As a side note, I'm not entirely happy with how OS X is designed. It's the whole problem of its lackluster performance (Aqua on top of Darwin/BSD on top of Mach gives you a nasty penalty; just look at the benchmarks).
Yeah, I kinda wonder why they bothered to work the freebsd kernel on top of mach. Even back then it wasn't exactly news anymore that context switches on mach suck big time. And freebsd already worked nice on both ppc and x86..
Well, considering how much of that Jet-A1 goop helicopters drink, I suspect it would be less bad for the environment just to leave those oxygen bottles there.
For the bodies, I guess its more about what the relatives feel or something like that..
Well, from the extremely short fluffy explanation in TFA, it thought it sounded more like a synchronous reluctance (SR) AC motor.
SR motors are supposed to be about the most efficient electric motors around (~97 %, IIRC), however to control them you need a microchip and a sensor (=expensive), or then you can have a somewhat fancier microchip (DSP) which measures the stator current from which the rotor angle can be estimated.
I recently bought an A-link WL54H PCI card (about 30 EUR), which has the rt2500 chip. See http://ralink.rapla.net/ for a big list of devices with the rt2500 chip.
Debian can drop almost 50% in one year? That's too many points in one year to be accepted as is. Being on the Debian mailing list I find very few mentions of people dropping or moving.
Well for some anecdotal evidence, after using Debian since 1996 I switched to Ubuntu a month ago. Otherwise I'm happy with Ubuntu, except that a security update to kde fucked up the kdelibs-data or was it knetworkconf package, and it still hasn't been fixed.
I bought an ASUS A8V (S939 for A64 cpu:s, via chipset) a few weeks ago. As expected, it had all kinds of things built-in, including GbEth and Ac97 audio, but no onboard video.
As far as the audio goes, IMHO the sound quality is better than with my old "SB Live value" card. It uses slightly more cpu time though, but with today's cpu:s, who cares? Unfortunately, I haven't got the onboard MIDI/gameport to work in Linux yet, not that I have tried that hard though.
My advice to the OP: As others have said, a motherboard with onboard stuff isn't really more expensive than a bare-bones board. So if money is tight, buy a board with lots of onboard stuff, and sell the extra cards on ebay or whatever.
Works just fine without the gui. However, the help browser in the GUI is quite good. Might be that the help pages are stored as html somewhere though, so you could read them with any web browser? And of course, the editor that comes with the GUI is close to useless, use the matlab.el matlab-mode for emacs that floats around the net somewhere..
I strongly suspect autovectorization is one of those few things in GCC that will work substantially better on PPC than x86. Apple is investing HUGELY in autovectorization in GCC, and PPC is far and away their primary goal.
Umm, no. The _hard_ thing about autovectorization is analyzing the source code and "seeing" which operations can be done in parallel. After you have done that choosing which asm instructions to generate (Altivec or SSE1/2/3) is a much simpler problem.
Re:Is anyone else curious what SSA trees are?
on
GCC 4.0.0 Released
·
· Score: 1
I see that you're using latex-beamer? Congratulations!;-)
Yes, I recognized it since I use it myself. It's really cool.
The game dynamics are such that if you don't expand at a ridiculous rate, the AI will eat you for breakfast. This is one thing that I find annoying in freeciv; your empire consists of a zillion size 2 cities.
2. how do I diagnose subtle problems caused by hibernate join failures? Exactly what sql is it running?
You turn on the sql output switch in the config file, and look at the sql which depending on your configuration is output to the console or the log file.
3. how much flexibility does it have?
A lot.:-)
Can it handle outer joins?
Yes.
Subselects?
Yes.
Group bys?
Yes.
Aggregate functions?
Hmm, I don't remember, but I suspect not. Of course it is possible and advisable to use both SQL and Hibernate in the same application.
