JPMorgan Rolls Out FPGA Supercomputer 194
An anonymous reader writes "As heterogeneous computing starts to take off, JP Morgan have revealed they are using an FPGA based supercomputer to process risk on their credit portfolio. 'Prior to the implementation, JP Morgan would take eight hours to do a complete risk run, and an hour to run a present value, on its entire book. If anything went wrong with the analysis, there was no time to re-run it. It has now reduced that to about 238 seconds, with an FPGA time of 12 seconds.' Also mentioned is a Stanford talk given in May."
Ah, progress... (Score:5, Funny)
Re: (Score:2)
Re: (Score:2)
you'll be glad to hear I have a 2TF super-computing chip crunching out massive frames-per-second as we read.
Engineers solve problems (Score:5, Insightful)
I would positively love to do something like this. The purpose of an engineer is to solve problems. That's what makes me happy at work. Solving problems. So here you have a very specific problem that required the construction of a custom computer made out of banks of FPGAs. Tell me that's not sexy! Who cares if it's for bankers. That is a damn nifty gadget to work on building.
Imagine building it yourself. Switching networks, Linux on ARM cores peppered here and there coordinating and dumping program code to the FPGA banks, writing the drivers to grab the data once the run is completed...
And at the end of the day a problem solved: What once took all day now takes a couple of minutes.
This would have been a thoroughly nifty machine to work on.
imagine helping JPMorgan destroy the economy (Score:5, Insightful)
there have been articles about the computer guys who were working inside the CDO machines of wall street in 2000-2008 before the whole thing came crashing down.
the only people who could hold their nose and not-care what their work was being used for are complete pscyopaths, who went through some kind of personality-cracking process so that they can act like normal people while they help destroy the planets economy.
Re: (Score:2)
It takes a special kind of pig-headedness to come up with a moral conclusion that doesn't allow for any doubt. The CDO guys could easily have concluded that what they were doing was useful to society. Or they could have decided that they weren't pious enough to come up with a completely solid moral case, just like we all do, all the time, with just about every decision we ever make, and said "well, I can't see any obvious harm" and gotten on with the work.
Re: (Score:2, Interesting)
Who cares if it's for bankers. That is a damn nifty gadget to work on building.
Replace "bankers" with "the Third Reich" and you would no doubt find some IBM engineers in the 1930s thinking exactly the same thing about the computers they were building.
Just because the technology is cool doesn't mean you're absolved of any moral or ethical dilemmas that may arise from working on it.
Speaking of scientists who worked with the Nazis.. (Score:2)
Tom Lehrer's "Wernher von Braun" song, as one satirical example of science being divorced from ethics.
Re: (Score:2)
I guess what I'm saying is that as an engineer solving a problem, especially if you have no advance knowledge of its usage: the ethical/moral dilemma is not yours. It's that of the person putting your end result to use.
Bullshit, ethical/moral dilemmas are always yours. You cannot abdicate responsibility for your own actions, they threw out the "only obeying orders" defence at the Nurenberg trials.
Re: (Score:2)
The German subsidiary's head, Heidinger, er, emotionally identified with his government clients:
""The physician examines the human body and determines whether...all organs are working to the benefit of the entire organism," asserted Heidinger to a crowd of Nazi officials. "We [Dehomag] are very much like the physician, in that we
Re:Engineers solve problems (Score:4, Informative)
Well, why don't you? The beauty if VHDL is that it's really, really basic. The rest is up to you to implement. Come join us in ##vhdl at irc.freenode.net or try fpga4fun.com. $100 should get you an FPGA training board which you can use to learn.
Re: (Score:2)
Re: (Score:2)
If you can't write a remote login shell in COBOL, or an application to organise your libraries, then you cannot program . Yes, I did both, on a platform (WANG VS) for which there was only a COBOL compiler installed. But I had access to all the OS calls through the compiler, so I could program what I wanted.
And I haven't seen anything in the last 10 years after I stopped with COBOL, that surpasses the ease to make proper financial calculations with it.
Re: (Score:2)
I can attest that Verilog is easy and intuitive, since it got me from a total newbie to a project that other people find useful, in a couple of weeks. Of course, some electronics and computing background helped a lot.
