MEP McCarthy said in a June analysis of the proposed directive that there were links between the patentability of computer-related inventions and the growth of IT industries in the United States. Such patents aided "in particular the growth of small and medium enterprises and independent software developers," she wrote, citing a study on the issue carried out for the European Parliament by London's Intellectual Property Institute.
Roughly analogous to the tobacco industry carrying out a study on the risk of cancer from smoking.
Does anybody have a copy of this study, on which Ms. McCarthy relies?
but thas because software patents benefit large corporations
That has nothing to do with U.S. software companies being big and rich. For the most part, they are big and rich because of copyright. No other protection is needed for the kind of software that allows companies to grow big and rich, its complexity ensures that.
It has been suggested that the Parliament's report will for the first time allow the patentability of computer-implemented inventions. This is simply not true. The patenting of computer-implemented inventions is not a new phenomenon. Patents involving the use of software have been applied for and granted since the earliest days of the European Patent Office (EPO).
Is he intentionally twisting the truth by implying that software patents are valid in Europe, when they aren't, or that the EPO has followed the letter and spirt of the law in granting those patents, when it clearly hasn't? One way to approach this is to give him the benefit of the doubt, and explain to him why his statement is incorrect. Give him a chance to prove his honesty. If he proves otherwise, at least we know where he stands, and perhaps his constituents would like to know as well.
In any event, perhaps it would be possible to find out something about his affiliations, does he have any obvious vested interest that would prompt him to distort the facts in this particular way? Is there any such thing as a public record of campaign contributions for an MEP? Do you know anybody who knows him personally?
At a time when many of our traditional industries are migrating to Asia and when Europe needs increasingly to rely on its inventiveness to reap rewards, it is important to have the option of the revenue secured by patents and the licensing out of computer-implemented technologies.
Glibly overlooking the fact that the net cost to the public of patents, in the form of lack of availability of free software, will certainly exceeds the revenue to be earned by the owners of patents. Furthermore software development in the EU will slow down, denying European software companies an advantage they could have had. If he wants to hand the advantage to Asia, he has certainly set out on the right path to accomplish it.
Perhaps he needs some help understanding the fact that there is no shortage of evidence that copyright provides all the protection nessary for a software company to survive and prosper.
ISDN - very expensive to start with (+- R2000-00 initial startup @ R 7.50 / $1) then you still pay for the call charges. If the config goes haywire you can end up with a bill of R 4000-00/month. ADSL is only available in certain areas - but there is a 3gig monthly cap. some guys can go through that in a day if they wanted to, and the service is being oversubscribed so quickly that the transfer rates are becoming dysmal. The only advantage is the 24x7 online connectivity (although they say that this is not guaranteed)
So most subscribers pay for 56k access (and we do pay for every local call made)
That pretty much describes the situation in Germany, up until two years ago. Barbaric. The weird thing is, the locals just never understood how badly Deutsche Telekom was abusing them until broadband came along and made it perfectly obvious how far the country was going to fall behind if things didn't change fast.
Local calls still carry toll charges, it's so stupid, as if the bandwidth needed to carry a voice call actually cost anything measurable these days. It's also still pretty much impossible to get flat rate 24/7 dialup access, as Canada has had since about 1995. So if you aren't in an urban area, you are still a digital hillbilly, you connect to the internet only on special occasions.
2. uClinux is not readily hackable, at least until you drift of it, and also know how to recover when this thing freezes. You can not just dive into it as if it was a linux PC.
You have never worked on any embedded system, for pay or otherwise.
are almost pointless, a 56k connection is bad enough without it being shared across several computers.
Oh, you won't be able to surf your pr0N, but it's good enough for email and irc, when your broadband goes down (it happens, and chances are, your POTS will still be there).
But that's not the point. The point is, you can run your own code on this thing. One application I've always wanted: a fanless unit sits on my dsl connection, always on, waits for me to connect from overseas, say, and powers up my server with wake-on-lan.
Since you have falsely accused me of plagiarism I have a hard time taking lessons from you on 'courtesy', but I will be a good sport and I will try to rephrase my future submissions to Slashdot if others agree with your position.
His position is about as nonexistent as a position gets. Please save your valuable time for something useful.
They do cheat, yes. There is no need to break backwards compatability, in fact the protocol already has what's needed, it's mostly a matter of XFree engineering and getting it effecient enough to not kill performance.
It will be pitifully slow without major changes to the rendering model. Window buffers are one approach, but there's a better way to do it: retained mode graphics where each window buffers a list of graphics primitives that are re-rendered on window expose or any visibility event, such as lying underneath a translucent window that changed. The primitives can be sorted into, e.g., a quadtree, to make partial exposes of complex windows efficient. This can be done with minimal impact on applications.
There are big advantages to this approach:
- Memory efficient
- Relatively processor-efficient, especially if accelerated with a 3D render card
- Subpixel rendering - buffered window bitmaps don't support that
- Easily extended to use standard 3D visibility and culling algorithms
As much as I love Open-Source/Free Software, it IS very very hard to make a living making it. I'm not talking about writing code for an employer, and then getting permission to release it; I'm talking about actually making your living DIRECTLY off of making, releasing and "selling" open source/free software, a la Red Hat (who just recently turned a profit for the first time). I'd love to hear some more stories from people who've actually made money by coding OSS/FS.
Practically all the core Linux and BSD contributors are paid full-time to hack on Free, open source code, sometimes with proprietary coding/design work also as a job duty, in many cases not.
This trend really started at the height of the Dot-bubble, and many thought it would end with it as well, but surprise, it didn't. There was a lot of job-changing to be sure, but nobody who could hack the kernel stayed without a patron for long. Reason: this stuff is strategic, and getting more so.