4. How do you leverage dbms-specific features? Yes, I know that it's best to stay generic, but when you need to add more analytical queries to your app (to deliver something beyond simple CRUD), you want to use olap functions, etc. Now, yeah, you could do this in your app. But it'll be slower. *Far* slower: you might have to process a hundred thousand rows - and the database will do it in less than a second. Your app? Maybe 5-10 seconds?
In general, Hibernate has pretty good support for stuff that you can find in standard SQL. Also, the architecture of Hibernate makes it extremely suitable for CRUD style stuff, e.g. the caching works very well if the app has any notion of working set, and Hibernate saves huge amounts of time when retrieving complicated object graphs compared to writing SQL by hand. For reporting, OLAP etc. Hibernate is less well suited. The general consensus seems to be to take advantage of Hibernate for CRUD stuff and do complicated reporting/OLAP/whatever in (vendor specific) SQL.
My recommendation is to learn sql, use standard sql whenever possible for portability. But learn the advanced sql required to deliver robust functionality. And when required be prepared to dip into vendor extensions.
My recommendation to you is to at least take a look at Hibernate; it makes you look less like a fool when people see that you have at least a vague idea of what you're talking about.;-)
Or stay away from huge, challenging, and powerful business applications.
Yeah, that's what I do these days, business apps tend to be very boring. Fortran (and computational physics) forever!
saippuakauppias (translated: soap dealer,or soap merchant, anyways, a guy that sells soap you get the point) is the most commonly used finnish palindrom.
Not to mention that a variant of that word, "saippuakivikauppias", is the worlds longest palindrome.
I liked the solid 2 year release cycle before woody, bo, slink, potato, woody.. all nicely timed.
Actually, if you look at the release dates, you see that each release cycle has taken about 6 months longer than the previous.
I think what hurts debian is not necessarily that the release cycle is longer than most distros, but that it is unpredictable. They have been promising that "this time we'll release faster" for ages, and nothing happens, every release drags and drags. I think they should make a schedule (precise details don't matter), and make sure that they follow it.
Normal cpu's already have pretty good fpu units, which are very fast for scalar code. Also, we have things like SSE2/Altivec for vectorizable code. And then there's things like gpgpu.org looking at using the massively parallel fpu capacity of modern gpu's for general purpose physics calculations (linear algebra), i.e. vector processing on a budget.
So where does this thing fit in? As expected, the "article" was nothing more than a thinly veiled marketing blurb, so no info there. Personally, I find it hard to believe that the PPU is competetive with FPU's and GPGPU for general purpose FP calculations. That leaves a chip optimized for certain operations, a bit like MDGRAPE. Or what am I missing?
yes, one single moving part that if it has a hairline fracture almost guarantees to shread your entire engine and most of the rest of your plane when it goes.
Yes, that is one failure mode of a jet.
However, what is relevant here is the reliability of the entire package, not dramatized descriptions of single failures. On average jets are about 7 times as reliable as the piston engines. Another benefit is that while jets are not as frugal as pistons, the fuel is vastly cheaper and safer than high octane avgas.
Also, can you agree that the features of processor A may be more suitable for building a 100 processor SMP/NUMA system than those of processor B or C?
No, since what sets 100 cpu systems apart is not the particular brand or instruction set of the CPU, but rather the NUMA chipset and assorted hardware used to connect them.
The reason big Suns are good is not due to some inherent superiority of the SPARC microarchitecture, but rather the kickass high-end systems design team Sun has.
Slashdot Mirror
As a side note, I'm not entirely happy with how OS X is designed. It's the whole problem of its lackluster performance (Aqua on top of Darwin/BSD on top of Mach gives you a nasty penalty; just look at the benchmarks).
Yeah, I kinda wonder why they bothered to work the freebsd kernel on top of mach. Even back then it wasn't exactly news anymore that context switches on mach suck big time. And freebsd already worked nice on both ppc and x86..
Well, considering how much of that Jet-A1 goop helicopters drink, I suspect it would be less bad for the environment just to leave those oxygen bottles there.
For the bodies, I guess its more about what the relatives feel or something like that..