My choice of language was partly influenced by the apparent verbosity of VHDL. It's the same reason why I preferred to learn Python instead of Java, for example. However, I also have the impression that VHDL is in some sense lower-level and more strictly defined, so you can have finer contro
Re: (Score:2)
ADA, the DOD's failed computer software language
Ada failed? If you can write SPARK Ada, I can point you at several companies that will hire you at an insane rate for entry level positions. The language has no serious competitors for situations where it's actually important that the software works.
Re: (Score:2)
The pilots control stick is nothing more than a computer entry device, every motion on the stick is interpreted by the software, which then actually tells the engines/ailerons/rudders/flaps what to do.
You wanna do that in C, be my guest...
Re: (Score:2)
I never thought I'd hear beauty and VHDL in the same sentence, unless there was a negative operator in there too.
Re: (Score:2)
I think it was "VHDL = beauty not;"
Of course that generates a compiler error.
Re: (Score:2)
I've been advocating FPGA supercomputers where I work for a while, but the problem I hit is that the people I try to sell the idea to are systems guys who absolutely refuse to learn VHDL or Verilog coding. They want tools to turn their C and Matlab code directly into something they can download to the supercomputer. There's lots of work being done on this, but I've yet to see a turnkey solution because people are still building their own supercomputers.
Re: (Score:2)
The purpose of an engineer is to solve problems.
Engineers solve real-world problems: when they make simplifying assumptions, they try to avoid transparent falsehoods like "a market functions with perfect transparency," and "all players in a market behave rationally." Hell, there's no way to even check the latter statement for truth: destroying a major company to build your relatively minor personal fortune may or may not be rational. If any solid engineering had been invested into the mortgage market of the past decade, the ultra-rich would not have cl
Re: (Score:2)
So here you have a very specific problem that required the construction of a custom computer made out of banks of FPGAs. Tell me that's not sexy!
I do the same thing, but for things IN SPACE!
Still not sexy, sorry.
Re: (Score:2)
The brightest heartless are inspired by the best computers to advance the problems of mankind.
I eagerly await the next big bombing in NYC.
sick (Score:2)
dude that is horrible
Re: (Score:2)
Banks use their r&d for competitive advantage. The chances of them freely releasing their work to help humanity (and their competitors) is not very high.
Re: (Score:2)
So now we can soon have economic swings turning from a hype to a depression in no time at all.
What takes 238 seconds today would have taken a month in old times.
I'm not sure if this is good or bad.
Re: (Score:2)
like how some tech started out with military funding because that's where the money is?
Re: (Score:3)
Mathless software?
That's like salt-free seasoning-salt, right?
More math, faster... (Score:5, Funny)
Re: (Score:2)
It just accidentally moved the decimal point on your account two places to the left.
But it did it hella fast.
Re: (Score:2)
Ok! Ok! I must have, I must have put a decimal point in the wrong place
or something. Shit. I always do that. I always mess up some mundane
detail.
Absolutely! More golfing, less bankrupting! (Score:3)
That will fix our banking system for sure!
I can't agree more. An increase in the number of golf matches and accelerated round-robin tournament configuration would go a long way to keeping those pesky bank officials occupied on the links and safely out of their offices. The Florida Professional Golfers' Association will benefit while also helping protect our nation's assets from executive malfeasance.
...
Wait a minute, what are we talking about here?
Cheers,
Re: (Score:3)
I think it was basically a pun on another meaning of the FPGA acronym.
Re: (Score:2)
Re: (Score:2)
That will fix our banking system for sure!
Banks for sure?
Re: (Score:3)
That will fix our banking system for sure!
GIGO at tremendous speed.
Re: (Score:2)
Paul Ehrlich: “To err is human, but to really foul things up you need a computer.”
The math is real simple (Score:2)
High school stuff.
Debt:
http://en.wikipedia.org/wiki/Exponential_function [wikipedia.org]
52 trillion dollars and counting.
The best and brightest (Score:2)
Weep, USA, weep! (Score:2, Insightful)
This technological breakthrough is an important milestone on the downfall of the mighty USA. When the brightest of a country are engaged into a completely nonproductive activity and wastes important resources (in terms of education, knowleddge and to some extent money) to achieve nothing which benefits the society, it's a high-mark - it's all downhill from here.
Let's come back here after 20 years and see how this comment stands up to time.
Weep, USA.
Re: (Score:2)
It's not just the USA. I periodically get job offers from financial organisations in the UK, offering 5-10 times what I currently make. It's easy to see why it's difficult to recruit people for more productive occupations.
