Kevin Morris over at FPGA Journal wrote a really good entry about the evolution of FPGA design and how the tools always target the intended market--even if they don't know the underlying execution engine is a reconfigurable array.
So if you want to make a product for the finance dudes, then you're building a spreadsheet--conveniently familiar reconfigurable array. The Accelerator thinks quantitative finance is a bunk field for Accelerated Computing enthusiasts. I say regardless of the underlying economic/ethical question he raises, having an arbitrary market for low-latency computation isn't necessarily bad. If some combination of low latency LAPACK routines lets me print money for my own biochemical endeavors then I'm willing to write them (incidentally, the 18.330 professor who taught me some of those LAPACK routines now works at DE Shaw).
We setup a public SVN to store our spreadsheet tools. See the Catalyst Accelerated Computing Trac page. Anonymous check-in enabled if you want to add a spreadsheet that does something nifty. I'll update this blog when we have releases.
Subscribe to:
Post Comments (Atom)
1 comment:
Thanks ggreat blog post
Post a Comment