While a lot of computational physics requires speed, I find myself on a daily basis needing to write simple programs to collect, filter, transform and plot data. I have found no better language for writing these quickie tools in than Python. I'm currently taking an introduction (for those who will continue with physics) course to mechanics, and almost all of the 12 mandatory assignments contained a bit of programming. We had to derive the necessary equations and then implement them in Python or Matlab.
Since programming wasn't a requirement for the course, each assignment included a "template" Python program where one just had to insert the equations we had found. We were however encouraged to modify it further (use RK4 instead of simple Euler etc).
We then had to find the results analytically and verify the numerical results.
Personally I found this mix very fun and also educational, and I found Python (with the Scitools package) to be good for the task.
I just had to try to put that into perspective, so I looked up the net annual usage of electricity in the US. If we had somehow captured, converted and stored all that energy, the US would spend about 45 million years using it up (assuming linear growth similar to the last 10 years, and I didn't screwed up the math).
The problem is in communicating effectively in spite of the inherent ambiguity, fuzziness, and confusion of human languages. I think it's very unlikely that any team is about to meet this threshold. Indeed, one of the easiest ways I've found to "out" these "intelligent" chat programs is to come up with some statement that is grammatically fine, but which just does not make any sense at all. Something like "I hear the wooden elves are jumping through fish". All programs I've tried so far assumes that the other party makes sense, and so it tries to find a sensible reply, and fail spectacularly.
he pretends that "everyone" is pushing the most naive, brute force approach to raytracing, in which you don't use any kind of bounding volume structure and just do intersection tests against triangles. I've seen literally no recommendations that do that: almost all involve some form of bounding volume heirarchy, and sparse voxel octrees are just one instance of that (perhaps a fairly ideal one, and that's great). Skimming over the article, I get the impression that he's not rendering triangles:
Now what is different about this versus a conventional ray tracing architecture is that it is a specialized data structure that you can ray trace into quite efficiently and that data structure brings you some significant benefits that you wouldn't get from a triangular structure. It would be 50 or 100 times more data if you stored it out in a triangular mesh, which you couldn't actually do in practice. Ray casting voxels obviously not a new thing, but perhaps his supposedly new data structure can make it work on next gen hardware.
[...] Windows seems to go out of it's way to make OpenGL unattractive or non-feasible. Actually, I would say that the ARB is doing just as much to make OpenGL unattractive or non-feasible for games. With the current state of OpenGL, it's hard to find a good (never mind the the optimal) way of doing things across platforms. Usually one has to resort to a lot of guesswork and testing, and what works now might just break or fall back to software mode (aka 1 frame per second) on the next driver version from vendor XYZ. While vendors obviously have the responsibility for writing drivers that work, OpenGL is currently a large beast with so many different ways of doing things that it is very difficult to cover all aspects.
For instance, there's nothing in the specs that disallows using VBOs as a source when accumulating a display list, but when I tried it, it failed on ATI hardware. Now, in ATI's defense, using VBOs in that way is not what they were made for and one might argue it's a bit silly, but still, it solved a problem I had (until I tested on ATI that is). This is one of those gazillion corner cases that's easy to miss.
The OpenGL forums is filled with people asking why something is slow, or the fastest way of doing something. And reading the web doesn't always clear things up, as what was fast a while ago (triangle strips, for instance) is now not so fast, or in many cases much slower (client side arrays vs VBOs). In most cases you'll end up trying out several ways of doing a thing.
OpenGL 3 was supposed to fix this mess by redesigning the API and cutting away non-essentials, but it's currently 7 months (or so) overdue, and the ARB has adopted the "absolute silence" approach to handling PR. This means that we most likely won't get any (useful) OpenGL 3 drivers until next year or so.
I used to be quite a fan of OpenGL, even though I primarily code for Windows. However if I had to start development of a Windows application today which had to use any non-fixed-function features, I would most certainly not use OpenGL.
That's kinda what they said about flying too. Of course there was a couple of brothers who kinda disputed that "laws of physics" Perhaps, I'm not that into history. However, things are a bit different now. Both the Standard Model and General Relativity have been tested extensively. While we know that they are both incomplete, we do know that within their domains, they are exceptionally accurate descriptions of nature. Since this experiment does not involve huge particle accelerators, reactors or say galaxies, one can be quite certain that it falls well within the domain of the Standard Model.
So while the experiment may very well involve some new and previously undiscovered phenomena, I think that we can be quite certain that it won't make us modify our current physical framework.
Why is it that Science is always trying to keep up with the inventors and then "explain" how it works when previously they (the scientists) said it was impossible and would violate all the known laws of "whatever"? Scientists have very good reasons for believing it's not a perpetual motion machine, and that it operates well within the physical laws we know. However that doesn't mean that the effect is easily explained, and may warrant further research.
