Just like to point out the limitations of Solaris.
I can see your point in the server space. Solaris does a lot of things very well. As you allude to however, Linux is stronger in the x86 driver arena.
If you're running Solaris friendly hardware then you're fine. If you have hardware that Solaris doesn't like (and I've found it *very* fussy... even with top shelf kit) then IMHO robust Linux distros are the way to go.
For some, having Solaris support dictate their hardware might be OK for the special joy the OS brings. If you're happy to limit your apps as a consequence then obviously your call too.
I'm way short of being an expert on the subject but I believe young humans only developed "concept of self" etc at around the 3 to 4yo mark. There's a lot that starts to click around this age.
I suspect that if it was chimps v's 5 year olds the results might have been a bit different.
The primary mirror will be able to alter its shape to compensate for deformation due to gravity at various elevations etc. That's "active optics" and you're right - that's no small challenge.
However the biggest outstanding problem is over coming turbulence in the atmosphere. That's Adaptive Optics and a hot research topic at the moment. Any telescope bigger than about 300mm isn't diffraction limited. It's limited by the atmosphere (Fried's coherence length aka r0).
There are some nasty requirements for AO. The detection of wavefront deformation and correction are huge engineering challenges. Most of the AO system deformable mirrors sit behind the secondary mirror although there's a fair bit of effort going into deformable secondaries atm eg. MMT and LBT. It costs big bucks for that kind of development and there are a *lot* of AO systems gathering dust because they were sooo expensive to keep tweaking with.
If I had the purse strings I'd want to see their AO design before they got a penny.
Hmmm... if you're only talking a single instrument. Very Long Baseline Interferometry (VLBI) when the VSPO mission was flying still holds the record for the largest telescope IMHO:) Around a 30,000km baseline is hard to beat!
You're right. The heart of Maya sweet. Modular and clean makes for excellent scripting capability. The GUI just calls the same code. Easy. Simple.
It's a work of art from a application point of view (and the quality of the results compared to the opposition speak for themselves - esp the soft models).
3DS represents the opposite.
I hope AD supports the ongoing development and steers clear of any thoughts to significantly change the AW approach to the application. This would be bound to keep WETA, Dreamworks, Pixar et al happy too!
I humbly suggest that those "at the highest" levels in this case are the ones in trouble.
"considered options and alternatives, and made a big dollar commitment" then they chose wrong. Use the *best* tool for the job. What that is depends on the job. Not always MS.
The "highest" levels have made a bad choice. Glad you got out. The company was run by fools.
Yeah, but as it goes onto say in that article there's currently not enough $'s in making HPC chips for anyone to find the job commercially worthwhile.
It'll be an interesting decision for the Japanese. The Earth supercomputer used custom NEC processors IIRC. It was a top performer for quite a while. The other current top performers are PowerPC, Itanium and Opterons. Maybe not vanilla processors but not custom either.
If you remove commerical pressures then maybe custom makes sense... we'll see I s'pose:)
Cheers Stevo
Re:LOFAR is going to be exciting
on
Software Telescope
·
· Score: 5, Informative
Hmmm. Wasn't Euro v's USA. It was the Dutch v's everyone else.
LOFAR was supposed to be the international forerunner to SKA for a lot of the tech.
Western Australia won the site selection. Dutch government said "if you build it here we'll throw in a bucket load of cash". Dutch reps took the bird in the hand (kind of understandable given the global spending habits of governments on peaceful science).
Everyone else in the original LOFAR weren't (and still aren't) too happy.
Still... it's a nice piece of kit.
SKA development continues. WA is again up for site selection. Speaking as an Oz astronomer I'm hopefull. It's a great site for radio astronomy.
For the person who asked, the antenna design is a folded dipole. Google it:)
Hmmm. Not sure Swordgeek. Your epee' is indeed an awesome duelling option (I've fenced) but not as generally robust as a slightly heavier grade sword (for the occasional horse strike or in a melee scenario). As a utility edge weapon I'd definately go jian.
However we both seem to agree that the distinct lack of thrusting in the SW movies is doing the lightsabre a disservice!:)
Of course the trick may be to investigate the development of the lightsabre. Maybe it developed from a single edge weapon and the Jedi style hasn't (yet) evolved?
My question is odd but then again... so is the audience =)
The lightsabre appears to lack an "edge" one finds on steel sabres, however even with the sabre you use a cut and thrust style.
I'm yet to see an example of thrust moves with a lightsabre - I'm happy to be corrected. The fight scenes all seem to involve moves typical of a single edge weapon (Sui/Tang style). Wouldn't you expect to use a lightsabre more like a Jian (double edge)?
For anyone else interested in yet another opinion...
