Slashdot Mirror
New Supercomputer By Star Bridge
Ronin Developer writes with word of this "interesting article on CNN about a new desktop-size super computer that reconfigures itself on the fly. The company name is 'Star Bridge.' Ring any bells? If I remember correctly, wasn't there something on /. about this a year ago?" Indeedy do -- Star Bridge seems to go straight from wacky-but-cool promises to Where are they now? (and back) with finesse. It's the the Arnold Schwarzenegger movie plot of hardware companies -- simultaneously head-scratchingly implausible, mildly compelling, and numbingly persistent.
That is kind of odd. Didn't their marketeers ever take a composition course? You're supposed to state your position unequivocably without resorting to "I think that..." or "We believe that..." or "In my opinion..."
Maybe it's different when you're talking about an actual product rather than academic theories, but I've never seen other vendors use such wishy-washy language.
Some reality is in order here.
FPGA computing is real and it has been shown to work for some problems. Take a look at TimeLogic. These guys have implemented search algorithms used in the human genome project on FPGAs.
Now let's look the difference between "works" and "price to performance". In the case of TimeLogic they have produced a "stand-alone appliance that end-users do not program (i.e. users do not program the FPGAs.) I beleive the reasons for this is that this programming abstraction (remember this) is not easy to master (i.e. it is not a mainstream programming language.) Nor is the "edit, complie, run cycle" easily reporduced on a desk top. (This time is perhaps the single most limiting factor in software production). So FPGA computing works, but is expensive to implement and program. It does not support cost effective general programing practices that are used today.(i.e. unless you are building specific purpose machine and can justify the software development costs based on a real market, the cost of programming for every day production environments is too expensive)
Which brings me to the main point. The issue is SOFTWARE. It is easy to build a Beowulf with 1000 processors and call it a supercomputer. It is hard (expensive) to write good software for this system. It is easy to string together a bunch of FPGAs and call it a supercomputer. It is hard (expensive) to write software for these things and it is harder (expensive squared) to write parallel software for these things.
In general, there is a huge (I mean really huge) investment in the supercomputer world in programming abstratcions that use FORTRAN (and to some extent C) Side Note:Before all you "FORTRAN is dead language boneheads" start hitting the reply button, remember that there are more than a few 100,000+ line FORTRAN programs that determine everything from airplane wings, to weather, to new drugs, that are not going to go away because you think XML is great way to go. Indeed, the cost of reprogramming these applications is almost an economic impossibility!
So where were we, ah yes, the software thing. My point is that until FPGA systems can take standard supercomputing FORTRAN or C applications and run them "out of the box" and thereby allow the tens of thousands of people who understand this type of programming to use FPGAs easily, they will remain application specific computers (albeit fast) and not realy a mainstream programmable computing devices. This is not to say in the future the FPGA computing will not dominiate (maybe it will), but there is a lot of work to be done on the software side before this will happen.
BTW: I sent the Starbridge guys some simple FORTRAN benchmarks a while ago. I did not receive a response.
Finally, remember this:
The general always eats the specific.
Any one remember a company called Symbolics?
HPC for Primates. Read Cluster Monkey
I have played with the FPGA's from Xilinx and while they are very cool they are slow @ lookups and vectors this is where people doing custom ASIC hardware will be better but most software problems does not require this
VHDL is nice but I always thought that the machine could do a better job if you described the problem better and arent so abstract
this is intresting but you have to understand that this is like software agents to hardware engineers
how many REAL applications of software agents are there ? this is all mangled up in the EXPERT systems design methodolgy and it become a real quagmire to sort out
if Star Bridge Systems keeps focus then they will be alright
all hell will break lose if they dream up fancy problems to solve, KISS is the order of the day
hope they get somewhere
oh and slashdot did run this story but it is nice that it is a story in its own right
regards
john jones
Go to starbridge's site and poke around a bit. The "HAL 300" was enough to make me spew coffee on my monitor I was laughing so hard.
The "faster than the IBM Pacific Blue (when simulating a 4-bit adder)" claims put the nail in their coffin for me. These guys are hucksters of the worst kind.
"Cause there's 40 different shades of black, so many fortresses and ways to attack, so why you complainin'?"
They were incredibly efficient, but Thompson couldn't understand why they worked. (He suspected such things as electromagnetic coupling and communication through the power supply.)
:-)
I think I remember reading about something like that - a guy removed "islands" from the circuit and it stopped working, and when he put them back in it worked. Crazy
Genetic algorithms (in software) tend to be like that. Evolution doesn't value parsimony or maintainability; it only cares about what works. Turns out genetically evolved software desperately needs "junk DNA" (as safe places to recombine bits from two parent algorithms).
On the other hand, much of the same thing seems to be true for the large C++ application I'm working on.-(
Stupid job ads, weird spam, occasional insight at
Does this mean the web will eventually become wholly self referrential and crawl up its' own arse? Oh, hang on.....
