Slashdot Mirror

...hosted by Xilinx even!

http://git.xilinx.com/cgi-bin/gitweb.cgi?p=linux-2.6-xlnx.git;a=summary

> This is a prototype!

But this 'product' makes no sense to me. They admit it is more useful at this point as an FPGA dev kit. But $1500 is a lot to plunk down for an introduction to FPGA develeopment.

This product direct from xilink makes a lot more sense for someone getting started. Ok, it only has 128MB instead of 256M, a single VGA port instead of dual DVI and a smaller FPGA. On the upside though the cheaper board is PCIe instead of PCI-X which is getting hard to find a machine to stick it into. But it is in the same family and when ya actually have a design that won't fit in the smaller part is when you should think about buying a bigger one.

you might be getting ripped off if you're paying $1500 for a Spartan-3 board.

I guess they don't really have the board volume to get low prices. But If you want a graphics card for $1500 that's probably less functional than an NVidia commodity card, I'm not gonna stop you.

OTOH, If you're interested in FPGA programming and a novice at it, you'll want to get a MUCH MUCH MUCH cheaper Spartan board (like 50 to 150). See http://digilentinc.com/ for good starter boards.
If you're serious about FPGA programming (or just willing to pay $1500 to $3000) you will definitely want to get a board with a Virtex or Stratix on board:
http://www.xilinx.com/products/devkits/HW-V5-ML501-UNI-G.htm

If you want to have it on PCIx:
http://www.xilinx.com/products/devkits/HW-V5-ML555-G.htm

You can also get FPGAs socketted for AMD's Hypertransport bus and Intel's FSB:
http://xtremedatainc.com/ (Altera FPGAs)
http://drccomputer.com/ (Xilinx FPGAs)
http://nallatech.com/
http://celoxica.com/

(some of these vendors also sell PCI solutions)

FPGA programming environments still mostly suck. it's a market impeded by proprietary standards and a whole lot of NP-Hard algorithms. We're working on it...

you might be getting ripped off if you're paying $1500 for a Spartan-3 board.

I guess they don't really have the board volume to get low prices. But If you want a graphics card for $1500 that's probably less functional than an NVidia commodity card, I'm not gonna stop you.

OTOH, If you're interested in FPGA programming and a novice at it, you'll want to get a MUCH MUCH MUCH cheaper Spartan board (like 50 to 150). See http://digilentinc.com/ for good starter boards.
If you're serious about FPGA programming (or just willing to pay $1500 to $3000) you will definitely want to get a board with a Virtex or Stratix on board:
http://www.xilinx.com/products/devkits/HW-V5-ML501-UNI-G.htm

If you want to have it on PCIx:
http://www.xilinx.com/products/devkits/HW-V5-ML555-G.htm

You can also get FPGAs socketted for AMD's Hypertransport bus and Intel's FSB:
http://xtremedatainc.com/ (Altera FPGAs)
http://drccomputer.com/ (Xilinx FPGAs)
http://nallatech.com/
http://celoxica.com/

(some of these vendors also sell PCI solutions)

FPGA programming environments still mostly suck. it's a market impeded by proprietary standards and a whole lot of NP-Hard algorithms. We're working on it...

Here we go, this should help you make a homebrew device of your choice way more powerful than what you did in 80s. Given that the price of the kit is fairly low, there is nothing stopping you from selling programmed devices as accessories for commercial products.

Current FPGAs seem to be running about 20-25% of a standard CPU's performance for clock rate (so figure possibly half that for "real" performance).

I don't know about you, but as long as it's not blocking the CPU (and with 8+ cores, even if it blocks one core) it should be plenty fast for most applications. What's a problem is the turnaround time between different usages.

First of all, Forth (unlike PERL) is not an acronym, so it shouldn't be written in caps. Second, it's much larger than either Pax or UnLambda. Even the versions for special Forth hardware are still considerably larger than either one. UnLambda has around a half dozen operations and only one data type (the function).

Forth and assembly language are both fairly low level, but neither is (usually) particularly minimal. Obviously some assembly languages are more elaborate than others, but about the simplest of which I'm aware is the Xilinx PicoBlaze, which still has 57 instructions and 5 data types. A more typical RISC (e.g. ARM, PowerPC) supports hundreds of instructions and a dozen or more data types.