It's a bit like if I told you to figure out the pattern behind a sequence of numbers. You could be fairly sure that I haven't invented some new branch of mathematics to generate it, however that doesn't mean it's immediately obvious to you how I did it.
I bought a new dual-core notebook [...] but damn is this Vista machine SLOW.
It's a laptop thing, that's my theory. Almost everyone I've seen/heard that complains about Vista has tried it on a laptop. Almost everyone I've seen/heard that thinks Vista is fine, uses a desktop PC.
Personally I made the jump two months ago, and while I can still dual boot, I haven't booted XP in 7 weeks. For me, Vista is significantly faster in several areas, while not noticeably slower in the others (with one exception: deleting files).
I can only speculate as to why this is so, but I suspect crappy OEM software and drivers, which is compounded by slower laptop disks and less RAM.
However, last weekend I bought a new laptop when my old one crapped out. Obviously it had Vista, so I tried to use it for a couple of days. Between the fact it was abysmally slow, consumed a gig of memory just sitting there [...]
I installed Vista a couple of weeks back, and so far my impressions are very good. For the things I do, it's anywhere from not much slower to much faster. However, I think I see a pattern in all this... Everyone I've talked to who says Vista sucks has used a laptop. Almost all those I've talked to who liked Vista has used a desktop PC. Guess what I'm using. A friend of mine who works at a local Dell tech support office has confirmed my pattern, saying that over 75% of Vista related issues are on laptops.
I guess one reason for this pattern is that more people have bought laptops this year, however from personal experience with a Vista laptop, I think wonky OEM software and drivers is a big part of the problem.
My list of positive things about Vista: - File caching seems much better, and ReadyBoost works like a charm. Reopening apps after gaming is much faster than in XP. All in all this makes some applications MUCH faster to use. - Kernel doesn't make my entire machine seem like a 286 just because an app hogs all the CPU @ normal pri. Even some single threaded apps (most notably VS2005) managed to do that. This only happened once I upgraded to a quad core, and I won't rule out some driver error, but still, was nice to have a smooth system again. - Gfx driver in userspace. Finally games etc won't crash my box (not that it happened that often in XP, but it did). Takes about 10 seconds to recover, and you continue as if nothing happened. - New task manager is a nice, with direct access to the services list for starting/stopping them.
And negative: - Main issue I have is Windows Explorer, which, as usual it seems, is the worst part of Windows. For instance, deleting files is quite slow (it seems to want to scan through the directory before asking the magic yes/no question). And I haven't yet figured out how to disable it's "intelligent view" thing, which makes every folder appear different. Since I'm used to having Detail view all over, it's quite annoying at times. - Can't make it turn off the HDs (power saving), because for some reason it tries to spin then up and down all the continously for some period. Annoying cause I have an archive disk which is a bit noisy, and which I seldom use.
No, I'm fairly sure they didn't. Deconvolution is only applicable to linear systems - this "twirl" filter is non-linear.
I admit I'm no expert, but from what I can see the twirl effect can be modeled as a spatially variant PSF, and from what I can see, deconvolution can handle this.
I've heard from people they feel it's easier to do a lot more things in DirectX than OpenGL. It's all opinion, of course.
While I strongly prefer OpenGL over DirectX, on some points you just can't deny that it's easier with DirectX, being a more extensive library. Sound and multi-media along with 3d and more in "one package". It also has functions for stuff which is complex to implement, ready to be used (for example the Precomputed Radiance Transfer API).
Another reason is that it's easier to verify that you're getting hardware accelerated features, and features that work (ATI has, in my experience, rather spotty OpenGL support). OpenGL 3 will fix the hardware detection issues, and hopefully make it easier to make good OpenGL drivers.
Again, I really like OpenGL, and OpenGL 3 looks very nice indeed, but if you're going for an xbox+pc game, I'd say it's almost a no-brainer to go with DirectX.
I have, several times. Common factor for all occasions: I'm alone, it's dark, and the movie doesn't break the illusion by using some "cheap" special effect monster or similar. Like Session 9, for example.
However I just can't be scared at all if I watch them with others around me. It just doesn't work at all.
It would be nice if the remote software were able to reject commands deemed likely to cause mission failure.
Or perhaps something like what they did to the display resolution dialogs after a while... Ie if communication is lost after a command for X time units, undo the command.
Sorry, but your operation systems FS works in 4k blocks anyways, so you dont actually save anything compared to 4k phys-layer blocks now.