The best way is to do your own TCO. Sure, read the arguments from both camps to make sure you've got a complete list of "factors". Also ensuring that they apply to your case.
But then use your *own* in-house numbers!:)
If you make a switch based on someone else's numbers and it was the wrong move... who do you blame?
Ocean Navigator carried an article a few years ago about how the number of people "lost at sea" reported by the US Coast Guard had *increased* since GPS was invented!
The typical response to was "the batteries went flat...". Hmmmm. Point taken re postage stamps and email but this is a lives-at-stake situation.
BTW, this is also why the US Navy still teaches celestial navigation and morse code.
I thoroughly agree. I'm a project manager that used to be a tech. I become a pm because I'd worked under too many of them that had *no* idea.
Projects are all about scope. Defining what it is that you're doing. Everone thinks that's bleedingly obvious... and they're right. But it ain't easy to do.
Once you're got the scope, the rest should be easy. But isn't. Another classic big project blunder is the lack of realistic funding and schedule. Nobody want's to say it's going to take megabucks and go for years.
So instead you end up with "it won't cost much or take very long". Guess what... budgets and schedules "blow out"! More likely they take as much and as long as someone who understood the aforementioned scope would have said in the first place.
Even when the basics are followed things go wrong. This is the final in the classic series of blunders. If something is starting to look bad - don't tell the project sponsor... we'll be right... maybe.
Big no no! Tell the project sponsor *now*! What's wrong, why it went wrong and how you intend to fix it!! You'll get more respect and less stress. Both of which make it more likely you'll get it sorted.
Ahhhh. I feel better now:) These lessons came to me through the development of a fair amount of scar tissue. These days I never employ a project manager that doesn't burst into tears when asked about their worst project.
Speaking as someone who also benchtests as part of what I do for a living, I'd like to add a few more points.
Test labs are useful for customers without the clout to engage vendors directly. Buying an $80 vid card ain't going to get you any attention from vendors, distributors or even sales outlets.
When you step up the ladder a few rungs though, things can get interesting.
For example, your grant has come thorough and you're about to assemble a cluster to run your nasty embarassingly parallel app. Now you can do some interesting things!
Chances are someone, somewhere is doing something vaguely similar already. Check the Net! Bound to be a mail list on a related topic. Ask a few loaded questions:)
For the chance of a big sale most vendors will lend you evaluation kit. When you know what sort of kit you'd like run - don't be shy... ask! Hardware, software, the whole shabang.
The best test is your application. Science, commercial whatever. SPECflopbongotatas with HT enabled switching fabrics don't mean squat! Cmme up with a good canned scenario for your app and benchmark with that. Wall clock time often sufficies as the measurable.
Things to consider with this approach. A lot of fuss is made about compiler flavours and "standardised" optimisation switches. Forget it. Go for the *most* optimised flavour you can muster. It's your app and the only thing that matters is the clock (for the sake of this example).
A nice little benefit from this approach is that it's often OK to let the vendors into the sandpit after you've played hard. Let them go crazy! Tweak drivers, compilers switches, CPU revisions etc etc. Doesn't matter (to a large extent) what they do. Your app is the reason you're doing this!!
Word to the wise though, don't share results from other tests until all vendors are finished. Sure you can let them know there are others but don't say who. This saves them from going into FUD-mode at this point. They'll do enough of that later when they're unsuccessful =)
Remember, you're in the cumfy chair. Use the power while you're got it (alas, it fades so quickly). The fun soon stops and you'll soon be grinding your new toy into the ground.
Slashdot Mirror
Just like to point out the limitations of Solaris.
I can see your point in the server space. Solaris does a lot of things very well. As you allude to however, Linux is stronger in the x86 driver arena.
If you're running Solaris friendly hardware then you're fine. If you have hardware that Solaris doesn't like (and I've found it *very* fussy... even with top shelf kit) then IMHO robust Linux distros are the way to go.
For some, having Solaris support dictate their hardware might be OK for the special joy the OS brings. If you're happy to limit your apps as a consequence then obviously your call too.
S
I actually just logged on so I could be on the record. (I just don't bother anymore... ) JIC anyone actually reads this that can exert some influence.
/. had good stuff. /. seemed to have a large audience with informed commentry. I was once proud of a 5 score.
Yeah.
What the parent said.
These days pseudo science and a never seemingly ending stream of gaming stories has turned me off.
Where have all the good articles and participants gone? No, really... I didn't get the memo and I want to join you!
I'm way short of being an expert on the subject but I believe young humans only developed "concept of self" etc at around the 3 to 4yo mark. There's a lot that starts to click around this age.
I suspect that if it was chimps v's 5 year olds the results might have been a bit different.
