could be a sort of really tiny ball grid array covered by
a sticker that could be peeled off. This would allow a controller chip to be customized by soldering a thin board on top.
Perhaps even laying on a layer of plastic to secure and obscure the added on board. You could add things like optical I/O,
wifi, ethernet, eeprom, ram...whatever. With current methods a lot of stuff can be put onto a minuscule board..just look
inside a small cellphone at those boards!
This grid that could allow layering a board or several atop a chip would be a first move in the direction of 3d chips instead
of current 2d designs...3d chip design is inevitable, might as well start now.
And why not use clear plastic as a covering for chips instead of always black? It would let you add some of those incredibly tiny sm
LEDs inside as diagnostic aids....also IR I/O could be added if the plastic was translucent to IR.
Chips could also be made with tiny socket holes on top so add on chips could just be pressed onto them...easy and so simple!
Why not???
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
HollyMinkowski said...
And why not use clear plastic as a covering for chips instead of always black? It would let you add some of those incredibly tiny sm
LEDs inside as diagnostic aids....also IR I/O could be added if the plastic was translucent to IR.
It's because silicon is light-sensitive, and any stray photons can affect performance in undesirable ways. Micron Semiconductor's image sensor line began with the discovery that their DRAM chips, if covered with a transparent window, could be used as crude imaging chips.
Plus the enhancements to the video shift registers and timers I suggested earlier, and the I/O strobes.
oohhh the things we will be able to do!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Chip....
Are you going to stay in a 40 pin dip package, or going to say plcc 84 to get the 'port b' on-line? Having both choices would be good; and it could be a packaging option.
If you are going 84 pin, we can get the port strobes without sacrificing general purpose I/O pins!!!!!!
are branches still going to take effectively one clock if taken / two if not?
If you need more cycles, how about adding delay slots, so that we can use the otherwise wasted cycles?
Chip Gracey (Parallax) said...
Bill Henning said...
Hmm... i surface from consulting work, and face a tough question...
Before I can intelligently decide which way to vote, I'd like to know how many cycles the hub instructions will take...
Also, I'm guessing you are targeting 160MHz... correct?
Hub instructions will take 2 clocks, so you can fit six regular instructions between them. 160MHz is the clock goal. Cogs are pipelined so instructions take 1 clock.
Chip, I like what I am seeing about the new propeller, although I have not been able to read through the entire 826 messages in this thread (so sorry if this has already been answered.) My focus is on education and training, and as such one of the more important aspects of a new device is its availability in DIP form - above 40 pin will still work, but PLCC and other packages like that will very much lower the usability of the device for my products (unless they are available in DIP format also, like the current Propeller). I have seen many places where people have mentioned that the DIP format is one great feature of the current Propeller, and I would have to agree (at least for my product lines.) I do realize that this will limit the number of I/O pins you can put in a chip, but ability to easily breadboard a new design is pretty important to people learning a new controller.
Thanks for your hard work on these great products.
Granz said...
the DIP format is one great feature of the current Propeller, and I would have to agree (at least for my product lines.) I do realize that this will limit the number of I/O pins you can put in a chip, but ability to easily breadboard a new design is pretty important to people learning a new controller.
It would be non-standard but I wonder why someone has not made chips in the 40pin DIP style that have extra rows of pins?
You could have 7 rows of 20 beneath the chip for 140 pins total. I know that it would need custom DIP sockets but surely some
Chinese manufacturer could whip some up in quantity for next to nothing each. The thing would drop right into the holes on a
standard breadboard.... a custom solderless breadboard would need to be designed for it though, but that is do-able and if the chip
design caught on they would soon be the same price as ordinary solderless breadboards.
Trace routing should still be OK for having custom boards made since BGA chips also have many pins (balls)
in close proximity in a grid.
I know it's a lot of pins, but just look at the forest of pins on the bottom of a quad core cpu.
140 pins and no sm soldering.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
If they can make a board to breakit out for a breadboard then I can still use it to test design but it's just plain getting hard to find lots of support chips in DIP format anyway. I keep finding great sounding ICs but there is no DIP version of it and thus I'm SOL until I get the ability to make PCBs in the garage.