So when you write a 4k page, you get 8 512byte sectors in a row. Fair enough. But what happens when one of the sectors is about to fail? From my understanding, the HD will then move the data to a fresh sector and mark the failing one as bad. If so, you suddenly need to move the head possibly twice while reading that page.
In context, it's clear the GP was referring to the NVidia cards that were reviewed by the article.
I realized this the second I hit submit (even after previewing). I tried to post a reply explaining that I'm a moron, but it seems to have been ignored:)
The Radeon X1950 beats the NVidia cards in every single test save for the "synthetic" crapmark test that has nothing to do with reality.
Interesting, that's not what I've been seeing in tests. In fact, in most tests it seems the 8800 GTX beats the X1950 XTX with a comfortable margin. Here's one review that goes through a number of games: http://www.xbitlabs.com/articles/video/display/gf8 800-games.html
Slashdot Mirror
Since programming wasn't a requirement for the course, each assignment included a "template" Python program where one just had to insert the equations we had found. We were however encouraged to modify it further (use RK4 instead of simple Euler etc).
We then had to find the results analytically and verify the numerical results.
Personally I found this mix very fun and also educational, and I found Python (with the Scitools package) to be good for the task.
I should have specified that I based it on electricity use. You make a good point though.
I just had to try to put that into perspective, so I looked up the net annual usage of electricity in the US. If we had somehow captured, converted and stored all that energy, the US would spend about 45 million years using it up (assuming linear growth similar to the last 10 years, and I didn't screwed up the math).
For instance, there's nothing in the specs that disallows using VBOs as a source when accumulating a display list, but when I tried it, it failed on ATI hardware. Now, in ATI's defense, using VBOs in that way is not what they were made for and one might argue it's a bit silly, but still, it solved a problem I had (until I tested on ATI that is). This is one of those gazillion corner cases that's easy to miss.
The OpenGL forums is filled with people asking why something is slow, or the fastest way of doing something. And reading the web doesn't always clear things up, as what was fast a while ago (triangle strips, for instance) is now not so fast, or in many cases much slower (client side arrays vs VBOs). In most cases you'll end up trying out several ways of doing a thing.
OpenGL 3 was supposed to fix this mess by redesigning the API and cutting away non-essentials, but it's currently 7 months (or so) overdue, and the ARB has adopted the "absolute silence" approach to handling PR. This means that we most likely won't get any (useful) OpenGL 3 drivers until next year or so.
I used to be quite a fan of OpenGL, even though I primarily code for Windows. However if I had to start development of a Windows application today which had to use any non-fixed-function features, I would most certainly not use OpenGL.
So while the experiment may very well involve some new and previously undiscovered phenomena, I think that we can be quite certain that it won't make us modify our current physical framework.
It's a bit like if I told you to figure out the pattern behind a sequence of numbers. You could be fairly sure that I haven't invented some new branch of mathematics to generate it, however that doesn't mean it's immediately obvious to you how I did it.
I bought a new dual-core notebook [...] but damn is this Vista machine SLOW.
It's a laptop thing, that's my theory. Almost everyone I've seen/heard that complains about Vista has tried it on a laptop. Almost everyone I've seen/heard that thinks Vista is fine, uses a desktop PC.
Personally I made the jump two months ago, and while I can still dual boot, I haven't booted XP in 7 weeks. For me, Vista is significantly faster in several areas, while not noticeably slower in the others (with one exception: deleting files).
I can only speculate as to why this is so, but I suspect crappy OEM software and drivers, which is compounded by slower laptop disks and less RAM.
However, last weekend I bought a new laptop when my old one crapped out. Obviously it had Vista, so I tried to use it for a couple of days. Between the fact it was abysmally slow, consumed a gig of memory just sitting there [...]
I installed Vista a couple of weeks back, and so far my impressions are very good. For the things I do, it's anywhere from not much slower to much faster. However, I think I see a pattern in all this... Everyone I've talked to who says Vista sucks has used a laptop. Almost all those I've talked to who liked Vista has used a desktop PC. Guess what I'm using. A friend of mine who works at a local Dell tech support office has confirmed my pattern, saying that over 75% of Vista related issues are on laptops.
I guess one reason for this pattern is that more people have bought laptops this year, however from personal experience with a Vista laptop, I think wonky OEM software and drivers is a big part of the problem.
My list of positive things about Vista:
- File caching seems much better, and ReadyBoost works like a charm. Reopening apps after gaming is much faster than in XP. All in all this makes some applications MUCH faster to use.
- Kernel doesn't make my entire machine seem like a 286 just because an app hogs all the CPU @ normal pri. Even some single threaded apps (most notably VS2005) managed to do that. This only happened once I upgraded to a quad core, and I won't rule out some driver error, but still, was nice to have a smooth system again.