If they didn't require me to relocate to the cesspit known as London, I'd almost be tempted, in spite of the soul-destroying nature of the work.
Re: (Score:2)
And there's nothing wrong wih London if you're earning a lot of money...
Re: (Score:2)
Well, OK, but what field do you work in?
A variety of things. I do some work on HPC compilers that are used, for example, by people mapping the human genome and doing protein folding simulations to search for cures for various diseases. I write books. I'm currently involved with a project to develop the IT infrastructure in Tanzania.
Because if it's anything commercial, it's pure snobbery to differentiate between (legally) making money at A or (legally) making money at B.
Yes, because nothing commercial affects anyone else in the world, and so everything is entirely equivalent. Most of the projects that I work on - including some that I work on for free - have a net positive impact
Re: (Score:2)
Uh, the point of producing things is improving the human condition. In the absence of production we all starve to death in a few weeks. Of course, in such a state we all revert to foraging, which is just very inefficient production without much trade. Then somebody figures out that if you plant crops and I keep out the foxes, then we can both share a lot more food than we'd have if we both just went around eating leaves or whatever. Then we figure out that the division of labor gets really complicated r
Re: (Score:3)
For me, it was pretty straightforward: Money, and opportunity. Software companies here in Australia tend not to gamble so much on fresh graduates, so the majority of opportunities that were attractive financially were in the consulting (Accenture, IBM, Infosys) or Financial space. As a poor student coming out of university, when I'm given the offers of $45k/yr vs. $60k/yr, plus bonuses, it's a pretty easy decision. Work hours tend not to be onerous, workplace conditions good, and the majority are in the mai
hahaha (Score:2)
there are people who live off the difference in those salaries.
what did you buy with that exra fifteen grand? did you need it?
Re: (Score:3)
Not that this justifies it, or that I disagree with you, but here in Australia the cost of living is vastly higher than the united states.
The US and AU dollars are at rough parity, but it goes a hell of a lot further in the US.
Re: (Score:2)
Re: (Score:2)
It doesn't matter what (s)he buys with that extra $15k... it produces jobs to make the goods that they bought. Granted, those jobs may be Chinese jobs depending on what they are buying, but that's not for you to decide for the person making the money.
Re: (Score:2)
A friend in the industry once remarked that some of the best and brightest in software engineering have been going into the financial industry as of late. It's hard not to wonder what they might have achieved in more productive areas of work...
I agree. Nationalise the so-called financial sector entirely and stop them inventing new derivatives-of-derivatives-of-derivatives, and just to stick to nice simple money going in and out of acounts. Then re-allocate the best minds into fields like medicine and space travel.
Huh? (Score:2)
The company also supports Excel and all different versions of Linux.
Riiiight.
Re: (Score:2)
Not that hard. Excel, front of house. Linux, back of house.
Re: (Score:2)
"All different versions"?
Re: (Score:2)
FYI there's a crazy number of front-office traders running their own "programs" if you call their VBA that. Whatever you think of it, Excel is an easy way to custom-build financial calculations, as well as a simple front-end for displaying information. Usually the guys are too busy to turn their sheet into something more rigorous.
Garbage in, Garbage out (Score:3)
Information asymmetry makes even the fastest analysis on the planet irrelevant as the data input is garbage. It is this lack of transparency which resulted in the housing bubble, etc.
The "too big to fail" banks regularly hide data from customers, regulators, and other branches of their own organization.
This is interesting because of the speedup of FPGAs, but don't be fooled by a second that it addresses an actual business need, other than PR.
Re: (Score:2)
Re: (Score:2)
Information asymmetry makes even the fastest analysis on the planet irrelevant as the data input is garbage.
Even garbage data tends to have real information hidden in it. And most financial data tends to have a quite a bit of real information in it.
Plus, what side of the "information asymmetry" is JP Morgan on? I bet they tend to have the advantage.
Re: (Score:2)
Human stupidity, greed and arrogance resulted in the housing bubble & subsequent crash. Banks such as Goldman Sachs and Deutsche Bank who had clever people that saw the whole thing coming got out while the going was good and made money on the way down - powered by intelligence, greed and arrogance. Transparency had nothing to do with it.
