Do not be too proud of this technological terror you have created. The power to store 51GB of data is insignificant next to the power of the market forces demanding pr0n.
It's possible. The big problem with using a GPU to handle the processing is that they only support single precision floats and aren't very good at branching algorithms. Not sure how a pixel shader would handle 16 bit grey scale pixels either.
1. Google indexes massive amounts of data. The telescope imagery will be a massive amount of data.
True enough, but google indexes massive amounts of data substantially different from imagery data. This would be something more akin to google earth, which is really nice but not particularly groundbreaking technology so far.
2. Google has huge data centers capable of a great amount of distributed processing. The telescope data will require a lot of possibly parallel data processing (multiple images, FFTs on the images, comparison between sequences, etc)
Yes and no. A substantial portion of LSST's computing power needs to be in close proximity to LSST. Every minute you are doing 2 frame captures + slew + settle + probably another frame capture. So roughly 3 images per minute, with a raw data size of 6 GBytes per image = 18 GBytes/minute all night long. Even if you could get a 6:1 compression on the images, you would need a pipeline to a remote area of Chile capable of 3GB/minute (~400Mbit/S) sustained. So for early event detection, the massive data centers in northern america will be of little use. You can only get that kind of throughput using dedicated lines between the computing center and the telescope.
Google will, however, be able to archive old images so that when something new is found, it will be possible to quickly look through old images of the same piece of sky to see if there was anything interesting there in the last 6 months or a year.
3. Google has a plethora of graduate level employees - who better than a bunch of PhD scientists to store / process / index massive amounts of scientific data?
To be fair, just about everybody working on LSST prior to Google's announcement is a Ph.D./Graduate Student (just look at the list of institutions involved in the project - lots of universities and a couple national laboratories). Many of them in astronomy and computer science, and many of them familiar with algorithms for analysis of astronomical imagery.
When I was working on it, I never once heard the name "google" dropped, so I don't know exactly the relationship. We were researching ways to have the computer identify phenomena based on pre-existing photometric pipelines already in use. In my case I was taking an existing algorithm and fitting it onto a super-parallel numeric processor and racing it against general purpose processors (okay, I only actually benched it against a P4, the drawback to these summer internships is they only last as long as the summer).
Correct me if I'm wrong, but from what I can tell of this project, it's going to do exactly that (and more), but on a larger scope, and with better accuracy?
Well, I was a very small cog for a very large telescope. But my understanding is pretty much exactly what you just said.
I actually did a small, insignificant portion of LSST's computation feasability study at LLNL during my internship there a couple summers ago. And yeah, the computational requirements were nothing to sneeze at. I'm not sure where they are at now, the specs changed seemingly every month, but when I left the CCD array was up to 3 gigapixels of 16 bit greyscale. I believe the observing cadence (at that time, again everything was changing on a regular basis) was two of those for the same piece of sky every 30 seconds. Wish I could have stayed... ahh well. I did get a really nice full-color research poster (that I had to design) out of it though!
When I graduated college, three places were chomping at the bit to hire me: Company X (a fortune 50 company), LLNL and one-of-those-UCs. To make a long story short, I took a moderate pay cut and went with the UC, expecting an atmosphere similar to LLNL (my previous internship, and managed by UC) only slightly more relaxed. It was a mistake.
Having only a bachelor's degree, I was the least educated person on staff. Despite being the only person on staff who had ever taken a product through to release, and despite being hired to bring a web product to public release, my opinion was worth practically nothing. When we were 9 months from release and I said "we need a management plan right now or we're never going hit that target, and we need to start defining what features will be at release" I was told by the PI that my comment was "inappropriate and premature." Seriously, I walk in 9 months from product release and they hadn't even defined requirements. We clashed over many other things. For the first time in my life I started having stress-related health problems - working at a University!
6 weeks into my UC employ, both Company X and LLNL made poaching offers. Both of them offered nearly 50% over what the UC was paying me. For purely financial reasons I went with Company X (too expensive to live near LLNL). Not a day goes by that I regret my choice (to leave UC, I still miss LLNL a little). The comment that got me censured at UC was actively encouraged at X. Focus on delivery dates that caused stress at UC, gets me recognition and bonuses at X. X is not perfect, there's still some bureaucratic nonsense, and I understand the area I'm in has a glass ceiling and I will have to transfer elsewhere in the company or stagnate at my current pay level. But all that is insignificant compared to "we want you to deliver this product in 9 months; and we're not really sure what the product will do!"