Dave
I write a blog now, you should be afraid.
Excerpting from this NASA press release that Slashdot linked to Friday:
And from this article that Slashdot linked to in the same writeup: I'm not trying to be a troll or start a flame war; I just think it's absurd that Slashdot's editors not only don't participate in posting comments (and claim they read them), but that they don't even their own articles!Strange how Slashdot was bought out, and now that our beloved editors are paid hefty sums with full editorial control, they still can't find the time to read their own site. This site was definately better back when it was Rob & Jeff posting stuff that interested them (and that they therefore actually read). It's still an amazing site, just not as amazing =(
-- Imagine how much more advanced our technology would be if we had eight fingers per hand.
so they invent a new supercomputer that's so smart it can reconfigure itself, and what do they name it? HAL.
they just never learn.
sean
--Mike--
From what I recall about parallel computing, the structure and using it are the hard part of parallel computing. So ,the "Viva" library is just as important as the hardware itself.
However, from the look of it, the library just makes parallel processing easy - it doesn't transform serial code into parallel algorithms. This makes porting pretty difficult.
From their web site:
This seems like the opposite of abstraction to me. Of course, I could be wrong
The main question is, when are these things going to be available? Seems like we've been reading about them for quite a while.
Slashdot noted the press release in this Slashdot story. I wouldn't be surprised if CNN found out about the story from Slashdot.
Check it out, you'll see what I mean.
Got Rhinos?
At age 12 Kent built a commercial-quality, space-invaders-type computer game.
Corollary to Moore's Law: The IQ of new computer owners is declining.
I'm sorry. I just noticed that I wrote "serialization" when I meant "parallelization" (which doesn't even come close to sounding like an actual word, but you know what I mean). Sorry for any confusion.
"I believe that a scientist looking at nonscientific problems is just as dumb as the next guy." -Richard Feynman
These things use Field Programmable Gate Arrays (FPGA's) in order to restructure themselves dynamically. This, in and of itself, is not a new concept. FPGA's have been used for years in prototyping or in the first products released. It's much cheaper and easier to reprogram an FPGA if a bug is found than it is to create a new chip design. Once the bugs are gone, FPGA's are replaced by hard-wired silicon in the rest of the line.
Now on to using FPGA's in supercomputers. First of all, an FPGA is slower than a hard-wired chip. These machines pick up speed from the fact that they can use portions of the chip that otherwise would have been on standby. It's super-charged serialization. By restructuring the circuitry for each task, they can take advantage of the majority of the chip at all times. This is not an easy task, and I find it quite impressive. (On a side note, the restructuring is software-controlled.)
When I read this story, I immediately associated it with an article from several years back about Inman Harvey and Adrian Thompson. Thompson was using an FPGA to run genetic algorithms for hardware development. Essentially, make a machine design the chip. He had some very interesting results. The chip designs took advantage of the physical chip rather than just the wiring. They were incredibly efficient, but Thompson couldn't understand why they worked. (He suspected such things as electromagnetic coupling and communication through the power supply.) This is all only moderately related, but it's very interesting, regardless. The article is from June, 1998 and can be found here if anyone is interested.
"I believe that a scientist looking at nonscientific problems is just as dumb as the next guy." -Richard Feynman
This is just laughable. Here's a snippet from the specs of their HAL-300 box, supposedly the superest-duperest computer in the world:
I bet ASCI White wishes it had a S-VHS connection.
"But Dave, I don't like Hollywood.......Dave? Aren't they just asking for trouble here?
I don't know about you, but I sure as heck hope that this bit is someone's April Fool's joke that launched a little early.
really
Check out the Vinny the Vampire comic strip
"It is a greater offense to steal men's labor, than their clothes"
From what I understand, it makes a lot of sense for some problems. Like, hard problems.
;)
Basicly, a FPGA can take up the properties of any chip that can be defined by VHDL or other such languages, with some restrictions, of course. So, theoretically, you use every last square inch of silicon for the problem at hand, minus whatever is there to make it reconfigurable.
So that's nice, because if you are doing floating point problems, you don't really need the integer unit. Things like that.
However, I suspect that the thing will, at least in the short term, be a pain in the arse to program efficently (Given that it's a completely different paradigm) and will probably be for specialized applications that suck on an ordinary computer.
And it isn't something that just one company thought up. It's been in the cookers in the academic part of the world for 5-6 years at least.
I mean, the best part about it is that all of your parts are off-the-shelf and cheap...
Gentoo Sucks
People have tried to use FPGAs for building supercomputers as long as they have been around. So far, they have never proven cost effective. You gain a lot in processing speed by not having to interpret instructions, but you lose a lot in speed and area because of the hooks needed for reprogrammability. It sounds good and attracts a certain brand of investor, but I wouldn't put my money on it. I think you are going to see massively parallel machines with single-package processor/memory combos succeed before FPGAs.