Deutsche and Goldman almost bit the dust (Score:4, Insightful)
if Morgan Stanley hadn't got bailed out by a Japanese bank, and if Bank of America hadn't bought Merrill, then Merrill would have failed, the Morgan, and Goldman and JPM would have fallen because of it.
Goldman's credit default swap business with AIG was also basically 100% bailed out by the taxpayer. Goldman would have lost massive amounts of money if it hadn't been for the deal the government gave them when it took over AIG.
.
A decade ago called... (Score:2)
...starbridge wants their concept back.
Hardware is faster than softare (Score:2)
This is a good idea. A hardware [austinchronicle.com] implementation of a risk analysis algorithm is always faster than software.
What's next for the FPGA supercomputer? (Score:2)
Reprogram this risk analysis computer into a Bitcoin miner and finish the remainder of the 26M BTC?
Re: (Score:2)
There are currently 14.2 million BTC left, which are worth $203 million USD in today's market values. JPmorgan made $17.37 billion last year. You do the math.
Re: (Score:2)
I wouldn't be surprised if someone already has, to be honest.
Re: (Score:3)
1 Virtex-6 SX475T could give you about 1 billion SHA-256 hashes/second clocked at 200MHz., will use 20% the power of the ATI GPU. but will cost about 4 times as much.
Re: (Score:2)
I don't think the FPGA's are as fast as the ATI GPU's for computing SHA hashes
Correct.
but they probably use a heck of a lot less electricity.
Correct.
But unfortunately their initial cost is much higher than ATI GPUs which is why no one started mass mining with FPGAs yet.
dollars vs increadibly intricate magic beans (Score:2)
Have they got the right algorithm though? (Score:2)
Re: (Score:2)
Who says they didn't see the collapse coming and knew the government would bail them out at the expense of the taxpayer?
Re: (Score:2)
Comment removed (Score:3)
Maxeler's partially owned by JP Morgan (Score:2)
Nifty, but remember: (Score:2)
A purpose built machine will often vastly outperform a general purpose computer.
The nice thing about an FPGA machine is that creating the purpose built machine is much faster.
Risk and "Risk" (Score:2)
Re: (Score:2)
The result of this will be that they'll get a number summing up their credit risk in 5 minutes instead of 8 hours. If there's anything way out of whack, they'll be able to react during the day, rather than having to react effectively 24 hours later.
From the article.... (Score:2)
"For the new Maxeler system, it flattened the C++ code down to a Java code."
???
Re: (Score:2)
By "flattened" it meant "bloated through interpretation and unnecessary recompilation".
C++ to Java? (Score:3)
For the new Maxeler system, it flattened the C++ code down to a Java code.
I hope to God that's a typo. C++ -> Java -> Java Bytecode -> Native code almost sounds like a programming language Rube Goldberg machine.
How would the pointer operations even translate?
Re:C++ to Java? (Score:5, Informative)
That's not how it works. The Maxeler FPGA compiler is actually a Java library for describing dataflow machines. They call it "stream computing" - basically just very deep pipelines. You describe your algorithm in the dataflow/streaming style, using the Maxeler Java library, then when you run your program the output is an HDL design to load onto the FPGAs. There's no automatic conversion from a serial program (in C++ or anything else) to a dataflow/streaming program. The developer reimplements the algorithm manually.
Re: (Score:2)
Re: (Score:2)
Java references are implemented with pointers.
FTFY. They are actually closer to C++ smart pointers in that you can't do crazy pointer arithmetic with them, but they will do garbage collection.
This means that while you certainly could implement Java references using pointers (and other stuff like GC), and I doubt there are any serious implementations using anything else, you can't reasonably implement pointers using Java references.
Baloney (Score:3)
The devs are going to run jobs while the machine is idle to corner the Bitcoin market.
But, wow, from the perspective of getting the Boss to buy awesome hardware for your pet projects - hey, we're not worthy.
Their risk metrics are less than worthless (Score:3)
JP Morgan, along with every other mega-bank, has no idea what is actually on it's balance sheet, and hasn't for 10-20 years.
The Shadow Banking system is too big, too complicated, and too interconnected for any of these risk metrics to mean anything.
JP Morgan did the same business as Goldman Sachs and the others, loading up with CDOs and credit default swaps and CLOs and the rest of it. JPM is portrayed as 'wiser' than the rest in the books and the articles about the crisis, but its not really true. They were less stupid than the stupidest people, that doesnt make them smart.