In fairness I should add that the lack of minimalism in both cases is because they're intended to be useful, which Pax and UnLambda really aren't. Granted, both (along with dozens more like them, going all the way back to Intercal) are somewhat interesting, especially in how they manage to be minimal yet (Turing) complete. Nonetheless, none of them is really intended for practical use at all. Some are simply experiments while others are explicitly intended to be the as awful as possible.

If you want something with DVI-in that could potentially be used to re-encode digital output, you can look at Xilinx's VIODC add-on module for the ML40x development platforms... http://www.xilinx.com/xlnx/xebiz/designResources/i p_product_details.jsp?key=HW-V4SX35-VIDEO-SK-US - it supports nearly all common video sources (other than HDMI) but only up to 720p. Actually, the link is for the ML402+VIODC bundle since the stand-alone VIODC card is (conveniently?) down. Of course, you need to supply your own application-specific firmware.

Ripping from HDMI/DVI/etc. is a little dumb: it needs insane amounts of storage, insane amounts of processing power for (near-)real-time re-encoding to make said storage requirements manageable, the re-encoded content can only be as good as the initial decoding and the re-encoding(s) will add its/their own lot of extra noise/blurriness/artifacts/etc.

If printer manufacturers get to that point it then becomes economical to buy a uClinux dev kit and hack up your own god damned controller that'll take a goddammed tank of the ink of your choice.

http://www.xilinx.com/products/boards/s3e1600e/ref erence_designs.htm

Hell with uClinux you could also have one hell of a net enabled printer.

>Look at that board... it uses "SmartMedia" yeah... that means that:
>
>1. This is OLD research
>2. The board developers didn't have a clue
>3. A very old development board is being used.

It's probably storing the Xilinx FPGA configuration bitstream on that card. Why is this a bad thing? What would you suggest they use instead?

http://www.xilinx.com/products/silicon_solutions/p roms/system_ace/

They want to remain in control of the platform, if people use mac or linux as their main os and use Windows to run one of those not-yet-supported programs the power of Microsoft wil start to degrade...

Xilinx's free Picoblaze core (source available in VHDL or Verilog, so you can learn about minimal CPU design) is a lot of fun and is perfect for interfacing little bits of digital hardware to a computer. The Spartan-3E $150 starter kit has a ton of hardware (LEDs, buttons, LCD display, VGA out, serial ports, ethernet, flash memory, ADCs, DACs, etc) and everything you need to get started. Use the free (as in beer) WebPack software from Xilinx, which you can get to work under Linux. Free (as in speech) Picoblaze assemblers and simulators are also available.

Xilinx's free Picoblaze core (source available in VHDL or Verilog, so you can learn about minimal CPU design) is a lot of fun and is perfect for interfacing little bits of digital hardware to a computer. The Spartan-3E $150 starter kit has a ton of hardware (LEDs, buttons, LCD display, VGA out, serial ports, ethernet, flash memory, ADCs, DACs, etc) and everything you need to get started. Use the free (as in beer) WebPack software from Xilinx, which you can get to work under Linux. Free (as in speech) Picoblaze assemblers and simulators are also available.

But in an industrial setting they are quickly replaced by JTAG-connected tools; ChipScope in particular (if you are a Xilinx slave) is great because it captures the signals into the local, very fast RAM, and then sends you the snapshot over a slower JTAG connection. The snapshot is true to what is really happening, and if you design a DDR controller (or faster) then just forget the external wires, they are useless at those speeds. And most of modern commercial designs push the devices to the limit. That's what makes standalone logic analyzers less appealing to a mass manufacturer. Logic analyzers in such conditions become tools of last resort, just like ICEs, where you have to spend a day just preparing your board for testing.

Myself, if I do not have an FPGA in between (and so ChipScope is not an option) then I just use an oscilloscope. I have a 4-channel, inexpensive Infiniium model, and 3 probes is the most I ever needed; staring at the schematic does the rest :-)

Based on my half remembered conversations from 10 years ago, FPGAs are great for prototyping, but not for flight systems, because they are power hogs.

When you measure your power consumption in surface area of solar panel and weight of battery that need to be put on orbit...

With your typical Xilinx FPGA, power consumption would be the least of your problems. Even with unlimited power, single event upset problems would prevent them from being usable in anything you put into orbit. There have been a number of white papers written on this problem, such as from Xilinx and Altera as well as Actel (sort of) [PDF warning].

