I have had great fun as a computer person for astronomical observatories. Have been treated quite well without the PhD, but there is certainly a ceiling without one.
The Square Kilometre Array is the current hotness, and is hiring. http://www.skatelescope.org/people-contacts/jobs/
Seriously, many of these news sites are publishing their content via RSS (News Corp. sites in particular). I assume that they will stop foisting their content on the interwebs, before unleashing the copyright flying monkeys.
Fifteen (!) years ago, I took a UC Extension class on Neural Networks taught by Stanford professor David Stork. He had developed a lip-reading system for communication in noisy environments, such as an airplane-repair facility. If you could do it 15 years ago with workstation-class desktops, I suppose you could do it with a smartphone today.
Terrorists stage a high-profile robbery to incite a police chase, which leads to the use of one of these EMP things, aimed at Mae West. Th' InterTubes go dark, civilization collapses.
FWIW, I've got some high-end Monster cables for speaker cables and some Blue Jeans cables for interconnect. The Blue Jeans cables are great. (Good audio cables isn't hard, just use some nice co-ax.)
About ten years ago I attended a workshop by Stanford professor David Stork. He mentioned some work on a system that was deployed for use by aircraft technicians: the system couldn't read the voice channel with the jet engine blasting away (the techs wear hearing protection). So it read lips. Ten years ago.
Sounds like TFA is talking about doing this in an embedded, consumer-electronics application. Rather than a fixed, industrial-military, hire-computer-scientists-to-maintain-it thing.
Not-so-coincidentally, David Stork is the author of the book, "HAL's Legacy"...
We target Mac OS X as a supported platform because our users demand it.
For our scientific applications, we support Linux and Mac OS X. Someday maybe there will be a Windows version, but I'm not expecting that any time soon.
I think that "-march=pentium-m -msse3" is WRONG, as it will favor x87 instructions rather than use the SIMD vector unit. Result is that the Core benchmarks slower than the Pentium 4. WRONG!
You're right that the Core Duo is based on the Pentium-M microarch, but it's had some major updates done to it. Fex, the SSE front and backend are completely redone. On the P-M, it took twice as long to decode SSE than X87. Core can handle up to three packed and micro-op SSE instructions at once, making using SSE the advantage. However, when you set -march=pentium-m, GCC prefers to generate x87 instructions. There's other changes that make Core more similar to Netburst than P-M when it comes to cost calculation, prefetch block size, etc., all of which are dependent on -march. Check out gcc/config/i386/i386.c and the IA32 Intel Architecture Optimization Reference Manual. -qed
At work use Linux predominately, but Macs are very popular and getting more so, particularly for laptops.
At the moment, a convenient way to move large data sets is to unplug the USB hard disk and move it to your other workstation, rather than pull that data across the network, because:
1) Portable machines which are used half the time "off net" do not lend themselves well to file-server usage. 2) Scientific data-reduction packages typically involve processing of multi-gigabyte data sets.
OS X 10.4 broke the ext2 project for Mac OSX, alas. As other Linux-supported filesystems: * HFS+ -- OK, but it doesn't support the journal. Risky? * FAT32 -- limited. File sizes less than 2GB, and volume sizes are limited, too. * UFS -- Linux doesn't yet support read-write; has endian issues when trying to share between Macs.
I'm hoping that some sort of Linux emulation on the Mac will allow Macs to access Linux data for filesystems that are not yet ported (ext3) or won't be ported anytime soon (Reiserfs).
Without their money many artists probably wouldn't get their albums published, so it's kind of a necessary evil that we have to deal with.
Well, this is wrong. Some friends of mine had to buy their album back from the record label in order to get it published:They ended up spending lunch money that paid the same marketing people who decided their album would not sell. Sure, they paid for the production costs (studio time, engineer time) too, but they were used to paying quite a bit less for that, and the cost of production goes down all the time. These fellas did not need studio musicians, they could have recorded the thing on a Tascam 10000 in their garage, and sold it at concerts. Would have saved everyone a lot of trouble.
What record labels and distributors still provide is a form of peer review. But iTunes' album and artist reviews, along with the simple hyperlinking and collaborative-filtering effect you get from published iMixes, makes iTunes a far more effective method of peer review than the record labels.
I don't think record labels help very much anymore. I want to support the production and distribution of music that I enjoy. So I purchase music on iTunes. But I want the record labels to get out of the way.
So I was a Gentoo zealot for two years, it helped to teach me (more than I wanted to know) about Linux.
I got fed up with power-management issues on my employer-supplied laptop computer (a nice machine, but not Linux-friendly) and purchased a Macintosh PowerBook. Very nice, not as clean as Gentoo, but it got me interested in *BSD.