So my experience with the university was: The pay was not competitive, the stress was unreasonably high, my first project was a sinking ship with no bilge pump, and I could not make it to anything resembling "director" unless I had at least a masters degree (which is going to happen anyway, but still).
Wouldn't help. The hippie gene is endemic in the human population. You could send a colony to mars, but within a few generations inevitably somebody will give birth to a child who has two copies of the recessive hippie gene. At that time the first fuel-inefficient VW buses will start production and it'll be all downhill from there. Until we come up with a treatment plan for hippiism, we're stuck with it.
The cell phone ping came 2 days after he left them.
Do you have a reference for this?
According to this, the cell phone ping in question occurred on Nov 26. It sounds like there is confusion between when the ping happened and when the tower that received the ping was located.
Okay, I finally searched down the reference for this. It looks a lot like trumping up a minor issue in order to make Kim's fateful trip seem more heroic. The helicopter pilot was following a single set of tire tracks (indicating someone probably went in and didn't come out) and also happened to see footprints. Had the footprints not been there, there's no reason to believe the result would have been different because the pilot was following the tire tracks.
If you really want to learn something from this scenario, telling people to spend a lot of money on something they'll be VERY unlikely to ever use is foolish.
Is insurance any less foolish? Is the cost of insurance much less? You have to evaluate honestly your own personality. Are you the kind of adventurous person who would turn a missed turn off into an opportunity for adventure on the road less traveled? I am. Do it all the time. Sounds like the Kim's were too because this is how they got where they were - they missed their main turn off and opted to take a road they were unfamiliar with. 9999 times out of 10000 this turns out good, you see beautiful country that few others do (I know, because I do this all the time). It's that 1 time in 10000 you need the PLB. If you're the type to enjoy this sort of thing, CARRY INSURANCE.
Fourth, the helicopter pilot that finally found the car actually spotted footprints in the snow, and followed them back to the car. If those were James Kim's footprints, then if he had stayed with the car, it might not have been found when it was.
This is like the fifth time I've read this just in this thread. Does anybody have a link from a news outlet stating this? Preferably words from the helicopter pilot? This is starting to feel like building heroics after the fact.
Everything I've read so far suggests that the helicopter did find footprints, but this was AFTER Kati had flagged him down with an umbrella with SOS written on it.
Well, he didn't leave the car for 7 days, and by that time the car had run out of gas and they were burning magazines, tires, etc to stay warm.
They burned the tires much earlier, which was another big mistake. Instead of waiting for a clear day to burn them, they burned them when it was snowing. The thick black smoke which can be seen for 10s of miles around from the tires was lost in the falling snow.
But remember that they probbably hadn't eaten much (if anything) for 7 days, and might have had a problem getting water (melting snow takes energy).
It was probably warmer in the car than outside. The sun could likely melt the water they needed. The fact that Kati was able to nurse her children suggests she had a good supply of water. A mother can nurse on her fat stores without food for a good many days, but without water she'll be in trouble in just 1 or 2.
Also, the family was only found because the helicopter pilot saw footprints in the snow (those of James Kim). The footprints lead him to the car of the family. If Kim hadn't gone out to try to find help, the whole family might have died from exposure.
This is the third time I've heard this. Do you have a source for this? I've googled three different phrases and they all mention James was tracked by his footprints, I can't find any that say that the footprints were found before Kati.
So, given the same situation are you really sure you'd just sit around forever and hope that you're rescued? I'm not sure I would have.
Being a father myself, I say it's a really REALLY hard decision. By this time he is probably thinking he is going to sit there for a few weeks and watch his children starve to death. Desperation leads people to do things that aren't in their best interest. Experts who study this stuff tend to agree though, your highest probability of rescue comes from staying put.
The guy who thinks that high-tech would have saved him might be right, but it's much cheaper, easier, and reliable to just stash some warm clothes in your car. They'll also be usefull when you're just plain cold.
A $600 personal locator beacon would have brought help directly to him within a few hours (assuming he didn't activate it while the snow was so thick that rescue operations couldn't be safely performed.)
Except that they only found James family by following the tracks he left in the snow back to them from the air.
Do you have a reference for this? Every google search I've done lists the two most prominent factors in the rescue as 1) The cell phone ping and 2) Kati waving an umbrella when a helicopter was nearby.
