 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
So I know there's a fairly diverse set of people in here so... Anybody
have any experience with using FPGAs? It's something I've often thought
of playing with, but I have no idea (1) how much it costs, and (2) how
difficult it is to make something that works.
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible wrote:
> So I know there's a fairly diverse set of people in here so... Anybody
> have any experience with using FPGAs? It's something I've often thought
> of playing with, but I have no idea (1) how much it costs, and (2) how
> difficult it is to make something that works.
>
just be careful you don't cause it to crack, leak and smoke....
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Mon, 19 May 2008 14:18:07 +0100, Invisible wrote:
> So I know there's a fairly diverse set of people in here so... Anybody
> have any experience with using FPGAs? It's something I've often thought
> of playing with, but I have no idea (1) how much it costs, and (2) how
> difficult it is to make something that works.
I'd start with Google or Wikipedia - both link to some fairly useful
resources. ;-)
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Tom Austin wrote:
> just be careful you don't cause it to crack, leak and smoke....
Thanks - I'll keep that in mind. :-P
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Jim Henderson wrote:
> I'd start with Google or Wikipedia - both link to some fairly useful
> resources. ;-)
Oh, sure, there's information out there. I was just wondering if anybody
around here had practical experience with such things.
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible wrote:
> Jim Henderson wrote:
>
>> I'd start with Google or Wikipedia - both link to some fairly useful
>> resources. ;-)
>
> Oh, sure, there's information out there. I was just wondering if anybody
> around here had practical experience with such things.
I know somebody trying to port SETI@Home to FPGAs. Slow progress.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Tue, 20 May 2008 09:12:12 +0100, Invisible wrote:
> Jim Henderson wrote:
>
>> I'd start with Google or Wikipedia - both link to some fairly useful
>> resources. ;-)
>
> Oh, sure, there's information out there. I was just wondering if anybody
> around here had practical experience with such things.
Ah. :-)
Jim
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible wrote:
> Anybody have any experience with using FPGAs?
...so that'd be a "no" them.
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible wrote:
> So I know there's a fairly diverse set of people in here so... Anybody
> have any experience with using FPGAs? It's something I've often thought
> of playing with, but I have no idea (1) how much it costs, and (2) how
> difficult it is to make something that works.
If you want to fork out $15, I suggest you subscribe to Circuit Cellar's
electronic edition (http://www.circuitcellar.com/). The June edition (not
yet shown on the website, though it is available for download) features a
raytracer implemented in an FPGA. It uses 24-bit fixed-point math, so is
not really useful for assisting POV, but still it's an interesting project
if you're curious about FPGA's. They also use the FPGA to implement a
framebuffer and generate VGA output.
NB you can purchase one-off editions, but that issue isn't yet linked.
-- Chris
PS in answer to your question, FPGA's cost anything from 'very little' to
'too much'. Around here we use some high-end Xilinx devices which IIRC cost
around $600 each - but that series are definitely not hobbyist devices.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Chris Cason wrote:
> If you want to fork out $15, I suggest you subscribe to Circuit Cellar's
> electronic edition (http://www.circuitcellar.com/). The June edition (not
> yet shown on the website, though it is available for download) features a
> raytracer implemented in an FPGA. It uses 24-bit fixed-point math, so is
> not really useful for assisting POV, but still it's an interesting project
> if you're curious about FPGA's. They also use the FPGA to implement a
> framebuffer and generate VGA output.
>
> NB you can purchase one-off editions, but that issue isn't yet linked.
>
> -- Chris
>
> PS in answer to your question, FPGA's cost anything from 'very little' to
> 'too much'. Around here we use some high-end Xilinx devices which IIRC cost
> around $600 each - but that series are definitely not hobbyist devices.
Heh. Chris Cason. Is there anything he can't do? ;-)
I've been to the Xilinx website and looked at the various "development
boards" they offer, and most of them seem to be around $2,000. Obviously
I don't have that kind of money. I'll have to hunt around further to see
if they offer anything at more sane prices.
