If the license is void, does that mean you now have no rights, and any use would be a violation of copyright? Isn't that the argument people use to justify the GPL - if it's invalid, you can't use/distribute the software at all?
1) That's a hacky solution. It's better to find the actual cause of the problem. 2) That would increase the time it takes to render a page, and performance is supposed to be a strong point of Gecko (or Firefox).
but then firefox changes the way it renders the page if you twiddle a nob? Shouldn't firefox consistently render it the same (broken) way every time?
In theory, yes. Unfortunately, there's a class of bugs called "reflow" bugs - reflow is basically the incremental rendering of pages as more of the HTML is downloaded.
When certain things happen at certain times, in certain orders, the layout ends up getting rendered incorrectly until you force a reflow (you can do this by changing the text size, resizing the window, etc).
The problem with these bugs is that they're very hard to track down. A lot of the time, you can't reproduce them on a [faster|slower] connection, and if the developers aren't experiencing it, they're stuck.
That's a really pathetic excuse - Mozilla is at 1.7.x (1.8 for trunk development), and the bugs are shared. Justifying holes with "oh, we haven't reached 1.0 yet" will just come back to bite you when 1.0 is released and more holes are discovered. Heck, Netscape is at version 7.2 and it is likely to share these holes.
Justify them as "we try hard to find them and fix them quickly", but not "they'll go away when we reach 1.0".
I've heard many people say this, but never anyone in the industry. It seems to me that if you had exceptional yeilds for high-speed parts, why not just drop the high-speed price and eat up the market share of your competitor?
CPUs are speed binned by the manufacturer based on rigourous tests done in worst-case conditions (highest allowed temperature, lowest voltage).
There are 3 things that let you overclock in normal situations: 1. If the CPU works at 2.99GHz, but not 3.0GHz, it has to be sold as one speed grade down. This CPU would be perfectly stable up to 2.99GHz. 2. If the environment you run in is not in the worst-case corner (you keep it cool, with good power supplied to the CPU), you'll be able to get a few extra percent. 3. When the manufacturer tests the CPU, they know all the worst-case instruction sequences and critical paths. When an overclocker does a stability test, it's extremely likely that they're missing various speed paths, and eventually something WILL use one of those paths, and you get data corruption. Using games as tests and seeing if they crash is absolutely not thorough - if every floating point operation was coming out slightly incorrect, you probably wouldn't notice, but the CPU is in fact not operating properly. Why is it that overclockers with "perfectly stable" overclocks always seem to end up having more apps crashing / more problems with "Windows sucking"?
If an OEM wants to sell a reliable machine, they'd have to do all the testing the CPU manufacturer does - the only thing they could do is guarantee a better max temperature/minimum voltage, but why bother? They're likely to gain at best 5% performance for significantly more effort.
There is a research paper here that gives a lot more information than the article linked (ironically enough, I happened to be reading it yesterday). They address many of the issues people have brought up (alignment, dust, etc.), and the paper really isn't a hard read.
They actually have a bunch of interesting papers in the parent directory here, mostly covering stuff about asynchronous/clockless computing.
It was ignored largely because a nearly-as-convincing fake can be created with DHTML, so blocking XUL doesn't add any real security. How many people really use different themes (so the DHTML fake would look wrong)?
I don't think it's that they didn't allow them, it's that the thunderbirds used different signalling than the original Athlons. Motherboards based on the AMD-750 (Irongate) chipset did work reliably with thunderbirds, but it was hit or miss with the Via KX133 (an otherwise better, faster chipset). There is some info on that here,
I don't know if the rest of the pinout was compatible enough to allow it either.
You do realize that there ARE disadvantages of moving to 90nm? In the past, a process shrink pretty much guaranteed large power savings, and a big speed boost. However, we're reaching a point where each new process brings with it SIGNIFICANT increases in device leakage, which can kill any potential power reduction, and the speed boosts aren't as great as they used to be.
The majority of IE-only problems I run into can be fixed by replacing "document.all.foo.*" or "document.foo.*" with "document.getElementByName('foo').*
RTFBug. Since MS decided programs should be able to register protocol handlers (e.g. irc://, telnet://), Mozilla behaves like a good little windows program, and passes any unknown protocols (shell://, vbscript://) to the OS. It's a flaw in the whole setup that windows uses here, and MS changed the behavior for XP SP2.