Educating the young is very important. But what about the elder ones, with wobbly hands and eyes (getting there). The Prop2 will have 100, or so pins in a quad package. Ok as long as there is a moduled version, breaking out into a managable pitch. Otherwise it would make it a one shot device, if it got put down ok once, it wouldn't relish the second, the third, ..
A bit big for any production wishes, I know, but the "bread board" could be used until the microscopes are got out
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
Those 40pin boards with the sm chips on them are nice but always so much more expensive! :-(
What we need are really low-cost plastic sockets that you can drop a sm processor into and snap down a cover to lock the tiny pins
down tight against contacts that are wired to rows of pins sized to drop into a breadboard.
If the prop 2 had a socket of this sort available for 2-3 dollars many more chips would be sold.
Sockets for chips like ARM9 are more expensive than a pile of the actual chips :-(
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
Granz said...
the DIP format is one great feature of the current Propeller, and I would have to agree (at least for my product lines.) I do realize that this will limit the number of I/O pins you can put in a chip, but ability to easily breadboard a new design is pretty important to people learning a new controller.
It would be non-standard but I wonder why someone has not made chips in the 40pin DIP style that have extra rows of pins?
You could have 7 rows of 20 beneath the chip for 140 pins total. I know that it would need custom DIP sockets but surely some
Chinese manufacturer could whip some up in quantity for next to nothing each. The thing would drop right into the holes on a
standard breadboard.... a custom solderless breadboard would need to be designed for it though, but that is do-able and if the chip
design caught on they would soon be the same price as ordinary solderless breadboards.
Trace routing should still be OK for having custom boards made since BGA chips also have many pins (balls)
in close proximity in a grid.
I know it's a lot of pins, but just look at the forest of pins on the bottom of a quad core cpu.
140 pins and no sm soldering.
A package like that would be *very* expensive!
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Toby Seckshund said...
Educating the young is very important. But what about the elder ones, with wobbly hands and eyes (getting there). The Prop2 will have 100, or so pins in a quad package. Ok as long as there is a moduled version, breaking out into a managable pitch. Otherwise it would make it a one shot device, if it got put down ok once, it wouldn't relish the second, the third, ..
A bit big for any production wishes, I know, but the "bread board" could be used until the microscopes are got out
I'm 67 and don't have any problems making PCBs for fine-pitch devices at home and mounting the chips. I actually prefer them to DIL devices - less holes to drill on the PCB.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Probably, but I think only because it would be a new design manufacturers would have to tool up for.
I don't think there is any reason why it would be much more expensive if it came into common usage.
Actually the best design I can think of for a chip that could be easily adapted to sockets that could be cheaply made
is a ball grid chip. A square hole in a plastic socket with rows of pins at the bottom that extend out the rear.
Drop the chip in and snap in a cap to press it down against the pins... the pin contacts could fan out to standard
breadboard dimensions or whatever.
Manufactured in quantity how expensive could such a simple socket be...you can get 40pin zif sockets delivered from
China for 1.25ea and that is a much more involved contraption.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
Incorporating pins like that in a package will be very expensive. It's one reason why PGA (pin grid array) packages were dropped for MCUs, although they are still used on expensive PC CPUs.
BGA sockets are made for testing purposes, but they are extremely expensive.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Groan, processors of all types are just ending up with too many pins!
Time to create processors with just 2 pins!
Positive V and Gnd pins.
Handle all the I/O you need with a single optical I/O port.
Add however many optically coupled I/O chips you need...a single simple one
or many huge ones fanning out to thousands of hardware I/O pins if you need them.
Basically a processor with however many pins you need and none you don't.
Very versatile processors that can be mounted any which way on a board since
there are just power pins to deal with...this would make board design much easier I think. Fiber optic cable is easy to use.
Might even be possible in many designs to just let the chips communicate by being in close proximity
without needing optical cable.
Optical I/O is very fast.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
Comments
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are now in stock. Only $30. PCB available for $5
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
If I understod Chip corect Logical COGs will have same speed as Physical
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Nothing is impossible, there are only different degrees of difficulty.