Also, if memory serves me correctly, the Xilinx software is a free
download; they make their money from selling you the hardware. Maybe I
should start there... although circuit cellar certainly seems worth a
look too.
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Orchid XP v8 wrote:
> Heh. Chris Cason. Is there anything he can't do? ;-)
Get betas out on time?
> I've been to the Xilinx website and looked at the various "development
> boards" they offer, and most of them seem to be around $2,000. Obviously
> I don't have that kind of money. I'll have to hunt around further to see
> if they offer anything at more sane prices.
Hmmm, thought they had some cheap ones. Check out Altera. I seem to recall
they have some small, cheapish (around the $100-$200 mark) boards.
Also, keep an eye out for a CPLD devkit. CPLD's are (simply put) FPGA's
with persistent configuration; i.e. you don't need to re-program them each
time they are powered up.
> Also, if memory serves me correctly, the Xilinx software is a free
> download; they make their money from selling you the hardware. Maybe I
Like Altera they probably have an advanced version of the software that
costs $$$. Back about ten years ago, even the base software version cost
$1000 or so, but then they woke up to the fact that they are a hardware
company and would probably sell more FPGA's if they didn't put barriers in
the way of people designing with them. (This may seem obvious, but to bean
counters, it apparently isn't: e.g. take the fact that Microsoft used to
charge a few grand for the Windows 1.0 and 2.0 SDK's before realising
they'd have more users if it was free).
-- Chris
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Orchid XP v8 wrote:
> I've been to the Xilinx website and looked at the various "development
> boards" they offer, and most of them seem to be around $2,000. Obviously
> I don't have that kind of money. I'll have to hunt around further to see
> if they offer anything at more sane prices.
Try Sparkfun:
http://www.sparkfun.com/commerce/product_info.php?products_id=8458
http://www.sparkfun.com/commerce/product_info.php?products_id=8460
http://www.sparkfun.com/commerce/product_info.php?products_id=8595
-- Chris
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Heh. Chris Cason. Is there anything he can't do? ;-)
>
> Get betas out on time?
OMG, PWN3D!!
[0wned BY YOURSELF! That's gotta be some uber-1337ness or something...]
> Also, keep an eye out for a CPLD devkit. CPLD's are (simply put) FPGA's
> with persistent configuration; i.e. you don't need to re-program them each
> time they are powered up.
OK, that's news. I was under the impression that an FPGA works something
like an EPROM - you use UV to erase it and then program in all the
circuit connections, and thereafter it just works.
>> Also, if memory serves me correctly, the Xilinx software is a free
>> download; they make their money from selling you the hardware.
>
> Like Altera they probably have an advanced version of the software that
> costs $$$.
Ah. Yeah, they probably have a free version, so you can see what it's
like, and then a super version that produces 10% more efficient layouts
or something. ;-)
> Back about ten years ago, even the base software version cost
> $1000 or so, but then they woke up to the fact that they are a hardware
> company and would probably sell more FPGA's if they didn't put barriers in
> the way of people designing with them. (This may seem obvious, but to bean
> counters, it apparently isn't: e.g. take the fact that Microsoft used to
> charge a few grand for the Windows 1.0 and 2.0 SDK's before realising
> they'd have more users if it was free).
Well, we charge people money = more money comes in. It seems like a
simple equation. Until you realise that the software is pretty much a
KILLER loss-leader... [It must surely cost them money to develop the
software of course, but it's fairly useless without the hardware.]
PS. I hear Xilinx bases their synthesis technology on Haskell. I don't
know how true that is...
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> I've been to the Xilinx website and looked at the various "development
>> boards" they offer, and most of them seem to be around $2,000. Obviously
>> I don't have that kind of money. I'll have to hunt around further to see
>> if they offer anything at more sane prices.