The problem comes from my friends. One morning, I sit down in front of my PC, boot it. Something comes up full-screen, immediately. I've been spywared. By no fault of my own. My less-than-savvy friends have just cost me an hour of my time downloading, updating and running AdAware/SpyBot S&D.
Perhaps you shouldn't have let your friends log in with an administrator account.
I think the problem is that there's no good indicator of performance for a non-computer person.
For a car, horsepower will give you a ballpark idea of its performance, and with 4 numbers (hp, torque, displacement, weight), you could probably get a very good idea of performance, there is no such analogy with computers.
A 1GHz CPU, backed up by the latest-and-greatest from ATI or nVidia will slaughter a 3GHz CPU backed up by an old ATI Rage128 or TNT2. To get a decent idea of a computer's performance, you need to look at: 1. CPU speed 2. RAM amount (ram speed doesn't usually have TOO much effect on performance) 3. HD size (not really performance-related, but important) 4. HD RPMs (or seek time) 5. Video card performance
Ignoring the fact that it's a lot of numbers, the real problem here is #5. I often see system requirements listed as "32MB video card" or "64MB video card". That's a ridiculous requirement, since you can buy a 64MB GeForce2, and a 32MB GeForce4... the GF4 will be faster in pretty much every situation. You can't even use model numbers... ATI's Radeon model numbers (8500, 9000, 9100, 9200, 9600, 9700, 9800) could be informative, if not for the fact that each card comes in "Pro", "SE", regular, and "AIW" editions... and a Sucky Edition 9800 will get its butt kicked by a 9600 Pro. nVidia's lineup is equally confusing.
With the cheating that has been going on in ATI and nVidia drivers, you can't even cite a benchmark (3dmark or ATI's "quack3"), and even if you did, do you really think manufacturers are going to be clear about scores? Or what a score of 10,000 means in 3dmark2003 vs 3dmark2001?
Realistically, you usually don't max out a fast connection with standard protocols. I've found that bittorrent does a good job, but guess what? It doesn't do sequential read/writes, but instead has a VERY random access pattern. My drive thrashes with bittorrent on just a 10Mbps connection - I really doubt you could sustain 100Mbps speeds with bittorrent accessing random locations across a 1gb file...
No other protocols usually offer large enough content (CD ISOs, movies, etc) that 100Mbps is much faster, and when they do, the server rarely has enough bandwidth to max out a 100Mbps connection.
As I understand it, you can't build a 4-layer motherboard with socket 940 (you have to use 6 layers). By changing the pin layout for socket 939, it is now possible to use 4 layers. This should reduce the cost of motherboards.
The cell phone companies are an oligopoly. There are only a few, very big players, and they're all equally evil. If you didn't want to buy a car from Ford, GM, Toyota, Saturn, or Honda, that leaves you with a VERY small selection of cars you could buy. Taking your business elsewhere only really works in markets with many players.
Power = C*V^2*F. C is a constant (capacitance switched), V is voltage, F is frequency. The 1GHz part runs at 1V, and is 6 watts typical: 6 Watts = C*1v*1v*1000000000hz C = 6/1000000000 13.125 = C*1.25v*1.25v*1400000000Hz = C*1.56*1400000000Hz
Since they're the same core, the factor C is the same. The reason it isn't exactly 14 watts is most likely the static (leakage) power... even when nothing is switching, a small amount of current is flowing, just producing heat.
These processors work in standard Socket A motherboards, so long as the motherboard can supply a core voltage as low as 1 volt. You don't really need a "desktop" version. It's up to marketing to determine what to call "embedded", "mobile", or desktop/consumer/server.
Geodes aren't really aimed at consumers, but rather manufacturers of products like printers, handheld computers, point-of-sale systems, thin clients, set-top boxes, etc.
Slashdot Mirror
If the license is void, does that mean you now have no rights, and any use would be a violation of copyright? Isn't that the argument people use to justify the GPL - if it's invalid, you can't use/distribute the software at all?
1) That's a hacky solution. It's better to find the actual cause of the problem.