Well, from the extremely short fluffy explanation in TFA, it thought it sounded more like a synchronous reluctance (SR) AC motor.
SR motors are supposed to be about the most efficient electric motors around (~97 %, IIRC), however to control them you need a microchip and a sensor (=expensive), or then you can have a somewhat fancier microchip (DSP) which measures the stator current from which the rotor angle can be estimated.
Sorry, you just lost the
Unfortunately for the Saudis, IIRC the Persian Gulf is very shallow.
Well, whatever floats your boat, I guess...
There is already a SQL 2003 standard.
There is an open source project which is developing the ralink drivers further, see http://rt2x00.serialmonkey.com/wiki/index.php/Main _Page. And yes, the driver works on 2.6 too, although it's not merged.
I recently bought an A-link WL54H PCI card (about 30 EUR), which has the rt2500 chip. See http://ralink.rapla.net/ for a big list of devices with the rt2500 chip.
IIRC DELL sells their computers with freedos if you ask for it?
Debian can drop almost 50% in one year? That's too many points in one year to be accepted as is. Being on the Debian mailing list I find very few mentions of people dropping or moving.
Well for some anecdotal evidence, after using Debian since 1996 I switched to Ubuntu a month ago. Otherwise I'm happy with Ubuntu, except that a security update to kde fucked up the kdelibs-data or was it knetworkconf package, and it still hasn't been fixed.
I bought an ASUS A8V (S939 for A64 cpu:s, via chipset) a few weeks ago. As expected, it had all kinds of things built-in, including GbEth and Ac97 audio, but no onboard video.
As far as the audio goes, IMHO the sound quality is better than with my old "SB Live value" card. It uses slightly more cpu time though, but with today's cpu:s, who cares? Unfortunately, I haven't got the onboard MIDI/gameport to work in Linux yet, not that I have tried that hard though.
My advice to the OP: As others have said, a motherboard with onboard stuff isn't really more expensive than a bare-bones board. So if money is tight, buy a board with lots of onboard stuff, and sell the extra cards on ebay or whatever.
Works just fine without the gui. However, the help browser in the GUI is quite good. Might be that the help pages are stored as html somewhere though, so you could read them with any web browser? And of course, the editor that comes with the GUI is close to useless, use the matlab.el matlab-mode for emacs that floats around the net somewhere..
I strongly suspect autovectorization is one of those few things in GCC that will work substantially better on PPC than x86. Apple is investing HUGELY in autovectorization in GCC, and PPC is far and away their primary goal.
Umm, no. The _hard_ thing about autovectorization is analyzing the source code and "seeing" which operations can be done in parallel. After you have done that choosing which asm instructions to generate (Altivec or SSE1/2/3) is a much simpler problem.
I see that you're using latex-beamer? Congratulations! ;-)
Yes, I recognized it since I use it myself. It's really cool.
Thanks, I'll have to try that strategy sometime.
I wonder how the freecivs ai compares to that
The game dynamics are such that if you don't expand at a ridiculous rate, the AI will eat you for breakfast. This is one thing that I find annoying in freeciv; your empire consists of a zillion size 2 cities.
Ok, here are a few reasons why I *wouldn't* want to use that hibernate syntax:
It's been a year or so since I last used Hibernate, but IMHO, it rocks compared to hand-written SQL. Now, for your points:
1. how do I separately validate the syntax & query results? Can I run it in on a command-line? In a query tool, etc?
There exists a GUI tool for this, see http://www.hibernate.org/259.html
2. how do I diagnose subtle problems caused by hibernate join failures? Exactly what sql is it running?
You turn on the sql output switch in the config file, and look at the sql which depending on your configuration is output to the console or the log file.
3. how much flexibility does it have?
A lot.
Can it handle outer joins?
Yes.
Subselects?
Yes.
Group bys?
Yes.
Aggregate functions?
Hmm, I don't remember, but I suspect not. Of course it is possible and advisable to use both SQL and Hibernate in the same application.