Being a father myself, I found the James Kim story particularly heart-wrenching. The urge to do SOMETHING for your family would have been absolutely overwhelming. The Kims did so much right (burning the tires), and so much wrong (driving unfamiliar roads in heavy snow). If there is one thing the average person might carry away from this tragedy, there is an expensive but manageable solution that would have saved the Kims in hours: a personal locator beacon. Detected by geo-stationary satellites and located by low-earth orbitting satellites, the PLB's distress signal can be pin-pointed within several meters in just 30 minutes and emits a low-frequency homing beacon to guide rescue crews to you. The beacon must be registered (with NOAA I believe) so that the authorities know who is in trouble. If you've got a family to protect and you'll be on roads you don't know in weather unfavorable to navigation, you owe it to yourself to have one of these.
IANAL, but something just feels weirdly wrong about this. Employers are only supposed to consider things necessary and relevant to the job. They can credit check you to make sure you are not carrying excessive debt or negligent with money or any of those things which might make their assets in danger of your spending habits. It just doesn't seem like they should be able to use a credit history with nothing negative on it against you. Did they give a reason why good credit is a requirement?
Here's a tasty morsel from PJ's mouth. This was in a followup post, not the original article, that I think speaks broadly to PJ's bias:
Linus has now said publicly that only one of the GPL's 4 freedoms matter to him. I doubt that Moglen will go along with that. That doesn't mean he hasn't listened. It means Linus has a lot of nerve to even say something like that.
Followed by:
This is, in my opinion, the enterprise trying to take over or muscle the GPL. I doubt they will succeed, but they do seem to be trying. Let them stay with v2 if they want. I don't care. But to try to sink the GPLv3 ship if they can't have what they want is not right. The enterprise probably hated v2 and still do, if they had full choice, but it's too late.
I don't think PJ's reporting of the whole GPLv3 thing has been even remotely balanced. As I said a couple posts up, she comes off to me as an FSF shill.
Yes, sadly this is what Groklaw has become. I think some of PJ's article posts when she came out against the general linux kernel community and its objection to GPLv3 are also shining examples of groklaw bias. Her hypocritical cries "unfair" to a couple responses just killed the shine on groklaw to me.
I guess we at least learned one thing. She isn't a shill for IBM (Stallman on the other hand...)
Slashdot Mirror
I've heard that Sony doesn't want it on Blu-Ray.
Do not be too proud of this technological terror you have created. The power to store 51GB of data is insignificant next to the power of the market forces demanding pr0n.
It's possible. The big problem with using a GPU to handle the processing is that they only support single precision floats and aren't very good at branching algorithms. Not sure how a pixel shader would handle 16 bit grey scale pixels either.
1. Google indexes massive amounts of data. The telescope imagery will be a massive amount of data.
True enough, but google indexes massive amounts of data substantially different from imagery data. This would be something more akin to google earth, which is really nice but not particularly groundbreaking technology so far.
2. Google has huge data centers capable of a great amount of distributed processing. The telescope data will require a lot of possibly parallel data processing (multiple images, FFTs on the images, comparison between sequences, etc)
Yes and no. A substantial portion of LSST's computing power needs to be in close proximity to LSST. Every minute you are doing 2 frame captures + slew + settle + probably another frame capture. So roughly 3 images per minute, with a raw data size of 6 GBytes per image = 18 GBytes/minute all night long. Even if you could get a 6:1 compression on the images, you would need a pipeline to a remote area of Chile capable of 3GB/minute (~400Mbit/S) sustained. So for early event detection, the massive data centers in northern america will be of little use. You can only get that kind of throughput using dedicated lines between the computing center and the telescope.
Google will, however, be able to archive old images so that when something new is found, it will be possible to quickly look through old images of the same piece of sky to see if there was anything interesting there in the last 6 months or a year.
3. Google has a plethora of graduate level employees - who better than a bunch of PhD scientists to store / process / index massive amounts of scientific data?
To be fair, just about everybody working on LSST prior to Google's announcement is a Ph.D./Graduate Student (just look at the list of institutions involved in the project - lots of universities and a couple national laboratories). Many of them in astronomy and computer science, and many of them familiar with algorithms for analysis of astronomical imagery.
When I was working on it, I never once heard the name "google" dropped, so I don't know exactly the relationship. We were researching ways to have the computer identify phenomena based on pre-existing photometric pipelines already in use. In my case I was taking an existing algorithm and fitting it onto a super-parallel numeric processor and racing it against general purpose processors (okay, I only actually benched it against a P4, the drawback to these summer internships is they only last as long as the summer).
Correct me if I'm wrong, but from what I can tell of this project, it's going to do exactly that (and more), but on a larger scope, and with better accuracy?
