The size of the install for Alias Sketchbook 1.1.1 is 11.1 Meg. And you could easily trim 2.1 Meg off that by deleting the Japanese help and Japanese resource dll.
And it has some pretty lightweight minimum hardware requirements -- 400 Mhz P2 with 256 Meg of Ram and a 1024x768 24 bit color display, but you'll want a wacom tablet of some form plugged in.
So it should be pretty portable...
I've tried 'em all. Intels is good, but when it comes to toe sort of code I was talking about in my original post, you can still get a factor of 2 or 3 over whan it can do with scalar code, if you havd code a vectorized MMX or SSE version.
I did explicitly say "if it's vectorizable", and "particularly if it uses small integer saturated math".
I doubt any autovectorization will recognize the conditionals necessary for saturated math and make efficient use of MMX instead.
(I worked on the back end of an optimizing compiler at IBM for 6 years, so I know a bit about how compilers work:-)
IBM has some pretty rockin good fortran compilers for their machines. Why do you think the Power5 benches so well on SpecFP? (The chip is pretty awesome, I admit, but it needs a good compiler to get the last bit of performance out of it)
The IBM XL compilers are just amazing.
I know you think you're joking, but you're actually right -- that's what the good programmers do. In my application, the performance critical routines (on the order of a dozen or so) are hand coded in vector assembler (Altivec on Mac, and MMX and SSE on Intel), and are about 3 to 4 times faster than the most optimized C or C++ implementation of those algorithms.
If you have code that is vectorizable, and especially if it is doing saturated small integer math, (blending, resampling etc), you can do way better than any existing compilers by hand coding in assembler.
The officials involved and the legislators too, probably, are looking and seeing a wealthy foriegn organization that wants them to take action which will result in money flowing out of the motherland. That will take some serious bribes to bring about that outcome.
You (and the moderators too) think you're joking. But I'm not. If, and as soon as, that capability is available in any widespread manner on a tablet PC (or a, purely hypothetical, tablet mac), Alias Sketchbook will exploit it in the way you describe. (There will probably be a preference to turn it off and/or adjust it's sensitivity. Wouldn't want it to happen accidentally.)
Cheers
It doesn't matter if you use --ansi or -wall, and ignore the warnings -- independent of compiler options, the above code will not work reliably across different compilers. And possibly not work the same with different options (think optimization) on the same compiler. Such blatant ignorance of one of the most important and common pitfalls of the language sends up a very big red flag for me.
And whether the question called for ansi-c, the solution was clearly written in c, and therefore it behooves the author of the code to ensure that it conforms to the relevant standards for that language -- especially with regards to undefined behaviour.
I'm afraid I can't hire you with that answer.
While shorter, it is clearly undefined in C/C++.
I quote (from memory) the relevant ANSI/ISO standard;
"Between the previous and next sequence point an object shall have its stored value modified at most once by the evaluation of an expression. Furthermore, the prior value shall be accessed only to determine the value to be stored."
You are modifying 'a' twice between sequence points. And NO, an assignment operator is NOT a sequence point.
I've found the Intel compilers to be truly excellent -- Both the front-end and back-end of them.
They do a great job on even the most complex templated metaprogrammed stuff -- code that makes the MS compiler fall over with an internal compiler error.
And their code generation is way ahead of everyone else.
And they listen carefully to bug reports and fix bugs on a regular basis. I've never seen a bug fix from MS on their compilers.
Don't generalize. Just because *you* can't do something or understand something doesn't similarly constrain the rest of us.
So you may not be able to tell by looking at a die photo, but others who know what to look for can definitely tell.
For a good example, go look at Hans DeVries excellent ChipArchitect website, specifically;
http://chip-architect.com/news/2003_04_20_Looking_ at_Intels_Prescott_part2.html
He also has an awesome analysis of the AMD chips.
Getting a good LCD monitor is simple -- I did it for my wife -- I went to a store (future shop), told the salsdriod what unit I wanted, and that I wanted to see it work with my laptop. I had him bring one out -- I opened the box, plugged it into my laptop, and what do you know -- 2 bad pixels -- I had him get another, and another (the third one was OK.)
I just refused to hand over any money until I saw that the unit I was buying had no bad pixels.
I call bullshit on that.
It first appeared in the prescott core. Die photos of northwood and earlier processors clearly do not have the room for AMD 64 bit extensions.