4. How do you leverage dbms-specific features? Yes, I know that it's best to stay generic, but when you need to add more analytical queries to your app (to deliver something beyond simple CRUD), you want to use olap functions, etc. Now, yeah, you could do this in your app. But it'll be slower. *Far* slower: you might have to process a hundred thousand rows - and the database will do it in less than a second. Your app? Maybe 5-10 seconds?
In general, Hibernate has pretty good support for stuff that you can find in standard SQL. Also, the architecture of Hibernate makes it extremely suitable for CRUD style stuff, e.g. the caching works very well if the app has any notion of working set, and Hibernate saves huge amounts of time when retrieving complicated object graphs compared to writing SQL by hand. For reporting, OLAP etc. Hibernate is less well suited. The general consensus seems to be to take advantage of Hibernate for CRUD stuff and do complicated reporting/OLAP/whatever in (vendor specific) SQL.
My recommendation is to learn sql, use standard sql whenever possible for portability. But learn the advanced sql required to deliver robust functionality. And when required be prepared to dip into vendor extensions.
My recommendation to you is to at least take a look at Hibernate; it makes you look less like a fool when people see that you have at least a vague idea of what you're talking about.
Or stay away from huge, challenging, and powerful business applications.
Yeah, that's what I do these days, business apps tend to be very boring. Fortran (and computational physics) forever!
saippuakauppias (translated: soap dealer,or soap merchant, anyways, a guy that sells soap you get the point) is the most commonly used finnish palindrom.
Not to mention that a variant of that word, "saippuakivikauppias", is the worlds longest palindrome.
I liked the solid 2 year release cycle before woody, bo, slink, potato, woody.. all nicely timed.
Actually, if you look at the release dates, you see that each release cycle has taken about 6 months longer than the previous.
I think what hurts debian is not necessarily that the release cycle is longer than most distros, but that it is unpredictable. They have been promising that "this time we'll release faster" for ages, and nothing happens, every release drags and drags. I think they should make a schedule (precise details don't matter), and make sure that they follow it.
Except that the IBM's xlf Fortran95 compiler only accepts
Why, why, oh why?! It is really, really annoying (and I've seen no flag to auto-convert
Uh, ever read the man page?
"xlf -qsuffix=f=f90 file.f90" does the trick.
Interface now uses GTK instead of wxWidgets
Any particular reason why? I haven't used pygtk2, but in my limited experience wxpython is nicer to use than pygtk1.
Normal cpu's already have pretty good fpu units, which are very fast for scalar code. Also, we have things like SSE2/Altivec for vectorizable code. And then there's things like gpgpu.org looking at using the massively parallel fpu capacity of modern gpu's for general purpose physics calculations (linear algebra), i.e. vector processing on a budget.
So where does this thing fit in? As expected, the "article" was nothing more than a thinly veiled marketing blurb, so no info there. Personally, I find it hard to believe that the PPU is competetive with FPU's and GPGPU for general purpose FP calculations. That leaves a chip optimized for certain operations, a bit like MDGRAPE. Or what am I missing?
yes, one single moving part that if it has a hairline fracture almost guarantees to shread your entire engine and most of the rest of your plane when it goes.
Yes, that is one failure mode of a jet.
However, what is relevant here is the reliability of the entire package, not dramatized descriptions of single failures. On average jets are about 7 times as reliable as the piston engines. Another benefit is that while jets are not as frugal as pistons, the fuel is vastly cheaper and safer than high octane avgas.
Also, can you agree that the features of processor A may be more suitable for building a 100 processor SMP/NUMA system than those of processor B or C?
No, since what sets 100 cpu systems apart is not the particular brand or instruction set of the CPU, but rather the NUMA chipset and assorted hardware used to connect them.
The reason big Suns are good is not due to some inherent superiority of the SPARC microarchitecture, but rather the kickass high-end systems design team Sun has.
And a year or so after that there'll be, uh, Königstiger?