Well, I was a very small cog for a very large telescope. But my understanding is pretty much exactly what you just said.
I actually did a small, insignificant portion of LSST's computation feasability study at LLNL during my internship there a couple summers ago. And yeah, the computational requirements were nothing to sneeze at. I'm not sure where they are at now, the specs changed seemingly every month, but when I left the CCD array was up to 3 gigapixels of 16 bit greyscale. I believe the observing cadence (at that time, again everything was changing on a regular basis) was two of those for the same piece of sky every 30 seconds. Wish I could have stayed... ahh well. I did get a really nice full-color research poster (that I had to design) out of it though!
Just a rebuttal based on my experience...
When I graduated college, three places were chomping at the bit to hire me: Company X (a fortune 50 company), LLNL and one-of-those-UCs. To make a long story short, I took a moderate pay cut and went with the UC, expecting an atmosphere similar to LLNL (my previous internship, and managed by UC) only slightly more relaxed. It was a mistake.
Having only a bachelor's degree, I was the least educated person on staff. Despite being the only person on staff who had ever taken a product through to release, and despite being hired to bring a web product to public release, my opinion was worth practically nothing. When we were 9 months from release and I said "we need a management plan right now or we're never going hit that target, and we need to start defining what features will be at release" I was told by the PI that my comment was "inappropriate and premature." Seriously, I walk in 9 months from product release and they hadn't even defined requirements. We clashed over many other things. For the first time in my life I started having stress-related health problems - working at a University!
6 weeks into my UC employ, both Company X and LLNL made poaching offers. Both of them offered nearly 50% over what the UC was paying me. For purely financial reasons I went with Company X (too expensive to live near LLNL). Not a day goes by that I regret my choice (to leave UC, I still miss LLNL a little). The comment that got me censured at UC was actively encouraged at X. Focus on delivery dates that caused stress at UC, gets me recognition and bonuses at X. X is not perfect, there's still some bureaucratic nonsense, and I understand the area I'm in has a glass ceiling and I will have to transfer elsewhere in the company or stagnate at my current pay level. But all that is insignificant compared to "we want you to deliver this product in 9 months; and we're not really sure what the product will do!"
So my experience with the university was: The pay was not competitive, the stress was unreasonably high, my first project was a sinking ship with no bilge pump, and I could not make it to anything resembling "director" unless I had at least a masters degree (which is going to happen anyway, but still).
In Korea, moon dust is only for old people.
there's nothing that bothers me more than having my intelligence insulted by trite propaganda
You must be new here! Welcome!
Wouldn't help. The hippie gene is endemic in the human population. You could send a colony to mars, but within a few generations inevitably somebody will give birth to a child who has two copies of the recessive hippie gene. At that time the first fuel-inefficient VW buses will start production and it'll be all downhill from there. Until we come up with a treatment plan for hippiism, we're stuck with it.
The cell phone ping came 2 days after he left them.
Do you have a reference for this?
According to this, the cell phone ping in question occurred on Nov 26. It sounds like there is confusion between when the ping happened and when the tower that received the ping was located.
I found it from the Wikipedia entry. http://en.wikipedia.org/wiki/James_kim
Okay, I finally searched down the reference for this. It looks a lot like trumping up a minor issue in order to make Kim's fateful trip seem more heroic. The helicopter pilot was following a single set of tire tracks (indicating someone probably went in and didn't come out) and also happened to see footprints. Had the footprints not been there, there's no reason to believe the result would have been different because the pilot was following the tire tracks.
If you really want to learn something from this scenario, telling people to spend a lot of money on something they'll be VERY unlikely to ever use is foolish.
Is insurance any less foolish? Is the cost of insurance much less? You have to evaluate honestly your own personality. Are you the kind of adventurous person who would turn a missed turn off into an opportunity for adventure on the road less traveled? I am. Do it all the time. Sounds like the Kim's were too because this is how they got where they were - they missed their main turn off and opted to take a road they were unfamiliar with. 9999 times out of 10000 this turns out good, you see beautiful country that few others do (I know, because I do this all the time). It's that 1 time in 10000 you need the PLB. If you're the type to enjoy this sort of thing, CARRY INSURANCE.
Fourth, the helicopter pilot that finally found the car actually spotted footprints in the snow, and followed them back to the car. If those were James Kim's footprints, then if he had stayed with the car, it might not have been found when it was.