For the people talking about use only in low-volume items: I recently looked at the innards of a couple of thoroughly mass-market oriented LCD TVs that each included an FPGA. A small FPGA isn't necessarily all that expensive, and can allow things like doing a single physical design that works with either PAL or NTSC, depending purely on how you program one part...

KILL THIS article. FPGA aren't new they were invented in 1984. See http://www.xilinx.com/company/history.htm

Sounds like fairly specialised stuff then, and not the kind of thing that would be run by the vast majority of corporate "grunts". On windows?

All of the specialized apps except one are for Windows. The one (Eagle CAD) is for Linux, but we don't use it any more (it's too simple for our needs.) Well, let me count who is locked in and who isn't. /me counting... Here is the result. 75% of our employees need a program that is available for Windows only. 25% can use Mac with Office and IE (if IE works - required for access to some government web sites, nothing else works.)

Some of our CADs may exist for Linux. For example, Xilinx ISE is available for Linux. But it has limitations; for example, MIG does not work on Linux, and we must have it. I have no idea if Linux native apps will work on Mac, under X or whatever. I don't even plan to try; games with a multi-thousand dollar tools are not something I am interested in, and the tools themselves are so fragile I'd have to be mad to even consider the possibility of trying it on an unsupported platform.

So what's the alternative? Now that Ubuntu is effectively productising Linux I suppose that's a possibility.

As soon as Win32 API for Linux is ready (in other words, WINE is out of alpha and into 1.x releases) then I will personally upgrade all our desktops to Ubuntu (or whatever other distribution is the best at the moment.) However considering that WINE development already took longer than MS needed to write Windows in the first place, I am unsure that I will live long enough to see it released.

I have to agree with you. When I saw the headline of the story, I thought, "Oh, why not an FPGA?", but upon opening and seeing the words "high school", the idea quickly evaporated. It's just too complex for not enough pay off; in order for it to be used with any sort of robotics system would require hours and hours of work by the teacher, and it would fly over most of the kids heads anyway. Even though there are available microcontrollers, such as Picoblaze, I've used the Spartan 3 starter kit myself, in college, and it went over the heads of some professors.

Sure, maybe one or two kids in the class would really enjoy it and just go nuts, but for the others, it would probably be the most boring thing they'd ever seen.

Er sorry, I misread the table here.. but the largest spartan is actually 5 million gates according to
With the smallest Spartan3 starting at 1.4 million.

For those of you who didn't quite follow greenrom's excellent (but rather technical) post, he's basically saying that doing a task in hardware is faster than doing it in software. What FPGAs allow you to do is to create nearly any form of hardware device just by uploading a new design. While you can use this ability to create a new CPU, it's likely to be much slower than a regular CPU. Thus FPGAs are more useful for hardware like network routers, graphics chip research, codecs, and other highly specialized hardware designs.

In fact, a common FPGA design is to have a regular CPU built into the FPGA chip which can then interface with whatever hardware you upload to the reconfigurable portion of the chip. This combination makes for the ultimate microcontroller as you get the performance of an ASIC CPU (a non-reconfigurable silicon chip) combined with the flexibility of a fully reconfigurable FPGA.

For example, here's a PowerPC chip with reconfigurable capabilities:

http://www.xilinx.com/products/silicon_solutions/f pgas/virtex/virtex4/capabilities/powerpc.htm

Again, the market for these chips is very specialized, but the potential uses are pratically limitless. You can basically implement any form of coprocessor you can possibily imagine, as long as it fits inside the available FPGA space.

Anyone who wants a low-volume run of custom chips. For runs up to a few thousand, FPGAs are cheaper than ASICs (and have the advantage of being firmware-upgradable). If you don't need latest-process speed or power efficiency then FPGAs are likely to be good enough. Take a look here for some of the people who use them.

Number one, would we know such an intelligence if we saw it? Furthermore, could we determine what such an intelligence was thinking if we could? I think we could deduce that something was behaving in an intelligent manner, if we applied the right tools to the purpose. The danger would be in not knowing if the object we are studying feels that such testing is against its interests and acts to stop the testing. For example, a theory of emergent behavior within large groups of people (think large bureaucracies or societal constructs) might indicate the possibility of a "group" or "hive" mind arising from interactions between the individuals involved, that is both of the individuals yet outside of it (same as Mind is to Neurons). The output of such a "mind" might seem to be intelligent, but is there any way to actually know what it is thinking, or how it is communicating? Can a neuron ever know of the human, or brain, or mind? What would we (as a human) do if a neuron suddenly could understand? Is it in any way possible such a group mind would act in the same manner? Would we understand it if it did? Can we assume that such a thing isn't happenning already in our increasingly connected and interacting world?