Sorry about the shameless plug, but (at the risk of slashdotting my company's servers) here's a link to the Sketchbook Home Page;
http://www.aliassketchbookpro.com
Actually that's not correct -- SGI had the MIPS Origin machines working in house before the aquisition of Cray. They called the interconnect "Cray Link" but it had nothing to do with Cray and everything to do with some pointy haired MBA bean counter type trying to "brand" something.
This said, most of the post above is correct, but one problem with scaling the opteron up to such large systems is the physical address space in current implementations. It's 40 bits. For large ccNUMA machines, you really want a minimum of 48 bits of physical address space.
The other thing the hub chip does on SGI machines is to handle the page directory stuff, counting remote vs local accesses, and handling automatically migrating pages to nodes that are making heavy use of them. Most of this cool stuff is patented, so I doubt the Opteron is doing it.
After a bit more research and reading the CARs, I see that you can be issued a restricted medical with color blindness. But you will be "restricted day flying only, requiring a two-way radio to fly in and out of controlled airports"
When I fly now, yes I do have a hand-held radio. Didn't that night though -- and it was real -- and yes, I was man enough, thank you:-) It was at a controlled field with the lights on. (At this field when the tower closes, the FSS takes over -- they're manned 24/7 -- and the lights stay on.) Failure happened just after I was cleared to join the downwind.
BTW I'm in Canada. The rules are slightly different here...
In Canada, Night is a seperate rating for your licence.
The requirements are; Private Pilot licence Class 1 or 3 medical certificate 20 hours, which includes: - 10 hour dual instrument - 5 hours dual night (2 hours of cross-country) - 5 hours solo night (10 take offs and landings)
And in Canada, you can't get an unrestricted medical certificate if you're color blind and can't demonstrate that you can discern the light signals. The wording on the restriction printed on your medical will be "restricted day flying only, requiring a two-way radio to fly in and out of controlled airports". And you will therefore not be allowed to hold a night rating.
Final authority on this here is the Canadian Air Regulations; 421.42, 401.42 and 401.43.
I stand by my statment -- flying color blind at night is asking for trouble.
Let me be direct -- You, sir, are *not* a safe pilot if you are flying at night colorblind. I hope that when (not if) this catches up with you that you are not carrying passengers.
As I mentioned in another post, this applies to the USA only. I'm pretty confident that in Canada colorblindness is disqualifying.
That said I had a friend who failed the eye test for colorblindness, but passed a SODA -- went to the far side of the airfield (about 7,000 ft from the tower) with a handheld, and told them what he saw.
So you're a pilot too -- ever had a power failure? A radio failure? (I've experienced both -- including power at night -- had to land holding a flashlight in my mouth so I could read the airspeed). You can't get an avaiation medical certificate if you are colorblind, and for good reason.
As for the colors on the ground -- don't want to land on a taxi-way, now, do we? (not all runways have center lights or strobed threshold lights or christmas trees, and if blue and white look the same to you, it can be easy to mistake the taxiway for the runway -- hell it's happened to people who can tell the difference, but who are tired).
As for telling the difference -- remember your light-gun signals for when your radio dies? (And yes, during that power failure, obviously the radios were not working too well). You know; red, green, white. Quick question -- what does flashing red mean when you're in the air? On the ground? You shold know them all without looking it up. (On short final in a dark cockput with a flashlight in your mouth, left hand on the control column, right on the power, flying the plane (compensating for a crosswind), watching for the light signals from the tower, is *NOT* a good time to be looking up things like this -- even if it is printed on the cheat-sheet on your kneeboard under a stack of other paperwork.)
The long and the short of it is that flying at night color blind is just asking for trouble.
If there's a radio failure, the control tower uses light signals -- under ordinary circumstances, you need to remember that airport lights (runway, taxiway, etc) are color coded. As a pilot, you *must* be able to tell the difference between red, green, yellow, blue and white lights.
Gabe at Penny Arcade uses Alias Sketchbook Pro (http://www.aliassketchbookpro.com) on a tablet PC for some of his work. I'm always amazed at what talented people can do with the software I write.
(Sketchbook also runs on a pc or a mac with a wacom tablet -- sorry -- no linux version.)