This is like the fifth time I've read this just in this thread. Does anybody have a link from a news outlet stating this? Preferably words from the helicopter pilot? This is starting to feel like building heroics after the fact.
Everything I've read so far suggests that the helicopter did find footprints, but this was AFTER Kati had flagged him down with an umbrella with SOS written on it.
Well, he didn't leave the car for 7 days, and by that time the car had run out of gas and they were burning magazines, tires, etc to stay warm.
They burned the tires much earlier, which was another big mistake. Instead of waiting for a clear day to burn them, they burned them when it was snowing. The thick black smoke which can be seen for 10s of miles around from the tires was lost in the falling snow.
But remember that they probbably hadn't eaten much (if anything) for 7 days, and might have had a problem getting water (melting snow takes energy).
It was probably warmer in the car than outside. The sun could likely melt the water they needed. The fact that Kati was able to nurse her children suggests she had a good supply of water. A mother can nurse on her fat stores without food for a good many days, but without water she'll be in trouble in just 1 or 2.
Also, the family was only found because the helicopter pilot saw footprints in the snow (those of James Kim). The footprints lead him to the car of the family. If Kim hadn't gone out to try to find help, the whole family might have died from exposure.
This is the third time I've heard this. Do you have a source for this? I've googled three different phrases and they all mention James was tracked by his footprints, I can't find any that say that the footprints were found before Kati.
So, given the same situation are you really sure you'd just sit around forever and hope that you're rescued? I'm not sure I would have.
Being a father myself, I say it's a really REALLY hard decision. By this time he is probably thinking he is going to sit there for a few weeks and watch his children starve to death. Desperation leads people to do things that aren't in their best interest. Experts who study this stuff tend to agree though, your highest probability of rescue comes from staying put.
The guy who thinks that high-tech would have saved him might be right, but it's much cheaper, easier, and reliable to just stash some warm clothes in your car. They'll also be usefull when you're just plain cold.
A $600 personal locator beacon would have brought help directly to him within a few hours (assuming he didn't activate it while the snow was so thick that rescue operations couldn't be safely performed.)
Except that they only found James family by following the tracks he left in the snow back to them from the air.
Do you have a reference for this? Every google search I've done lists the two most prominent factors in the rescue as 1) The cell phone ping and 2) Kati waving an umbrella when a helicopter was nearby.
Being a father myself, I found the James Kim story particularly heart-wrenching. The urge to do SOMETHING for your family would have been absolutely overwhelming. The Kims did so much right (burning the tires), and so much wrong (driving unfamiliar roads in heavy snow). If there is one thing the average person might carry away from this tragedy, there is an expensive but manageable solution that would have saved the Kims in hours: a personal locator beacon. Detected by geo-stationary satellites and located by low-earth orbitting satellites, the PLB's distress signal can be pin-pointed within several meters in just 30 minutes and emits a low-frequency homing beacon to guide rescue crews to you. The beacon must be registered (with NOAA I believe) so that the authorities know who is in trouble. If you've got a family to protect and you'll be on roads you don't know in weather unfavorable to navigation, you owe it to yourself to have one of these.
I wonder why I never see the JVM start up when I compile with javac.
Because you aren't paying attention. Javac is in the package com.sun.tools, see documentation here.
When you're done reading, download jikes and use that instead.
IANAL, but something just feels weirdly wrong about this. Employers are only supposed to consider things necessary and relevant to the job. They can credit check you to make sure you are not carrying excessive debt or negligent with money or any of those things which might make their assets in danger of your spending habits. It just doesn't seem like they should be able to use a credit history with nothing negative on it against you. Did they give a reason why good credit is a requirement?
Depending on your definition of "colonize," antaractica is colonized.
They need to be so fast that going to and from Pluto should take no more than an hour.
186,282.397 miles per second. It's not just a good idea, it's the law!
I can see we're going to get nowhere here, so I'm going to leave with a simple "bias doesn't mean reporting incorrect facts."
Followed by:
I don't think PJ's reporting of the whole GPLv3 thing has been even remotely balanced. As I said a couple posts up, she comes off to me as an FSF shill.
Yes, sadly this is what Groklaw has become. I think some of PJ's article posts when she came out against the general linux kernel community and its objection to GPLv3 are also shining examples of groklaw bias. Her hypocritical cries "unfair" to a couple responses just killed the shine on groklaw to me.
I guess we at least learned one thing. She isn't a shill for IBM (Stallman on the other hand...)
Did you read his thread on Google groups? I'm not speaking just about his blog entry, I'm talking about the entirety of what was done.