My server was running Gentoo SeLinux until last week. I've installed FreeBSD 5.2.1 and I am *very* happy with it as a stable, secure server platform.
Linux, Apache, etc. have lent legitimacy to Open Source, and BSD license is attractive to many who cannot otherwise use Open Source. So *BSD is helping spread Open Source, and to otherwise improve the quality of the aggregate code base.
Since Gentoo was developed by someone who liked BSD but wanted the device-driver support of Linux, I feel that most of my skills transfer very quickly. I feel that my learning curve on FreeBSD helps me better understand Mac OS X, which has an installed base of about 12 million computers (if Apple is to be believed).
Well, in our field we have lots of custom software that we use in-house for scientific data-reduction and for equipment monitor/control.
So we use Java for custom applications - GUI apps as well as web services.
James Gosling and some other Sun people came out here to visit, and perhaps got the impression that everyone in science is using Java on Linux as much as we are. Shrug. We like it.
Analysis techniques discussed in this paper are used by the Search for Extra-Terrestrial Intelligence to compute the likelihood that a given signal stream contains semantic content.
But they were not aware of this particular application of the research, and now they are thanks to Slashdot; a SETI researcher emails:
It will be very interesting to read this paper. We had looked at the Shannon entropy of octave music compated to languages, but we were not aware of a Zipf plot of it. Much thanks!
http://www.newegg.com part #N82E16817145316 External firewire enclosure $ 50 each part #N82E16822152011 Samsung 160GB hard drive $105 each
http://www.pccables.com part #70924 FIREWIRE HUB 6-Port $ 35 each
With the six-port hub, we set up a topology such that we have one hard disk per port, minus one port which goes to the host computer. Then we have five disks per hub in the array; if one of them fails, we can pull it off-line without touching the cables for the other drives, since they all go to the hub.
So it is $155 per 160GB element of the array, plus $35 for the hub, amortized over the five disks (35/5=7) is $162 for 160GB.
At RAID 5, we have (N-1) effective storage capacity, or 4*160=640GB for $810.
Since most every firewire host adapter has two ports, we can have two of these hooked up to the computer, for a total of 1.3TB of RAID5 on-line for about a dollar a Gigabyte.
It's a bloooody mess, with twelve power-supply bricks, but it is remarkably cheap for a hot-swappable personal server.
Note that we can scale this topology if we want to; we could implement RAID 10, by adding elements to each of the hub ports in a daisy-chain fasion; FireWire supports up to 63 devices... for a personal server, I don't anticipate saturating the firewire bandwidth. We would implement RAID 10 or other 2-D RAID topologies by means of combining RAID with Logical Volume Manager (LVM) -- a RAID5 array per hub, with each element of the RAID5 array being an LVM expandable volume. You would have to add to the LVM in "slices" of five disks, or otherwise keep the RAID5 elements the same size...
Actually I'm not sure that this last point is meaningful; I would not plan to do such a thing. But it is worth noting that I can dynamically attach other devices to the firewire array by connecting to the spare ports on the back of the disk enclosures.
works with Gentoo i686 glibc-2.2.5 gnome2
on
GCC 3.2 Released
·
· Score: 2, Informative
I've been using GCC 3.2_pre on a Gentoo Linux PIII laptop for about a month. Everything seems to work just fine.
That is, gcc 3.2 is the ONLY gcc on this computer. So the ABI interop issue isn't a problem, I suppose.
If you want to compartmentalise build a small
reasonably verifyable core and run Linux
unpriviledged on it - its one of the uses for
real microkernels (not Mach but something thats
actually -micro-).
I Am Not A Kernel Hacker (IANAKH) but..
Jonathan Shapiro's Ph.D. thesis work impressed me in this regard - a "nanokernel"-based OS that attempts to better support low-level security:
It would be nice to implement a linux-like system on top of this, or perhaps use such a system to provide virtual machines on top of the hardware (like IBM mainframes or VMWare GSX Server).
Jim Gray has been
collabortaing with the astronomical data community for some time.
The spacial-indexing schemes Jim helped develop for Terraserver will be key to performant queries for a Virtual Observatory
Jim Gray was well-known in the database community as the guru of pe rforman ce metrics, long before joining Microsoft
The take-home lesson from the Virtual Observatories conference was that the amount of data required to do science with a "virtual observatory" leads to interesting problems in computer science, problems which are only tractable when analyzed by collaborations between statisticians, computer science people, and the astronomers themselves.
