Disqus - Latest Comments for phaze1955

Re: http://futurememes.blogspot.com/2008/12/status-of-life-sciences.html

phaze1955 — Fri, 23 Jan 2009 00:57:14 -0000

Hi Melanie,
I agree with your comment about the individuals contributing data. It is key to the spread of knowledge and the ability of the crowd to understand it and turn it into real advances that benefit humanity.
Today I received the newsletter from pacific bio. I am sure many thousands did also. Imagine , a 30,000 times decrease in the speed of sequencing. So what do we do?
The amount of data and the subsequent compute resources required to analyze, understand, correlate etc this amount of data makes the Swiss/French collider look like a minor experiment.
I would suggest a nonprofit be set up that buys one or more of their boxes and has one Peta flop of crunch tied to the back end to see if it can be run in real time and posted to a national data base. The posting is key so that ip is not horded and that real data is provided to those that need it most.
Personally I am going to build my own 100 teraflop test box so that I can fine tune input and output. One of the issues that sticks with me is the standardization of software based output. Personally I would like the NIH to take the lead on this issue.
The above example is similar to the smart grid issue. How do we sample 100 million nodes in real time and compute the load balance etc. I hope to seek direction/participation and advice this next month in Austin.
Many thanks for your blog. You are so on the right track.
Peace
Tim

Re: http://futurememes.blogspot.com/2008/12/status-of-life-sciences.html

phaze1955 — Sun, 04 Jan 2009 18:50:17 -0000

Hi Lablogga:
It's interesting that you mentioned simtk and pacific bio. They have been on my mind a lot recently. I am most interested in whole genome population work. The problem is standards. How I am supposed to tune a 4 teraflop workstation of a larger 100 teraflop box to really do some positive work with the pacific system ( or other) if there are no standard. The flood of whole genome data will be like nothing we have seen before. And it will grow exponentially. I am inclined to fund these with a try at social capital. I.e. no vc, just normal people taking micro risk. We need to move faster that prior systems allowed, so large scale committed groups will be needed to get projects to the commercial show me stage. I also think they need to be small and fast.
Right now I or others can build 10^16 for under 5 million, heck 100 teraflops can be had for 250k. it's getting to fast to quick.
But that good, it opens the creative mind
Tim

Re: http://futurememes.blogspot.com/2008/12/status-of-life-sciences.html

phaze1955 — Mon, 29 Dec 2008 23:03:48 -0000

I think that real time mass genomic s will have a huge impact. with two 1k genomes looking like there set for 2009 i think we are on the edge of a huge change.
Real cell simulations, less drug trial misses. multimedia content at theater quality at the cost of cartoons. self evolving dialog, long gone actors digitally sunthesized and step over the uncanny valley

Re: Top 10 Computing Trends for 2009

phaze1955 — Mon, 29 Dec 2008 22:58:42 -0000

so we only need 10 of them. and with nice gpu systems already here the 2 *10^16 is only a year or 18 months away.
Real time genomic s on mass numbers. self evolving ai etc.. the singularity is looks a little more like double exponential