They got bailed out just like all the other megabanks. Why? Because they had no idea what was on their books. They are running a black box. All the supercomputers in the world cannot make up for a complete and utter lack of transparency. And that is what the world of Credit Default Swaps (invented at JPM no less) are. A gigantic black box. The rest of the Shadow Banking system is the same, and JP Morgan (and the rest of the big banks) are up to their necks in it.
Just because you can calculate lies faster, doesnt mean they aren't lies.
Re: (Score:2)
But that is the whole point of the system. To have a system that can calculate the risks, so you have an excuse. "But the Supercomputer we had in the closet told us the risks were acceptable, and we can prove it with the algorithms we used" "Ah ok, here are another 500$ Billion bailout money".
Re: (Score:2)
When your business is selling lies, plausible deniability is a big part of your business strategy. That unfathomable black box is pure gold.
Because a man with a Supercomputer (Score:2)
With apologies to Don Henley.
Re: (Score:2)
Um wait what? (Score:2)
8 hours (or 28,800 seconds) to 238 seconds!
So it's 121x faster, in reality, not in some theoretical limit.
Either that was some shitty software they were running, or they were trying to do the processing on that spare 286 they had collecting dust in the corner. Seriously, I'd like to know what they were using for comparison.
Re:Pictures of the rig (Score:4, Informative)
Do not click. Some sort of phony coprophile spam.
Re: (Score:2)
Re: (Score:2, Informative)
It seems unlikely that they're doing anything that would prevent them from doing it significantly faster with less power draw on a GPU, and with lower initial hardware costs to boot.
No, FPGAs use significantly less power and provide greater performance than GPUs. The initial capital cost is higher though. Here's an article that gives a bit more detail: http://www.xilinx.com/publications/archives/xcell/issue74/FPGAs-speed-computation-complex-credit-derivatives.pdf [xilinx.com]
Re: (Score:2)
>> No, FPGAs use significantly less power
Depends on the actual hardware being compared. The important metric isn't raw power consumption, but FLOP/watt.
>> and provide greater performance than GPUs
Not if the algorithm can be (and is!) expressed in kernels that can be operated in a massively parallel fashion. Not even close.
However if you just want to run Plain Old Code... then yes the FPGA is going to dominate. Also if you are looking for super low latency, then FPGA is the way to go.
I guess m
Re: (Score:2)
GPUs are much more power hungry compared to FPGA and provide a fraction of the performance.
At the end of the day, GPUs are designed for gaming machines... the whole GPGPU thing is a side show for the graphics market. It's just not optimized in any way for this sort of computation. There's little money to be made building supercomputers compared to selling gaming machines.
However, an FPGA can be completely customized to suit your exact needs, you will make efficient use of the entire chip. It won't be a mere
Re:Pretty stupid approach. (Score:4, Insightful)
sometimes, in some industries, first post IS important.
Re: (Score:2)
I think you're underestimating the bank.
The cost of this solution might have been low enough to warrant the immediate gains in performance.
The lock-in you describe might not exist, as the algorithms and the accelerated bits are a small portion of the entire code-base (but take 99% of the run-time).
It will very likely be the case that the cost of not going with this solution is far far greater than going for it.
Re: (Score:2)
and thrown commodity hardware at the problem. Hardware will improve over time and it becomes an off the shelf improvement, not some horrible one-off solution they're stuck with barring great expense and yet another complete reworking.
... because FPGA's are highly specialised one-off hardware items that are never improved on and will be obsolete within a year?
Re: (Score:2)
Re: (Score:2)
Just because they're FPGAs doesn't mean they can't be upgraded over time. Most FPGAs can be re-programmed many times in the event that they improve upon their algorithm. Further, while FPGAs are improved upon at a slower rate, they are improved over time and I would suspect that their design could be ported to newer FPGA versions as they become available without to much trouble - just the expense of buying a new system. And given how much money the banks have, and the value they seem to be placing in this
Re: (Score:2)
No. ALL FPGAs can be reprogrammed. Field-Programmable gate arrays.
FPGAs don't have programming memory, they need an external chip that holds the program, and downloads it to the FPGA on boot. Ignore OP, he doesn't know shit.
Re: (Score:2)
And if you are real good you get to do the books for disaster area.
Re: (Score:2)
For the new Maxeler system, it flattened the C++ code down to a Java code.
Just one Java code? I bet it was '.'