Cheers
Stevo
Sort of... SAR? Synthetic Aperature = yes, Radar = no.
To be more pedantic... All in the "world" / on the Earth? Then the VLA would win at 35km.
S
The primary mirror will be able to alter its shape to compensate for deformation due to gravity at various elevations etc. That's "active optics" and you're right - that's no small challenge.
:)
However the biggest outstanding problem is over coming turbulence in the atmosphere. That's Adaptive Optics and a hot research topic at the moment. Any telescope bigger than about 300mm isn't diffraction limited. It's limited by the atmosphere (Fried's coherence length aka r0).
There are some nasty requirements for AO. The detection of wavefront deformation and correction are huge engineering challenges. Most of the AO system deformable mirrors sit behind the secondary mirror although there's a fair bit of effort going into deformable secondaries atm eg. MMT and LBT. It costs big bucks for that kind of development and there are a *lot* of AO systems gathering dust because they were sooo expensive to keep tweaking with.
If I had the purse strings I'd want to see their AO design before they got a penny.
Still... good luck to 'em
Cheers
Stevo
Hmmm... if you're only talking a single instrument. Very Long Baseline Interferometry (VLBI) when the VSPO mission was flying still holds the record for the largest telescope IMHO :) Around a 30,000km baseline is hard to beat!
Cheers
Stevo
Good luck to ya matey :)
:/
You're right. The heart of Maya sweet. Modular and clean makes for excellent scripting capability. The GUI just calls the same code. Easy. Simple.
It's a work of art from a application point of view (and the quality of the results compared to the opposition speak for themselves - esp the soft models).
3DS represents the opposite.
I hope AD supports the ongoing development and steers clear of any thoughts to significantly change the AW approach to the application. This would be bound to keep WETA, Dreamworks, Pixar et al happy too!
I've crossed my fingers
Cheers
Stevo
Vendetta
http://www.imdb.com/title/tt0434409/
I humbly suggest that those "at the highest" levels in this case are the ones in trouble.
"considered options and alternatives, and made a big dollar commitment" then they chose wrong. Use the *best* tool for the job. What that is depends on the job. Not always MS.
The "highest" levels have made a bad choice. Glad you got out. The company was run by fools.
S
Yeah, but as it goes onto say in that article there's currently not enough $'s in making HPC chips for anyone to find the job commercially worthwhile.
:)
It'll be an interesting decision for the Japanese. The Earth supercomputer used custom NEC processors IIRC. It was a top performer for quite a while. The other current top performers are PowerPC, Itanium and Opterons. Maybe not vanilla processors but not custom either.
If you remove commerical pressures then maybe custom makes sense... we'll see I s'pose
Cheers
Stevo
Hmmm. Wasn't Euro v's USA. It was the Dutch v's everyone else.
:)
LOFAR was supposed to be the international forerunner to SKA for a lot of the tech.
Western Australia won the site selection. Dutch government said "if you build it here we'll throw in a bucket load of cash". Dutch reps took the bird in the hand (kind of understandable given the global spending habits of governments on peaceful science).
Everyone else in the original LOFAR weren't (and still aren't) too happy.
Still... it's a nice piece of kit.
SKA development continues. WA is again up for site selection. Speaking as an Oz astronomer I'm hopefull. It's a great site for radio astronomy.
For the person who asked, the antenna design is a folded dipole. Google it
Cheers
Stevo
Duuude
:)
Just means you've got some leg work to do. You've got the coder's name and organisation. Hunt him down
Most science pro's are prepared to at least talk about their projects if not share.
Cheers
Stevo
> So nearly a week later, I reply. Don't know if you'll ever see this, but anyhow...
:)
;)
Oh indeed. I'm still here
In a one-on-one I agree - small dueling sword indeed (prob italian grip as I enjoy the pointing focus it instills).
As a day-to-day, carry on the hip weapon I'd go the Jian. So when me and my mates meet you and your mates on the street... you get the idea
S
Hmmm. Not sure Swordgeek. Your epee' is indeed an awesome duelling option (I've fenced) but not as generally robust as a slightly heavier grade sword (for the occasional horse strike or in a melee scenario). As a utility edge weapon I'd definately go jian.
:)
However we both seem to agree that the distinct lack of thrusting in the SW movies is doing the lightsabre a disservice!
Of course the trick may be to investigate the development of the lightsabre. Maybe it developed from a single edge weapon and the Jedi style hasn't (yet) evolved?
Cheers
Steve
My question is odd but then again... so is the audience =)
The lightsabre appears to lack an "edge" one finds on steel sabres, however even with the sabre you use a cut and thrust style.