Finally, note that this year's historic increase in the National Science Foundation budget is largely due to the new Information Technology Research Initiative. The need for new methods of data managment in the sciences is real.
Slashdot Mirror
I have had great fun as a computer person for astronomical observatories. Have been treated quite well without the PhD, but there is certainly a ceiling without one.
The Square Kilometre Array is the current hotness, and is hiring.
http://www.skatelescope.org/people-contacts/jobs/
Nice article here explains how to roll your own:
http://www.cnx-software.com/2012/06/13/hardware-packs-for-allwinner-a10-devices-and-easier-method-to-create-a-bootable-ubuntu-12-04-sd-card/
Anyone who thinks this is a silly idea needs to spend two weeks in a city without modern plumbing.
Seriously, many of these news sites are publishing their content via RSS (News Corp. sites in particular). I assume that they will stop foisting their content on the interwebs, before unleashing the copyright flying monkeys.
Fifteen (!) years ago, I took a UC Extension class on Neural Networks taught by Stanford professor David Stork. He had developed a lip-reading system for communication in noisy environments, such as an airplane-repair facility. If you could do it 15 years ago with workstation-class desktops, I suppose you could do it with a smartphone today.
Terrorists stage a high-profile robbery to incite a police chase, which leads to the use of one of these EMP things, aimed at Mae West. Th' InterTubes go dark, civilization collapses.
Slobodkin is GREAT!
http://www.slobodkin.org/books/index.html
The Turning Place
But We Are Not of Earth
http://www.childrensliteraturenetwork.org/birthbios/brthpage/07jul/7-29karl.html
She was a brilliant author and editor. Under-appreciated.
Also:
Runaway Robot by Lester Del Ray
FWIW, I've got some high-end Monster cables for speaker cables and some Blue Jeans cables for interconnect. The Blue Jeans cables are great. (Good audio cables isn't hard, just use some nice co-ax.)
About ten years ago I attended a workshop by Stanford professor David Stork. He mentioned some work on a system that was deployed for use by aircraft technicians: the system couldn't read the voice channel with the jet engine blasting away (the techs wear hearing protection). So it read lips. Ten years ago.
Sounds like TFA is talking about doing this in an embedded, consumer-electronics application. Rather than a fixed, industrial-military, hire-computer-scientists-to-maintain-it thing.
Not-so-coincidentally, David Stork is the author of the book, "HAL's Legacy"...
We target Mac OS X as a supported platform because our users demand it.
For our scientific applications, we support Linux and Mac OS X. Someday maybe there will be a Windows version, but I'm not expecting that any time soon.
I think that "-march=pentium-m -msse3" is WRONG, as it will favor x87 instructions rather than use the SIMD vector unit. Result is that the Core benchmarks slower than the Pentium 4. WRONG!
From the Gentoo Wiki http://gentoo-wiki.com/HARDWARE_Apple_MacBook#CFLA GS
This seems to be confirmed by posts to the GCC mailing list. http://gcc.gnu.org/ml/gcc-help/2006-06/msg00080.ht ml
At work use Linux predominately, but Macs are very popular and getting more so, particularly for laptops.
At the moment, a convenient way to move large data sets is to unplug the USB hard disk and move it to your other workstation, rather than pull that data across the network, because:
1) Portable machines which are used half the time "off net" do not lend themselves well to file-server usage.
2) Scientific data-reduction packages typically involve processing of multi-gigabyte data sets.
OS X 10.4 broke the ext2 project for Mac OSX, alas. As other Linux-supported filesystems:
* HFS+ -- OK, but it doesn't support the journal. Risky?
* FAT32 -- limited. File sizes less than 2GB, and volume sizes are limited, too.
* UFS -- Linux doesn't yet support read-write; has endian issues when trying to share between Macs.
I'm hoping that some sort of Linux emulation on the Mac will allow Macs to access Linux data for filesystems that are not yet ported (ext3) or won't be ported anytime soon (Reiserfs).
P1265 =
12-inch silicon wafer
65-nm process
Well, this is wrong. Some friends of mine had to buy their album back from the record label in order to get it published:They ended up spending lunch money that paid the same marketing people who decided their album would not sell. Sure, they paid for the production costs (studio time, engineer time) too, but they were used to paying quite a bit less for that, and the cost of production goes down all the time. These fellas did not need studio musicians, they could have recorded the thing on a Tascam 10000 in their garage, and sold it at concerts. Would have saved everyone a lot of trouble.
What record labels and distributors still provide is a form of peer review. But iTunes' album and artist reviews, along with the simple hyperlinking and collaborative-filtering effect you get from published iMixes, makes iTunes a far more effective method of peer review than the record labels.