2) That would increase the time it takes to render a page, and performance is supposed to be a strong point of Gecko (or Firefox).
but then firefox changes the way it renders the page if you twiddle a nob? Shouldn't firefox consistently render it the same (broken) way every time?
In theory, yes. Unfortunately, there's a class of bugs called "reflow" bugs - reflow is basically the incremental rendering of pages as more of the HTML is downloaded.
When certain things happen at certain times, in certain orders, the layout ends up getting rendered incorrectly until you force a reflow (you can do this by changing the text size, resizing the window, etc).
The problem with these bugs is that they're very hard to track down. A lot of the time, you can't reproduce them on a [faster|slower] connection, and if the developers aren't experiencing it, they're stuck.
That's a really pathetic excuse - Mozilla is at 1.7.x (1.8 for trunk development), and the bugs are shared. Justifying holes with "oh, we haven't reached 1.0 yet" will just come back to bite you when 1.0 is released and more holes are discovered. Heck, Netscape is at version 7.2 and it is likely to share these holes.
Justify them as "we try hard to find them and fix them quickly", but not "they'll go away when we reach 1.0".
Has anyone in industry SAID most Celeron 300As were down-binned 450s, or did they just overclock really well?
I've heard many people say this, but never anyone in the industry. It seems to me that if you had exceptional yeilds for high-speed parts, why not just drop the high-speed price and eat up the market share of your competitor?
CPUs are speed binned by the manufacturer based on rigourous tests done in worst-case conditions (highest allowed temperature, lowest voltage).
There are 3 things that let you overclock in normal situations:
1. If the CPU works at 2.99GHz, but not 3.0GHz, it has to be sold as one speed grade down. This CPU would be perfectly stable up to 2.99GHz.
2. If the environment you run in is not in the worst-case corner (you keep it cool, with good power supplied to the CPU), you'll be able to get a few extra percent.
3. When the manufacturer tests the CPU, they know all the worst-case instruction sequences and critical paths. When an overclocker does a stability test, it's extremely likely that they're missing various speed paths, and eventually something WILL use one of those paths, and you get data corruption. Using games as tests and seeing if they crash is absolutely not thorough - if every floating point operation was coming out slightly incorrect, you probably wouldn't notice, but the CPU is in fact not operating properly. Why is it that overclockers with "perfectly stable" overclocks always seem to end up having more apps crashing / more problems with "Windows sucking"?
If an OEM wants to sell a reliable machine, they'd have to do all the testing the CPU manufacturer does - the only thing they could do is guarantee a better max temperature/minimum voltage, but why bother? They're likely to gain at best 5% performance for significantly more effort.
This was one of the security fixes (arguably the only exploitable hole) that was included in yesterday's releases, 1.7.2, 0.9.3, and 0.7.3.
There is a research paper here that gives a lot more information than the article linked (ironically enough, I happened to be reading it yesterday). They address many of the issues people have brought up (alignment, dust, etc.), and the paper really isn't a hard read.
They actually have a bunch of interesting papers in the parent directory here, mostly covering stuff about asynchronous/clockless computing.
It was ignored largely because a nearly-as-convincing fake can be created with DHTML, so blocking XUL doesn't add any real security. How many people really use different themes (so the DHTML fake would look wrong)?
I don't think it's that they didn't allow them, it's that the thunderbirds used different signalling than the original Athlons. Motherboards based on the AMD-750 (Irongate) chipset did work reliably with thunderbirds, but it was hit or miss with the Via KX133 (an otherwise better, faster chipset). There is some info on that here,
I don't know if the rest of the pinout was compatible enough to allow it either.
You do realize that there ARE disadvantages of moving to 90nm? In the past, a process shrink pretty much guaranteed large power savings, and a big speed boost. However, we're reaching a point where each new process brings with it SIGNIFICANT increases in device leakage, which can kill any potential power reduction, and the speed boosts aren't as great as they used to be.
The majority of IE-only problems I run into can be fixed by replacing "document.all.foo.*" or "document.foo.*" with "document.getElementByName('foo').*
RTFBug. Since MS decided programs should be able to register protocol handlers (e.g. irc://, telnet://), Mozilla behaves like a good little windows program, and passes any unknown protocols (shell://, vbscript://) to the OS. It's a flaw in the whole setup that windows uses here, and MS changed the behavior for XP SP2.