I'm yet to see an example of thrust moves with a lightsabre - I'm happy to be corrected. The fight scenes all seem to involve moves typical of a single edge weapon (Sui/Tang style). Wouldn't you expect to use a lightsabre more like a Jian (double edge)?
Cheers
Steve
Right on X.
Give me a comment that says *why* you're doing it... not *what* you're doing.
S
Maybe it's just me but I'd prefer pain from a hot chick in leather with a whip...
It is just me? OK. Sorry...
Stevo the Devo
how far behind is the space station at this point?
:o
How about measured in billions of dollars?
S
You're right WD :)
:)
For anyone else interested in yet another opinion...
The best way is to do your own TCO. Sure, read the arguments from both camps to make sure you've got a complete list of "factors". Also ensuring that they apply to your case.
But then use your *own* in-house numbers!
If you make a switch based on someone else's numbers and it was the wrong move... who do you blame?
Cheers
Stevo
.. is the "we ain't screwing the planet" lobby starting to sound like the tobacco lobby in the late 70's / early 80's ?
"There's no definitive proof that smoking causes lung cancer..." Remember that one?
Hmmm.
We ain't going to Mars until we fix what we've done to this planet!
Stevo
Ocean Navigator carried an article a few years ago about how the number of people "lost at sea" reported by the US Coast Guard had *increased* since GPS was invented!
The typical response to was "the batteries went flat...". Hmmmm. Point taken re postage stamps and email but this is a lives-at-stake situation.
BTW, this is also why the US Navy still teaches celestial navigation and morse code.
Stevo
I thoroughly agree. I'm a project manager that used to be a tech. I become a pm because I'd worked under too many of them that had *no* idea.
:) These lessons came to me through the development of a fair amount of scar tissue. These days I never employ a project manager that doesn't burst into tears when asked about their worst project.
Projects are all about scope. Defining what it is that you're doing. Everone thinks that's bleedingly obvious... and they're right. But it ain't easy to do.
Once you're got the scope, the rest should be easy. But isn't. Another classic big project blunder is the lack of realistic funding and schedule. Nobody want's to say it's going to take megabucks and go for years.
So instead you end up with "it won't cost much or take very long". Guess what... budgets and schedules "blow out"! More likely they take as much and as long as someone who understood the aforementioned scope would have said in the first place.
Even when the basics are followed things go wrong. This is the final in the classic series of blunders. If something is starting to look bad - don't tell the project sponsor... we'll be right... maybe.
Big no no! Tell the project sponsor *now*! What's wrong, why it went wrong and how you intend to fix it!! You'll get more respect and less stress. Both of which make it more likely you'll get it sorted.
Ahhhh. I feel better now
Yepo!
:)
The Hayden display is based on a package called Partiview. Google away
Stevo
Speaking as someone who also benchtests as part of what I do for a living, I'd like to add a few more points.
:)
Test labs are useful for customers without the clout to engage vendors directly. Buying an $80 vid card ain't going to get you any attention from vendors, distributors or even sales outlets.
When you step up the ladder a few rungs though, things can get interesting.
For example, your grant has come thorough and you're about to assemble a cluster to run your nasty embarassingly parallel app. Now you can do some interesting things!
Chances are someone, somewhere is doing something vaguely similar already. Check the Net! Bound to be a mail list on a related topic. Ask a few loaded questions
For the chance of a big sale most vendors will lend you evaluation kit. When you know what sort of kit you'd like run - don't be shy... ask! Hardware, software, the whole shabang.
The best test is your application. Science, commercial whatever. SPECflopbongotatas with HT enabled switching fabrics don't mean squat! Cmme up with a good canned scenario for your app and benchmark with that. Wall clock time often sufficies as the measurable.
Things to consider with this approach. A lot of fuss is made about compiler flavours and "standardised" optimisation switches. Forget it. Go for the *most* optimised flavour you can muster. It's your app and the only thing that matters is the clock (for the sake of this example).
A nice little benefit from this approach is that it's often OK to let the vendors into the sandpit after you've played hard. Let them go crazy! Tweak drivers, compilers switches, CPU revisions etc etc. Doesn't matter (to a large extent) what they do. Your app is the reason you're doing this!!
Word to the wise though, don't share results from other tests until all vendors are finished. Sure you can let them know there are others but don't say who. This saves them from going into FUD-mode at this point. They'll do enough of that later when they're unsuccessful =)
Remember, you're in the cumfy chair. Use the power while you're got it (alas, it fades so quickly). The fun soon stops and you'll soon be grinding your new toy into the ground.
Enjoy.
Stevo
The Xbox doesn't make a good cluster for a few reasons (as the article points out).
:)
There are much better bang for buck options out there.
So speaketh the man building a cluster at home with dual Slot 1 P3 mobos (in a frame to save the cost of multiple cases