I don't think record labels help very much anymore. I want to support the production and distribution of music that I enjoy. So I purchase music on iTunes. But I want the record labels to get out of the way.
> ...USC Berkeley...
Awesome. One incorrect letter -- the most succinct way of pissing off Northern Californians and Southern Californians I've yet encountered.
So I was a Gentoo zealot for two years, it helped to teach me (more than I wanted to know) about Linux.
I got fed up with power-management issues on my employer-supplied laptop computer (a nice machine, but not Linux-friendly) and purchased a Macintosh PowerBook. Very nice, not as clean as Gentoo, but it got me interested in *BSD.
My server was running Gentoo SeLinux until last week. I've installed FreeBSD 5.2.1 and I am *very* happy with it as a stable, secure server platform.
Linux, Apache, etc. have lent legitimacy to Open Source, and BSD license is attractive to many who cannot otherwise use Open Source. So *BSD is helping spread Open Source, and to otherwise improve the quality of the aggregate code base.
Since Gentoo was developed by someone who liked BSD but wanted the device-driver support of Linux, I feel that most of my skills transfer very quickly. I feel that my learning curve on FreeBSD helps me better understand Mac OS X, which has an installed base of about 12 million computers (if Apple is to be believed).
BSD is dead? Hmm. I rather doubt it.
Well, in our field we have lots of custom software that we use in-house for scientific data-reduction and for equipment monitor/control.
So we use Java for custom applications - GUI apps as well as web services.
James Gosling and some other Sun people came out here to visit, and perhaps got the impression that everyone in science is using Java on Linux as much as we are. Shrug. We like it.
http://www.newegg.com
part #N82E16817145316 External firewire enclosure $ 50 each
part #N82E16822152011 Samsung 160GB hard drive $105 each
http://www.pccables.com
part #70924 FIREWIRE HUB 6-Port $ 35 each
With the six-port hub, we set up a topology such that we have one
hard disk per port, minus one port which goes to the host
computer. Then we have five disks per hub in the array; if one of
them fails, we can pull it off-line without touching the cables for
the other drives, since they all go to the hub.
So it is $155 per 160GB element of the array, plus $35 for the
hub, amortized over the five disks (35/5=7) is $162 for 160GB.
At RAID 5, we have (N-1) effective storage capacity, or 4*160=640GB
for $810.
Since most every firewire host adapter has two ports, we can have two
of these hooked up to the computer, for a total of 1.3TB of RAID5
on-line for about a dollar a Gigabyte.
It's a bloooody mess, with twelve power-supply bricks, but it is
remarkably cheap for a hot-swappable personal server.
Note that we can scale this topology if we want to; we could
implement RAID 10, by adding elements to each of the hub ports in a
daisy-chain fasion; FireWire supports up to 63 devices... for a
personal server, I don't anticipate saturating the firewire
bandwidth. We would implement RAID 10 or other 2-D RAID topologies by
means of combining RAID with Logical Volume Manager (LVM) -- a RAID5
array per hub, with each element of the RAID5 array being an LVM
expandable volume. You would have to add to the LVM in "slices" of
five disks, or otherwise keep the RAID5 elements the same size...
Actually I'm not sure that this last point is meaningful; I would not
plan to do such a thing. But it is worth noting that I can
dynamically attach other devices to the firewire array by connecting
to the spare ports on the back of the disk enclosures.
I've been using GCC 3.2_pre on a Gentoo Linux PIII laptop for about a month. Everything seems to work just fine.
That is, gcc 3.2 is the ONLY gcc on this computer. So the ABI interop issue isn't a problem, I suppose.
Has anyone tried using rsync/rdiff coupled with version control?
http://www.stanford.edu/~bescoto/rdiff-backup/Alan Cox wrote:
I Am Not A Kernel Hacker (IANAKH) but..
Jonathan Shapiro's Ph.D. thesis work impressed me in this regard - a "nanokernel"-based OS that attempts to better support low-level security:
It would be nice to implement a linux-like system on top of this, or perhaps use such a system to provide virtual machines on top of the hardware (like IBM mainframes or VMWare GSX Server).
I attended the Virtual Observatories of the Future conference this past summer and would like to note that:
The take-home lesson from the Virtual Observatories conference was that the amount of data required to do science with a "virtual observatory" leads to interesting problems in computer science, problems which are only tractable when analyzed by collaborations between statisticians, computer science people, and the astronomers themselves.
Finally, note that this year's historic increase in the National Science Foundation budget is largely due to the new Information Technology Research Initiative. The need for new methods of data managment in the sciences is real.