- Gfx driver in userspace. Finally games etc won't crash my box (not that it happened that often in XP, but it did). Takes about 10 seconds to recover, and you continue as if nothing happened.
- New task manager is a nice, with direct access to the services list for starting/stopping them.
And negative:
- Main issue I have is Windows Explorer, which, as usual it seems, is the worst part of Windows. For instance, deleting files is quite slow (it seems to want to scan through the directory before asking the magic yes/no question). And I haven't yet figured out how to disable it's "intelligent view" thing, which makes every folder appear different. Since I'm used to having Detail view all over, it's quite annoying at times.
- Can't make it turn off the HDs (power saving), because for some reason it tries to spin then up and down all the continously for some period. Annoying cause I have an archive disk which is a bit noisy, and which I seldom use.
No, I'm fairly sure they didn't. Deconvolution is only applicable to linear systems - this "twirl" filter is non-linear.
I admit I'm no expert, but from what I can see the twirl effect can be modeled as a spatially variant PSF, and from what I can see, deconvolution can handle this.
Surely Interpol's top-secret image-unblurring technology is just a matter of applying the Twirl effect in the opposite direction at the same location
I'm fairly certain they used deconvolution.
Not being able to play a game because I don't have internet that is Single Player is evil.
Apparently they have this offline mode which allows you to do just that...
I've heard from people they feel it's easier to do a lot more things in DirectX than OpenGL. It's all opinion, of course.
While I strongly prefer OpenGL over DirectX, on some points you just can't deny that it's easier with DirectX, being a more extensive library. Sound and multi-media along with 3d and more in "one package". It also has functions for stuff which is complex to implement, ready to be used (for example the Precomputed Radiance Transfer API).
Another reason is that it's easier to verify that you're getting hardware accelerated features, and features that work (ATI has, in my experience, rather spotty OpenGL support). OpenGL 3 will fix the hardware detection issues, and hopefully make it easier to make good OpenGL drivers.
Again, I really like OpenGL, and OpenGL 3 looks very nice indeed, but if you're going for an xbox+pc game, I'd say it's almost a no-brainer to go with DirectX.
I've never been scared by a movie, ever.
I have, several times. Common factor for all occasions: I'm alone, it's dark, and the movie doesn't break the illusion by using some "cheap" special effect monster or similar. Like Session 9, for example.
However I just can't be scared at all if I watch them with others around me. It just doesn't work at all.
It would be nice if the remote software were able to reject commands deemed likely to cause mission failure.
Or perhaps something like what they did to the display resolution dialogs after a while... Ie if communication is lost after a command for X time units, undo the command.
Does anybody actually use alternative URIs other than http: file: https: mailto: and ftp:?
Personally I use news: a lot, MS HTML help uses ms-help:, and I've found the res: handy as well for some programs.
Kind of like Windows 3.1 being 16bit when it wouldn't run on anything older than a 386 (32bit) anyway.
I ran 3.1 on my 286 12Mhz for years...
And, I guess, if they're shot down, there's no pilot captured by the baddies, which is a tad harder to explain away than a bunch of debris.
I'm continually surprised at how low-spec so many systems are in the Steam surveys.
I know several people who play CS and similar games because it runs on their old hardware.
What if its a software I need to use now and I'm running a compile which takes a whole day to finish?
I think the only app I've installed on this XP box that actually NEEDED the reboot in order to work, was daemon tools.
In my experience, 99% of the time you can just say "No" and go right ahead and use it.
Sorry, but your operation systems FS works in 4k blocks anyways, so you dont actually save anything compared to 4k phys-layer blocks now.
:)
So when you write a 4k page, you get 8 512byte sectors in a row. Fair enough. But what happens when one of the sectors is about to fail? From my understanding, the HD will then move the data to a fresh sector and mark the failing one as bad. If so, you suddenly need to move the head possibly twice while reading that page.
Then again, I'm no HD expert
Or develop for both DX9 and DX10, like Flight Simulator.
Indeed, just like several other games have a DX8 and a DX9 rendering path (STALKER, FEAR, WoW and many others).
In context, it's clear the GP was referring to the NVidia cards that were reviewed by the article.
:)
I realized this the second I hit submit (even after previewing). I tried to post a reply explaining that I'm a moron, but it seems to have been ignored
The Radeon X1950 beats the NVidia cards in every single test save for the "synthetic" crapmark test that has nothing to do with reality.
8 800-games.html
Interesting, that's not what I've been seeing in tests. In fact, in most tests it seems the 8800 GTX beats the X1950 XTX with a comfortable margin. Here's one review that goes through a number of games: http://www.xbitlabs.com/articles/video/display/gf