Secondly, I think the other problem with building such an artificial mind is that of design and construction. Interestingly, we likely have both at hand. For design, I subscribe to the view that the mind (or at the very least, the cerebrum) is nothing more than a pattern recording and playback machine, as detailed by Jeff Hawkins in his book On Intelligence. I am pretty certain that this idea is spot-on, and is something that should be investigated much further. As for construction, the design of Dr. Hugo de Garis's CAM-Brain Machine (CBM), as realized by Genobyte, seems to be the approach to use to build a system similar to what is described in On Intelligence. These machines were actually built, shipped, and used in a few research institutions around the world. Whether they still exist or not, or are buried in a back room, is anyone's guess. The fact is that they aren't a standard design for a computer, and furthermore they utilized Xilinx FPGAs that isn't manufactured anymore (whether a similar machine could be built using a different Xilinx FPGA is another matter), leads me to wonder what will happen to these machines as they end their useful lives and/or have hardware failures. Also, it doesn't appear that Genobyte is in business anymore, though their website still maintains "ghostship" status.

Maybe I am reading too much into either of these ideas? Maybe both are a bunch of hooey (indeed, the whole CAM-BRAIN machine thing is something that I am not sure whether to completely believe or not - I seem to remember a /. article a long time ago in which another company linked to this - STARLABS - was seen to be a hoax or something?). Even so, the ideas seem sound, even if the implementations don't exist in fact (although, all the research I have done seems to indicate that these systems do in fact exist).

I have a Xilinx FPGA experimentation/evaluation board. The total cost involved in getting started: $100. I ordered the board itself off of Xilinx's web site, and they offer the synthesis and simulation software (native Linux version included!) as a free download (it's also included in the box). The board comes with a manual, and features a "200,000-gate" FPGA, two serial interfaces (one RS-232 port and a bare connector), a 3-bit VGA interface (I've managed 800x600 so far), a PS/2 port (keyboard or mouse), a number of on-board switches and LEDs, four 7.1-segment displays, and three 40-pin external interfaces. The FPGA can handle internal clock rates in excess of 200 MHz, driven by an external 50MHz crystal oscillator. There's a socket for user-installed crystals as well.

I can't seem to find the original board I ordered, but you can find a much better one (64MB DDR SDRAM! Includes a CPLD on-board! I kind-of wish I'd waited a bit...) for $150 + shipping in their online store. The software can be downloaded free from their website (free registration required).

I have a Xilinx FPGA experimentation/evaluation board. The total cost involved in getting started: $100. I ordered the board itself off of Xilinx's web site, and they offer the synthesis and simulation software (native Linux version included!) as a free download (it's also included in the box). The board comes with a manual, and features a "200,000-gate" FPGA, two serial interfaces (one RS-232 port and a bare connector), a 3-bit VGA interface (I've managed 800x600 so far), a PS/2 port (keyboard or mouse), a number of on-board switches and LEDs, four 7.1-segment displays, and three 40-pin external interfaces. The FPGA can handle internal clock rates in excess of 200 MHz, driven by an external 50MHz crystal oscillator. There's a socket for user-installed crystals as well.

I can't seem to find the original board I ordered, but you can find a much better one (64MB DDR SDRAM! Includes a CPLD on-board! I kind-of wish I'd waited a bit...) for $150 + shipping in their online store. The software can be downloaded free from their website (free registration required).

Hm. There are two issues here and I'm a bit confused regarding which you mean.

My impression is that Open Source does exist to do at least part of this job. I don't know how good it is.

...for a custom application, I'd rather look at FPGA.

Lots of other comments have made clear the point that it's not easy to program this kind of hardware. Typical software programs run in a very sequential manner. In fact, trying to get cooperative parallel execution of threads is known to be a major sticking point in the average programmer's education.