Slashdot Mirror
The size of the install for Alias Sketchbook 1.1.1 is 11.1 Meg. And you could easily trim 2.1 Meg off that by deleting the Japanese help and Japanese resource dll. And it has some pretty lightweight minimum hardware requirements -- 400 Mhz P2 with 256 Meg of Ram and a 1024x768 24 bit color display, but you'll want a wacom tablet of some form plugged in. So it should be pretty portable...
I've tried 'em all. Intels is good, but when it comes to toe sort of code I was talking about in my original post, you can still get a factor of 2 or 3 over whan it can do with scalar code, if you havd code a vectorized MMX or SSE version.
I did explicitly say "if it's vectorizable", and "particularly if it uses small integer saturated math". I doubt any autovectorization will recognize the conditionals necessary for saturated math and make efficient use of MMX instead. (I worked on the back end of an optimizing compiler at IBM for 6 years, so I know a bit about how compilers work :-)
IBM has some pretty rockin good fortran compilers for their machines. Why do you think the Power5 benches so well on SpecFP? (The chip is pretty awesome, I admit, but it needs a good compiler to get the last bit of performance out of it) The IBM XL compilers are just amazing.
I know you think you're joking, but you're actually right -- that's what the good programmers do. In my application, the performance critical routines (on the order of a dozen or so) are hand coded in vector assembler (Altivec on Mac, and MMX and SSE on Intel), and are about 3 to 4 times faster than the most optimized C or C++ implementation of those algorithms. If you have code that is vectorizable, and especially if it is doing saturated small integer math, (blending, resampling etc), you can do way better than any existing compilers by hand coding in assembler.
The only bad publicity is an obituary -- and even then only if you're actually dead.
The officials involved and the legislators too, probably, are looking and seeing a wealthy foriegn organization that wants them to take action which will result in money flowing out of the motherland. That will take some serious bribes to bring about that outcome.
You (and the moderators too) think you're joking. But I'm not. If, and as soon as, that capability is available in any widespread manner on a tablet PC (or a, purely hypothetical, tablet mac), Alias Sketchbook will exploit it in the way you describe. (There will probably be a preference to turn it off and/or adjust it's sensitivity. Wouldn't want it to happen accidentally.) Cheers
I can't believe nobody's jumped on the opportunity :-)
In Soviet Russia, Cars drive you! :-)
It doesn't matter if you use --ansi or -wall, and ignore the warnings -- independent of compiler options, the above code will not work reliably across different compilers. And possibly not work the same with different options (think optimization) on the same compiler. Such blatant ignorance of one of the most important and common pitfalls of the language sends up a very big red flag for me. And whether the question called for ansi-c, the solution was clearly written in c, and therefore it behooves the author of the code to ensure that it conforms to the relevant standards for that language -- especially with regards to undefined behaviour.
I'm afraid I can't hire you with that answer. While shorter, it is clearly undefined in C/C++. I quote (from memory) the relevant ANSI/ISO standard; "Between the previous and next sequence point an object shall have its stored value modified at most once by the evaluation of an expression. Furthermore, the prior value shall be accessed only to determine the value to be stored." You are modifying 'a' twice between sequence points. And NO, an assignment operator is NOT a sequence point.
Thanks for the kind words.
I've found the Intel compilers to be truly excellent -- Both the front-end and back-end of them. They do a great job on even the most complex templated metaprogrammed stuff -- code that makes the MS compiler fall over with an internal compiler error. And their code generation is way ahead of everyone else. And they listen carefully to bug reports and fix bugs on a regular basis. I've never seen a bug fix from MS on their compilers.
Don't generalize. Just because *you* can't do something or understand something doesn't similarly constrain the rest of us. So you may not be able to tell by looking at a die photo, but others who know what to look for can definitely tell. For a good example, go look at Hans DeVries excellent ChipArchitect website, specifically; http://chip-architect.com/news/2003_04_20_Looking_ at_Intels_Prescott_part2.html
He also has an awesome analysis of the AMD chips.
Getting a good LCD monitor is simple -- I did it for my wife -- I went to a store (future shop), told the salsdriod what unit I wanted, and that I wanted to see it work with my laptop. I had him bring one out -- I opened the box, plugged it into my laptop, and what do you know -- 2 bad pixels -- I had him get another, and another (the third one was OK.) I just refused to hand over any money until I saw that the unit I was buying had no bad pixels.
I call bullshit on that. It first appeared in the prescott core. Die photos of northwood and earlier processors clearly do not have the room for AMD 64 bit extensions.