For every stupid question there is at least one intelligent answer.
Don't guess - ask instead.
If you don't ask you won't know.
If your gonna construct something, make it·as simple as·possible yet as versatile as posible.
Sapieha
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
propmod_us and propmod_1x1 are now in stock. Only $30. PCB available for $5
Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
It pretty clear, from posts and third party products like the hydra 512 xtreme, that there is a strong demand for more memory.
There are tons of posts where someone is trying to add more memory without sacrificing too many gpio or speed. That's a pretty good sign of user needs.
Also, it's understandable that there is a desire to get away from windows (get a mac or use Linux). But it's not clear that commiting prop resources to puting the develop environment inside the chip is a good idea.
The whole team I work with would probably run away as far as possible from
the chip that had the vxworks development environment built into it. It would also prevent it from being used in critical systems, not that the prop is in this market, because development tools can't reside in them by many regulations. If it was a good idea we would probably see that more often.
Just curious - have the Prop II's specs been frozen yet?
thanks
- H
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
There already is a usable development environment that can run standalone on the Propeller I (Sphinx). It could use a better editor and some I/O driver work, but it runs and can be used to maintain itself (compile the compiler and I/O drivers).
There should not be a problem with critical systems and regulations unless the regulations are written very poorly. The fact that a development system is stored in masked ROM on the chip doesn't mean that it's usable without specific attached hardware and the memory that stores the "user" program doesn't have to be modifiable (it can be write protected).
Howard,
It's a bad idea. There was a long discussion earlier in this thread about such a scheme and it would really mess up the determinacy of the current Propeller design which is a deliberate design decision / advantage.
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
1986 Chevy EL Camino·· No prop yet
1984 Suzukie GS1100GK No prop yet
512kb eeprom on the chip to get rid of the external eeprom.
64kb ram for each cog
Maybe one much more capable video generator for all cogs to have access to
instead of or in addition to the ones now in each cog....full HD maybe.
And why can't a fab process be invented to integrate a xtal on chip or some way to get
precision frequency control without an external xtal? xtals on the board seems sorta last century.
There just has to be a way....maybe approximate freq control that is fine tunable like some
AVRs do with the internal oscillator...only more precise.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
Mike> It's a bad idea. There was a long discussion earlier in this thread about such a scheme
> and it would really mess up the determinacy of the current Propeller design which is a deliberate design decision / advantage.
Mike,
ah, sorry, I may have unwittingly used a silicon-industry jargon word: "frozen" --- I was merely asking if the architectural design of the Prop II was decided and set in stone yet.
Phil> Maybe it takes a wooden stake and a silver bullet to keep this thread buried!
No, Phil, that's not powerful enough ...
it will take a wooden stake, silver bullet, and a Prop II to end this beast ! [noparse]:)[/noparse])
- H
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
At the min, I'm not sure why EEProm on chip is a big deal. Most other MCUs have it. It would be nice to get rid of those parts.
Precision internal oscillator would be nice but I don't think other chips have it now.
Internal USB to replace the FTDI chip would be nice too but probably asking too much there.
A fab process needs to be developed that gives you a standard dip or sm chip but has a set of connectors on top that
connects to all the interesting leads of the silicon wafer inside. It could be a sort of really tiny ball grid array covered by
a sticker that could be peeled off. This would allow a controller chip to be customized by soldering a thin board on top.
Perhaps even laying on a layer of plastic to secure and obscure the added on board. You could add things like optical I/O,
wifi, ethernet, eeprom, ram...whatever. With current methods a lot of stuff can be put onto a minuscule board..just look
inside a small cellphone at those boards!
This grid that could allow layering a board or several atop a chip would be a first move in the direction of 3d chips instead
of current 2d designs...3d chip design is inevitable, might as well start now.
And why not use clear plastic as a covering for chips instead of always black? It would let you add some of those incredibly tiny sm
LEDs inside as diagnostic aids....also IR I/O could be added if the plastic was translucent to IR.