Hardware, on the other hand, is massively parallel. All the "gates" (*) are all running all the time. It's like multi-threading a program, taken to the limit of infinity. However, if designed correctly, this thing can scale beyond belief, since it's all parallel.

It's also important to note that it's a Virtex4 on that card. That's one hell of an FPGA, they sure aren't cutting any corners. I'm not sure which one they're using, but some Virtex4 chips have PowerPC processors at 450 MHz.

This is definitely a niche product for now, due mainly to the lack of people who can write code in Hardware Description Languages (HDLs). But if you can figure it out, and you have an application that works on a massive scale, this may be for you.

Oh, and for all you detractors who are saying "that thing only runs at 500 MHz! How is it supposed to be faster than my 2 GHz AMD chip?" You're forgetting one very important factor. Your AMD chip executes one instruction at a time, and the important instructions are surrounded by instructions whose sole purpose is to control program flow or move data back and forth. However, the XtremeDSP slices of a Virtex4 can each execute a multiply and an add in a single cycle, and there are up to 512 of them in the most hardcore Virtex4 chip, and other logic executing in parallel can control the "program flow" and ferry data back and forth across the bus.

*: Modern FPGAs are actually built out of SRAMs that can implement arbitrary logic functions. They're no longer arrays of gates, so to speak.

The Virtex 4 FPGAs can be clocked at up to 500MHz, so we are talking about ~10-15 times slower than the processor, depending on the application. Even a simple digital filter would be faster when implemented in the FPGA, and this would only take a small fraction of the FPGA resources.

If you can't do it something like a Microchip PIC, then try a Xilinx FPGA.

If you've ever: tried to synthesize FPGA code... Most of that stuff is FREE to corporate customers, companies will voluntarily lose money just to get people to try to use their product. However for people on the street, or companies too small to be "real", they will charge thousands upon thousands of dollars for these materials, if they will let you have them at all.

Xilinx provide the basic ISE synthesis tools[1] for their smaller FPGAs for free, no matter who you are, and for their more powerful tools/larger FPGAs it's ludicrously expensive no matter who you are.

However, corporate users do get support for free, where as hobbyists and students don't. But that seems like good business practice, if you ask me; spend your time with the people most likely to spend lots of money!

[1] Which are buggy as hell when I was using them last year (and I was using the full version of ISE and paying through the nose for it).

Why not make a CPU with a built-in FPGA, then load bits of the kernel into that hardware?

Were you thinking of something a lot different from the Xilinx Virtex 4 FX, Altera Excalibur or Atmel part (referred to elsethread)?

Needless to say, this is not a beginners project by any means. But is it possible for a few hundred dollars and a few hundred hours.

Nonsense! Beginners do it all the time. All you need is $99 to get yourself a Spartan 3 Starter Kit, the free WebPack ISE tools that Xilinx bundles, and a lot of perseverance in learning VHDL, Verilog, JHDL, or some other Hardware Design Language of choice.

As a bonus, the Starter Kit comes with manuals targetted right at newbies to hardware design. It's so easy to understand the hardware layouts of the board that a monkey could do it, and the booklet explaining the history and usage of FGPAs is most informative. Pick up an understanding of Logic Gates from Wikipedia, and you should be ready to get started. (Personally, I think JHDL is the best place to start because it forces you to deal with the actual gates. Once you get that clear in your head VHDL and Verilog become much easier to master.)

A field programmable gate array is a little (fairly) inexpensive chip with hundreds of thousands of gates that can be programmed into lots of different types of hardware, and reprogrammed at your convience.

I've worked with stuff from Xilinx and it's pretty impressive.

The other bonus to this is that you can take the Verilog or VHDL langauge (used to write hardware) and simulate it with great accuracy.

You can license ARM and PowerPC cores -- but they will probably get a bit cheaper if this one is available for free.

Right now Xilinx and Altera make user-configurable FPGA processors. Most of the processor is fixed, but you can encode what happens for special instructions. Here's one: http://www.xilinx.com/products/silicon_solutions/f pgas/virtex/virtex_ii_pro_fpgas/capabilities/power pc.htm

Now if Sun is giving away the processor, there's no reason for you to pay more for a PowerPC-based design -- someone will make a "cheapo" FPGA-extendable UltraSPARC.