Sorry about the shameless plug, but (at the risk of slashdotting my company's servers) here's a link to the Sketchbook Home Page; http://www.aliassketchbookpro.com
Actually that's not correct -- SGI had the MIPS Origin machines working in house before the aquisition of Cray. They called the interconnect "Cray Link" but it had nothing to do with Cray and everything to do with some pointy haired MBA bean counter type trying to "brand" something. This said, most of the post above is correct, but one problem with scaling the opteron up to such large systems is the physical address space in current implementations. It's 40 bits. For large ccNUMA machines, you really want a minimum of 48 bits of physical address space. The other thing the hub chip does on SGI machines is to handle the page directory stuff, counting remote vs local accesses, and handling automatically migrating pages to nodes that are making heavy use of them. Most of this cool stuff is patented, so I doubt the Opteron is doing it.
After a bit more research and reading the CARs, I see that you can be issued a restricted medical with color blindness. But you will be "restricted day flying only, requiring a two-way radio to fly in and out of controlled airports"
When I fly now, yes I do have a hand-held radio. Didn't that night though -- and it was real -- and yes, I was man enough, thank you :-)
It was at a controlled field with the lights on. (At this field when the tower closes, the FSS takes over -- they're manned 24/7 -- and the lights stay on.) Failure happened just after I was cleared to join the downwind.
BTW I'm in Canada. The rules are slightly different here...
In Canada, Night is a seperate rating for your licence.
The requirements are;
Private Pilot licence
Class 1 or 3 medical certificate
20 hours, which includes:
- 10 hour dual instrument
- 5 hours dual night (2 hours of cross-country)
- 5 hours solo night (10 take offs and landings)
And in Canada, you can't get an unrestricted medical certificate if you're color blind and can't demonstrate that you can discern the light signals. The wording on the restriction printed on your medical will be "restricted day flying only, requiring a two-way radio to fly in and out of controlled airports". And you will therefore not be allowed to hold a night rating.
Final authority on this here is the Canadian Air Regulations; 421.42, 401.42 and 401.43.
I stand by my statment -- flying color blind at night is asking for trouble.
Let me be direct -- You, sir, are *not* a safe pilot if you are flying at night colorblind. I hope that when (not if) this catches up with you that you are not carrying passengers.
As I mentioned in another post, this applies to the USA only. I'm pretty confident that in Canada colorblindness is disqualifying.
That said I had a friend who failed the eye test for colorblindness, but passed a SODA -- went to the far side of the airfield (about 7,000 ft from the tower) with a handheld, and told them what he saw.
Not in Canada. In Canada, you can't get a class 3 medical if you're color blind. And no medical, no licence.
So you're a pilot too -- ever had a power failure? A radio failure? (I've experienced both -- including power at night -- had to land holding a flashlight in my mouth so I could read the airspeed). You can't get an avaiation medical certificate if you are colorblind, and for good reason.
As for the colors on the ground -- don't want to land on a taxi-way, now, do we? (not all runways have center lights or strobed threshold lights or christmas trees, and if blue and white look the same to you, it can be easy to mistake the taxiway for the runway -- hell it's happened to people who can tell the difference, but who are tired).
As for telling the difference -- remember your light-gun signals for when your radio dies? (And yes, during that power failure, obviously the radios were not working too well). You know; red, green, white. Quick question -- what does flashing red mean when you're in the air? On the ground? You shold know them all without looking it up. (On short final in a dark cockput with a flashlight in your mouth, left hand on the control column, right on the power, flying the plane (compensating for a crosswind), watching for the light signals from the tower, is *NOT* a good time to be looking up things like this -- even if it is printed on the cheat-sheet on your kneeboard under a stack of other paperwork.)
The long and the short of it is that flying at night color blind is just asking for trouble.
If there's a radio failure, the control tower uses light signals -- under ordinary circumstances, you need to remember that airport lights (runway, taxiway, etc) are color coded. As a pilot, you *must* be able to tell the difference between red, green, yellow, blue and white lights.
(Yes, I am a pilot)
Gabe at Penny Arcade uses Alias Sketchbook Pro (http://www.aliassketchbookpro.com) on a tablet PC for some of his work. I'm always amazed at what talented people can do with the software I write.
(Sketchbook also runs on a pc or a mac with a wacom tablet -- sorry -- no linux version.)
Ian Ameline,
Alias Sketchbook Tech Lead.