Try Win+R and Win+D
Win+D and Win+M are slightly different in their behaviors.
The problem comes from my friends. One morning, I sit down in front of my PC, boot it. Something comes up full-screen, immediately. I've been spywared. By no fault of my own. My less-than-savvy friends have just cost me an hour of my time downloading, updating and running AdAware/SpyBot S&D.
Perhaps you shouldn't have let your friends log in with an administrator account.
I think the problem is that there's no good indicator of performance for a non-computer person.
For a car, horsepower will give you a ballpark idea of its performance, and with 4 numbers (hp, torque, displacement, weight), you could probably get a very good idea of performance, there is no such analogy with computers.
A 1GHz CPU, backed up by the latest-and-greatest from ATI or nVidia will slaughter a 3GHz CPU backed up by an old ATI Rage128 or TNT2. To get a decent idea of a computer's performance, you need to look at:
1. CPU speed
2. RAM amount (ram speed doesn't usually have TOO much effect on performance)
3. HD size (not really performance-related, but important)
4. HD RPMs (or seek time)
5. Video card performance
Ignoring the fact that it's a lot of numbers, the real problem here is #5. I often see system requirements listed as "32MB video card" or "64MB video card". That's a ridiculous requirement, since you can buy a 64MB GeForce2, and a 32MB GeForce4... the GF4 will be faster in pretty much every situation. You can't even use model numbers... ATI's Radeon model numbers (8500, 9000, 9100, 9200, 9600, 9700, 9800) could be informative, if not for the fact that each card comes in "Pro", "SE", regular, and "AIW" editions... and a Sucky Edition 9800 will get its butt kicked by a 9600 Pro. nVidia's lineup is equally confusing.
With the cheating that has been going on in ATI and nVidia drivers, you can't even cite a benchmark (3dmark or ATI's "quack3"), and even if you did, do you really think manufacturers are going to be clear about scores? Or what a score of 10,000 means in 3dmark2003 vs 3dmark2001?
Realistically, you usually don't max out a fast connection with standard protocols. I've found that bittorrent does a good job, but guess what? It doesn't do sequential read/writes, but instead has a VERY random access pattern. My drive thrashes with bittorrent on just a 10Mbps connection - I really doubt you could sustain 100Mbps speeds with bittorrent accessing random locations across a 1gb file...
No other protocols usually offer large enough content (CD ISOs, movies, etc) that 100Mbps is much faster, and when they do, the server rarely has enough bandwidth to max out a 100Mbps connection.
Andrew Linux, used at Carnegie Mellon, is heavily network-based.
As I understand it, you can't build a 4-layer motherboard with socket 940 (you have to use 6 layers). By changing the pin layout for socket 939, it is now possible to use 4 layers. This should reduce the cost of motherboards.
That makes my point even better ;). I thought they were supposed to be "a different kind of car company".
Don't do business with them? Tell your friends?
The cell phone companies are an oligopoly. There are only a few, very big players, and they're all equally evil. If you didn't want to buy a car from Ford, GM, Toyota, Saturn, or Honda, that leaves you with a VERY small selection of cars you could buy. Taking your business elsewhere only really works in markets with many players.
Power = C*V^2*F. C is a constant (capacitance switched), V is voltage, F is frequency. The 1GHz part runs at 1V, and is 6 watts typical:
6 Watts = C*1v*1v*1000000000hz
C = 6/1000000000
13.125 = C*1.25v*1.25v*1400000000Hz = C*1.56*1400000000Hz
Since they're the same core, the factor C is the same. The reason it isn't exactly 14 watts is most likely the static (leakage) power... even when nothing is switching, a small amount of current is flowing, just producing heat.
These processors work in standard Socket A motherboards, so long as the motherboard can supply a core voltage as low as 1 volt. You don't really need a "desktop" version. It's up to marketing to determine what to call "embedded", "mobile", or desktop/consumer/server.
Geodes aren't really aimed at consumers, but rather manufacturers of products like printers, handheld computers, point-of-sale systems, thin clients, set-top boxes, etc.