You know what I think? I think you should actually TRY designing a CPU. You may not find things to be as cut and dry as you think. Here's some great tools to get you started. Hope you have a good grasp on the parallelism of circuit design, and understand microcode, instruction decoders, on-chip cache, and pipelining. When you're ready to put your masterpiece in fabric, go grab an FPGA and check to see if you made efficient enough use of silicon to make it fit.

When you're done, show us your design and we'll show you how the existing stuff has been done better.

I hate to break it to you, but in the "real world", an inexpensive dev board can become a product unto itself. Take the Spartan 3 Dev Kit for example. It costs $99. (Which is actually incredibly cheap for a dev board.) Xilinx probably moves thousands of these kits, making the venture actually profitable. While many of their customers may be hobbyists, those hobbyists will remember the inexpensive Xilinx solutions and recommend those for their day job. Even if they do it only as amatuers and never expect to go into the field, they *still* generate buzz about Xilinx products. And buzz == free advertising. Advertising == Product Awareness. Product Awareness == $$$.

Let me put it this way. This kit was just featured on Slashdot, a site with hundreds of thousands of members, and probably MANY more non-members. Were this board affordable (i.e. $300-$500), they'd already be moving hundreds of them from this story alone. At $1000-$1500, they'd probably still move a hundred or so boards. At $3000 everyone is going to say, "The technology is cool, but it's too expensive for the moment." and move on.

The reasons the PPC did not become the x86 killer it was predicted to be are largely business-based rather than grounded purely on technical merit. The PPC is superior in a number of areas like SIMD however, IBM did not market it to Microsoft in the same way that Intel pushed the x86 for Windows. IBM placed its bets instead on the lesser(but very capable) players, Apple and Linux, who did not achieve enough market share.

However, the PPC is long from dead. IBM has been strongly cementing the PPC as the ideal platform for embedded computing in an effort to displace other commonly used processors such as MIPS and ARM. The biggest example of this is that both the PS3 and Xbox 360 will feature what are essentially PPC upgrades. Another good example is the embedding of PPCs in Xilinx FPGAs to provide a complete SoC solution.

The Field-programmable Port Extender (FPX) project.

The Field-programmable Port Extender (FPX) is an open platform that augments a network with reprogrammable hardware. It enables new data-processing hardware to be rapidly developed, prototyped, and deployed over the Internet. A diagram of the FPX combined with the Washington University Gigabit Switch (WUGS) is shown above. This enhanced system enables research, development, and implementation of new hardware-based networking applications, intelligent packet processing, custom data processing, and real-time systems.

Xilinx Virtex 2 pro FPGA with gigE support. http://www.xilinx.com/xlnx/xil_prodcat_landingpage .jsp?title=Virtex-II+Pro+FPGAs

for those who don't know what FPGAs are: very basically, it's a chip with many logic elements (mostly look-up tables). You can configure the contents of the look-up tables and how they connect to each other and other specialist circuits on the chip. The result is that you can configure pretty much any digital ciruit you can design. Obviously the complexity is limited by the chip in use.

The one I linked above has up to 99000 logic elements: capable of simulating millions of gates.

The important thing is that while it's programmable, it's still hardware. It all runs in parallel as if you built the circuit. It's not firmware. It's definately not software.

Finally, once programmed, the memory need not be connected to anything so it can't be hacked without physically opening the box.

Anyway. Back to my main point: What you want can be done with existing tech but I couldn't find any for sale. This isn't to say they don't exist, it's just that they don't tend to advertise how they work. FPGAs are very popular to networking and DSP so I'm sure an FPGA-based firewall is commercially available.

Sorry for babbling.

I've been looking at getting this kit, or waiting for this kit to become available. There are plenty of Verilog tutorials out on the web. Not sure of any good sites for teaching digital design... you might want to try MIT's OpenCourseWare for 6.111.

I've been looking at getting this kit, or waiting for this kit to become available. There are plenty of Verilog tutorials out on the web. Not sure of any good sites for teaching digital design... you might want to try MIT's OpenCourseWare for 6.111.

Although using FPGAs for reconfigurable computing applications still has a number of drawbacks, utilizing FPGAs for embedded applications is some really cool stuff.

For example, an entire system can be dynamically built right into the FPGA -- including processor, OPB, memory buses, and any other devices such as interrupt controllers, timers, etc. Aside from RAM and Flash, you almost have an entire embedded system built right into a chip.