>
> Try Sparkfun:
>
> http://www.sparkfun.com/commerce/product_info.php?products_id=8458
> http://www.sparkfun.com/commerce/product_info.php?products_id=8460
> http://www.sparkfun.com/commerce/product_info.php?products_id=8595
That sounds more like it...
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> OK, that's news. I was under the impression that an FPGA works something
> like an EPROM - you use UV to erase it and then program in all the
> circuit connections, and thereafter it just works.
Some are OTP (one time programmable) but most are flash based today.
Atmel makes CPLDs and FPGAs, too.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> OK, that's news. I was under the impression that an FPGA works something
>> like an EPROM - you use UV to erase it and then program in all the
>> circuit connections, and thereafter it just works.
>
> Some are OTP (one time programmable) but most are flash based today.
A PROM is OTP; an EPROM (Erasable PROM) can be reused.
An EPROM usually works by having a little lense over the die surface,
through which you fire UV radiation to cause chemical changes in the
chip material which either 'connects' or 'disconnects' circuit pathways,
thus forming a ROM.
I was under the impression that FPGAs work in the same way, except
instead of opening and closing contacts between data lines and the power
rails to make bit patterns, you actually connect and disconnect blocks
of circuitry to make a custom circuit configuration.
Of course, flash RAM uses chemical changes to permanently store data
too; you just don't need the lense and the UV radiation. But I didn't
think you could use that technology to physically change connections...
> Atmel makes CPLDs and FPGAs, too.
Right, OK.
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible <voi### [at] dev null> wrote:
> >> OK, that's news. I was under the impression that an FPGA works something
> >> like an EPROM - you use UV to erase it and then program in all the
> >> circuit connections, and thereafter it just works.
> >
> > Some are OTP (one time programmable) but most are flash based today.
>
> A PROM is OTP; an EPROM (Erasable PROM) can be reused.
True
>
> An EPROM usually works by having a little lense over the die surface,
> through which you fire UV radiation
Correct
> to cause chemical changes in the
> chip material which either 'connects' or 'disconnects' circuit pathways,
> thus forming a ROM.
Partly right. You erase with UV-radiation and program with a higher than usual
voltage
The programming works by trapping electrons inside an insulated gate of the
field effect transistors used to form the logic gates thus enabling the
connection. The erasing via UV gives the trapped electrons enough energy to
leave the gate and thus disabling the transistor
>
> I was under the impression that FPGAs work in the same way, except
> instead of opening and closing contacts between data lines and the power
> rails to make bit patterns, you actually connect and disconnect blocks
> of circuitry to make a custom circuit configuration.
THat's how it works. You have switch matrices which connect logic blocks and I/O
pins with each other and also configure the logic blocks.
>
> Of course, flash RAM uses chemical changes to permanently store data
> too;
Wrong. Works the same as in eeprom with electricity and trapped electrons
> you just don't need the lense and the UV radiation. But I didn't
> think you could use that technology to physically change connections...
>
you don't physically change connections but control switches in the connections.
> > Atmel makes CPLDs and FPGAs, too.
>
> Right, OK.
>
> --
> http://blog.orphi.me.uk/
> http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Aydan wrote:
>> to cause chemical changes in the
>> chip material which either 'connects' or 'disconnects' circuit pathways,
>> thus forming a ROM.
>
> Partly right. You erase with UV-radiation and program with a higher than usual
> voltage
> The programming works by trapping electrons inside an insulated gate of the
> field effect transistors used to form the logic gates thus enabling the
> connection. The erasing via UV gives the trapped electrons enough energy to
> leave the gate and thus disabling the transistor
Interesting. I was under the impression that you use UV to program it
too - much the same way that you use a laser to burn a CD. (Obviously
you don't actually "burn" anything; you induce a chemical change in the
ink.)
>> Of course, flash RAM uses chemical changes to permanently store data
>> too;
>
> Wrong. Works the same as in eeprom with electricity and trapped electrons
Again, I was under the impression that you use a large current to
chemically change the components such that they now respond to
electricity differently, and that's how you store information.