Chips could also be made with tiny socket holes on top so add on chips could just be pressed onto them...easy and so simple!
Why not???
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
-Phil
Chip, I like what I am seeing about the new propeller, although I have not been able to read through the entire 826 messages in this thread (so sorry if this has already been answered.) My focus is on education and training, and as such one of the more important aspects of a new device is its availability in DIP form - above 40 pin will still work, but PLCC and other packages like that will very much lower the usability of the device for my products (unless they are available in DIP format also, like the current Propeller). I have seen many places where people have mentioned that the DIP format is one great feature of the current Propeller, and I would have to agree (at least for my product lines.) I do realize that this will limit the number of I/O pins you can put in a chip, but ability to easily breadboard a new design is pretty important to people learning a new controller.
Thanks for your hard work on these great products.
Art
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
It would be non-standard but I wonder why someone has not made chips in the 40pin DIP style that have extra rows of pins?
You could have 7 rows of 20 beneath the chip for 140 pins total. I know that it would need custom DIP sockets but surely some
Chinese manufacturer could whip some up in quantity for next to nothing each. The thing would drop right into the holes on a
standard breadboard.... a custom solderless breadboard would need to be designed for it though, but that is do-able and if the chip
design caught on they would soon be the same price as ordinary solderless breadboards.
Trace routing should still be OK for having custom boards made since BGA chips also have many pins (balls)
in close proximity in a grid.
I know it's a lot of pins, but just look at the forest of pins on the bottom of a quad core cpu.
140 pins and no sm soldering.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
A bit big for any production wishes, I know, but the "bread board" could be used until the microscopes are got out
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
Doh!
I should have thought of that
They would have to mask off the silicon by spraying it with opaque paint or something to keep the light away from
sensitive areas.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
In the webinar Chip talks about a 44 pin dip for the Prop II. When did that get ruled out?
Rich H
What we need are really low-cost plastic sockets that you can drop a sm processor into and snap down a cover to lock the tiny pins
down tight against contacts that are wired to rows of pins sized to drop into a breadboard.
If the prop 2 had a socket of this sort available for 2-3 dollars many more chips would be sold.
Sockets for chips like ARM9 are more expensive than a pile of the actual chips :-(
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
It's only gonna have 44 pins?!?!
Seems like many more I/O pins should be added.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
He has ruled it out on this forum, IIRC.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
A package like that would be *very* expensive!
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
I'm 67 and don't have any problems making PCBs for fine-pitch devices at home and mounting the chips. I actually prefer them to DIL devices - less holes to drill on the PCB.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Probably, but I think only because it would be a new design manufacturers would have to tool up for.
I don't think there is any reason why it would be much more expensive if it came into common usage.
Actually the best design I can think of for a chip that could be easily adapted to sockets that could be cheaply made
is a ball grid chip. A square hole in a plastic socket with rows of pins at the bottom that extend out the rear.
Drop the chip in and snap in a cap to press it down against the pins... the pin contacts could fan out to standard
breadboard dimensions or whatever.
Manufactured in quantity how expensive could such a simple socket be...you can get 40pin zif sockets delivered from
China for 1.25ea and that is a much more involved contraption.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -
BGA sockets are made for testing purposes, but they are extremely expensive.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/17/2009 10:28:49 PM GMT
July 2008 - (page 30 of this thread) Mike Green says no Dip.
March of 2009 - (webinar) Chip indicates that there will be.
August 2009 - ???
Rich H
Time to create processors with just 2 pins!
Positive V and Gnd pins.
Handle all the I/O you need with a single optical I/O port.
Add however many optically coupled I/O chips you need...a single simple one
or many huge ones fanning out to thousands of hardware I/O pins if you need them.
Basically a processor with however many pins you need and none you don't.
Very versatile processors that can be mounted any which way on a board since
there are just power pins to deal with...this would make board design much easier I think. Fiber optic cable is easy to use.
Might even be possible in many designs to just let the chips communicate by being in close proximity
without needing optical cable.
Optical I/O is very fast.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- Some mornings I wake up cranky.....but usually I just let him sleep in -