Earlier this spring I had the opportunity to work on a project that required this very embedded setup. Using the MicroBlaze soft-processor from Xilinx built into the Spartan3 FPGA and only 8MB of SDRAM, I got uCLinux running -- completely tailored to my hardware setup!

I can't tell you how much time and money would have been wasted trying to design and fabricate the same setup on a PCB.

• comp.arch.fpga Google archive of Usenet groups, where people interested in FPGA hang out.
• Opencores A set of free IP cores that can be implemented in FPGAs
• Comprehensive tutorial on FPGA
• A comprehensive list of FPGA CPUs
• A good FPGA tools overview
• FPGAworld news, jobs, forums, demos etc.(http://www.fpgaworld.com)
• FPGA Basics by Ray Andraka
• Fpga4Fun various fpga projects
• FPGA Boards
• AP100 PCI Platform FPGA Development Board
• Information about signal processing on FPGA by RF Engines
• FPGA manufacturers
  • Xilinx Xilinx has traditionally been the FPGA leader. Their general philosophy is to provide all the features possible, at the cost of extra complexity.
  • Altera Altera is the second FPGA heavyweight. Their philosophy is to provide the features that most people want while keeping their devices easy to use.
  • Lattice Lattice's focus is on low-cost, feature-optimized FPGAs and non-volatile, flash-based FPGAs.
  • Actel (http://www.actel.com/) and QuickLogic have antifuse (programmable-only-once) products.
  • Cypress
  • Atmel
  • Debian FPGA.

I can think of one use right off the top of my head. Anyone remember the console design I suggested? Well, if these chips are cheap enough, it may actually make sense to go back to cartriges! Which means that copious quantities of graphics (including videos and prerecorded music) could be used in games for an inexpensive console system!

Anyone else have any good ideas for this chip?

P.S. Definition of an antifuse. Usually the type of thing you only learn about when you're playing with FPGAs, ASICs, and CPLDs. (The "history of programmable hardware" book that comes with Xilinx's Starter kit gives a good overview of the different technologies including antifuse chips.)

P.P.S. If I'm doing my math right, 1-GBit of memory is ~119 megabytes. 128 megabytes if you're calculating 1-GBit == 2^30.

They should try the Spartan3Es. Much cheaper, with support for cheap PROMs. 1.2M gates for $9 isn't bad at all. Heck, if they call and let Xilinx know that their vision is to have every Linux server use a video card with a Xilinx on it, Xilinx might design it for them for free. And a sale price point of $99 is easily achievable - after all, the Spartan3E Starter Kit is only $149 and has a lot more stuff!

They should try the Spartan3Es. Much cheaper, with support for cheap PROMs. 1.2M gates for $9 isn't bad at all. Heck, if they call and let Xilinx know that their vision is to have every Linux server use a video card with a Xilinx on it, Xilinx might design it for them for free. And a sale price point of $99 is easily achievable - after all, the Spartan3E Starter Kit is only $149 and has a lot more stuff!

Do you use only existing transistors and processing chips, or do you design your own and then have them fabricate a run of those chip designs before having those installed in your board designs?

Actually, you just use an FPGA. They're completely programmable processors that are very similar in design to static RAM. They can be reprogrammed on the fly, and can represent any chip desired. (Limited only by the number of logic units.) They used to be used only for prototyping due to high cost and low speed, but today they are very competitive on the market. Many a manufacturer has taken to shipping the FPGA instead of paying for the manufacture of a custom chip (usually an ASIC).

You should go purchase an FPGA board and see all of the fun stuff you can program it to do! :-)

GPUs are not complex? Then why do we only have a very small number of companies making them? And, what tricky work is done in software? Shading? Bump mapping? All of the big functions are performed in hardware.

A 3S200 is not that small of a chip.

It is a small chip when you're talking about GPUs. Xilinx states that it contains 200,000 system gates. If you have ever worked with FPGAs, you'll know that typically only a max of 75% of the resources can be used if you would like to be able to route your FPGA and still maintain decent clock speeds. This leaves around 150,000 gates. At an average of 4 transistors/gate, this is equivalent to ~600,000 transistors. Compare this with the latest offering from NVidia and ATI, which are pushing the 300 million transistor mark. So, you need 500 FPGAs to get the equivalent resources (at a reduced horse power).

GPUs can NOT be programmed onto FPGAs. At least, not in an economically feasible fashion.