>> you just don't need the lense and the UV radiation. But I didn't
>> think you could use that technology to physically change connections...
>
> you don't physically change connections but control switches in the connections.
Mmm, OK.
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible <voi### [at] devnull> wrote:
> Again, I was under the impression that you use a large current to
> chemically change the components such that they now respond to
> electricity differently, and that's how you store information.
PROMs work this way. Either by fuse and current or antifuse and voltage.
The change is more physical than chemical though.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible wrote:
> OK, that's news. I was under the impression that an FPGA works something
> like an EPROM - you use UV to erase it and then program in all the
> circuit connections, and thereafter it just works.
nope. FPGA == Field Programmable Gate Array, with the emphasis on 'Field'.
i.e. it has to be re-configured each time it powers on. typically used in
applications where the FPGA (for obvious reasons) is not needed to get the
system booted. (Actually you can attach serial EEPROM's to them to auto-
configure them at startup if you wish). Some newer FPGA's incorporate this
device into the FPGA itself. Configuration is done serially (basically just
a bitstream).
Here's a write-up on them anyhow:
http://en.wikipedia.org/wiki/Field-programmable_gate_array
-- Chris
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Chris Cason wrote:
> nope. FPGA == Field Programmable Gate Array, with the emphasis on 'Field'.
> i.e. it has to be re-configured each time it powers on.
Field-programmable just means it can be reprogrammed "in the field".
Doesn't say how often. ;-) But yeah, OK...
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible wrote:
> So I know there's a fairly diverse set of people in here so... Anybody
> have any experience with using FPGAs? It's something I've often thought
> of playing with, but I have no idea (1) how much it costs, and (2) how
> difficult it is to make something that works.
NB at it appears you are interested in getting into hardware, you might be
interested in looking at the Hammer board:
http://www.tincantools.com/
Particularly their hammer kit, which comes with a very nice prototyping
board. I have one of these and it's easily the best of the various embedded
development kits I've had over the years. My previous platform for personal
projects (up until a few months ago) was the gumstix (gumstix.com) but
while it's OK for plug-and-play, I found doing custom hardware work with it
to be a bit of a pain (particularly due to its small connectors and overall
mechanical setup).
The nice thing about the hammer (apart from the dev board) is the fact it's
got a standard 40-pin DIP form-factor, meaning it's very easy to wire into
custom projects. Of course you lose some flexibility in that you have very
few I/O pins to play with, but currently it's got all I need.
-- Chris
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Chris Cason wrote:
> NB at it appears you are interested in getting into hardware, you might be
> interested in looking at the Hammer board:
>
> http://www.tincantools.com/
>
> Particularly their hammer kit, which comes with a very nice prototyping
> board. I have one of these and it's easily the best of the various embedded
> development kits I've had over the years. My previous platform for personal
> projects (up until a few months ago) was the gumstix (gumstix.com) but
> while it's OK for plug-and-play, I found doing custom hardware work with it
> to be a bit of a pain (particularly due to its small connectors and overall
> mechanical setup).
>
> The nice thing about the hammer (apart from the dev board) is the fact it's
> got a standard 40-pin DIP form-factor, meaning it's very easy to wire into
> custom projects. Of course you lose some flexibility in that you have very
> few I/O pins to play with, but currently it's got all I need.
Mmm, interesting...
I seem to have misplaced it, but *somewhere* I have a box of 7400s and
some breadboard. If you chop up lots of wire and hot A LOT of patience,
you can wire up arbitrary logic devices with it. Latches, counters,
mutiplexers, etc. I always wanted to sit down and wire up an actual
processor. I mean, it's one thing thinking you know roughly how to do
it. But the proof is in DOING it. ;-)
I just thought an FPGA sounds easier than plugging in millions of wires
and then trying to figure out why nothing is working. :-}
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