It's not just kernel work either. For example, being a member of the Samba team is pretty much a ticket to full employment. There are dozens of high profile projects in the same category. And finally, any FOSS credits on your resume are money in the bank.
I was thinking the same thing. The top prize is $2000. Over four months that's less than $3 an hour. And that's if you win. Screw that.
I estimate that for over half the world's population, it's over half a year's per-capita income. (Per capita GDP by country here. )
While you are unlikely to do this purely for the money you live in a rich, industrialized country, for you, the resume item is easily worth the effort. And what the heck, even if you don't win, you are going to learn enough to make it worthwhile.
Re:business vs tech presss
on
SCO Roundup
·
· Score: 1
The reason the stock has been going up is because a company known as Integral Capital Partners has apparently been purchasing a shitload of stock and filed with the SEC that they had over 5% interest in SCO as of 8/22. Interestingly enough, ICP owns the majority of shares in Drugstore.Com, and Melinda Gates (Bill's wife) sits on the board of Drugstore.com.
Sorry for replying twice, but this is important. Are you able to substantiate your comments on who's buying the SCO stock? I see that Integral Capital Partners is based in Seattle, what a coincidence.
Re:business vs tech presss
on
SCO Roundup
·
· Score: 1
The reason the stock has been going up is because a company known as Integral Capital Partners has apparently been purchasing a shitload of stock and filed with the SEC that they had over 5% interest in SCO as of 8/22. Interestingly enough, ICP owns the majority of shares in Drugstore.Com, and Melinda Gates (Bill's wife) sits on the board of Drugstore.com.
Nice research. So, assuming that Integral Capital Partners is actually a proxy for BillG himself, it's just another way of pumping the stock to finance SCO and its officers. How blatant. I'm still wondering how Darl McBride is getting his payoff, since he himself isn't dumping stock like the others.
I would say that EISA was a lot LESS common then the Vesa Local Bus slots. I have a lot of PC Magazines of the time - they all advertised VLB slots, not EISA slots.
But Eisa outlived the VESA bus. After PCI came in the VESA bus wasn't needed any more because PCI was fast enough[1]. But it didn't cost much to throw an Eisa slot or two onto a motherboard beside the PCI slots since they also handle ISA cards. Lots of manufacturers did, and I'm still running a machine today with PCI and Eisa slots in it. I never did see an Eisa card in one of my machines though.
[1] When graphics bandwith outgrew the PCI bus, AGP was devised, filling much the same role as VESA did.
I think that this is the perfect environment for an anti-worm. If the spread of such a worm was limited to the college's netblock, it could be easily controlled (luckily computer viruses don't spontaneously mutate) and it could be set to download all needed patches from a campus server, and destroy itself on command from the same server. Something like this could also be worthwhile on corporate networks. Why haven't antivirus companies caught on to this?
Once the machine is owned by a virus, patching it and pretending everything is ok is just plain stupid. You have no idea how many trojans the virus installed. Once an infected machine is found, it should be blocked from the net immediately, physically disconnected, shut down, and reinstalled from scratch, including all applications. Basically, the only safe approach is to boot from a CD and wipe the disk.
Even with all of the above, you're still not 100% safe, your BIOS may have been trojaned (i.e., reflashed). The best approach is prevention: just don't run an OS that leaves you wide open like that. The second time it happens to you, you might as well put Linux on the machine. You're obviously going to save time in the long run, not to mention keeping your valuable data safe from snooping or perhaps total loss.
Kernel 2.4.20 fails to compile on my alpha using Gcc 3.3, 2.95 works though. I'm sure it's fine for x86, but I find the ports lacking in refinement.
This would be because (some of) the 2.4 ports haven't been cleaned up, and possibly never will be. I should have mentioned, I was talking about 2.6. Though to be sure, it's important that 2.4/x86 just works, so the vast majority of users can use gcc 3.2+ without worrying. Since it's installed by default on most distributions, you have to go out of your way not to use it anyway.
To transition to a new compiler, it's essential that the new compiler support both the most current release in the previous stable series as well as all the releases in the new stable series, at least for x86 which is, like it or not, "mainstream". Where goes x86, so go all the ports, eventually, even if it may be slightly less smooth to make the transition.
And what happens when the DHS begins to use Linux/Solaris/et al and the attackers focus their attention on these products and find numerous and obvious vulnerabilities?
If they are obvious, then we already found them. Numerous... I don't think so, not in the core system. When a new Linux vulnerability comes out, it's big news and dozens of hackers descend on it immediately. Then when the fixes go out, they are *easy* to apply and highly unlikely to break anything unrelated in your system.
Any new features that go into core systems get heavily peer-reviewed for security impact. That's *proactive* security. This process has been going on for 30 years (long before Linux appeared) and you might say, it's reached a state of comparative maturity.
This is the difference between security as an afterthought and security as a process. Besides that, Linux 2.6 has a gleaming new plug-in security harness. This allows the user to tailor their own security system. For example, mandatory access controls allow the administrator to limit the actions of any process, even root. The impetus for this originally came from the NSA. You can bet that's interesting to government departments across the board.
Anyone know why they still require gcc 2.95? Or is this a minimum? Will it compile and run with gcc 3.3.x without problems? I was under the impression they tried to target the current stable version of gcc on each new major release.
There is still an architecture or two that requires gcc 2.95 to compile properly (unless you're running Sparc 32 you are probably OK) and there are some developers still fond of it because of 20% or so faster build speed. The cord will likely be cut in the next cycle.
Gcc 3.x has worked just fine for me for the past couple of years. I switched at 3.0.7 and didn't have any problems with kernel builds, though 3.2+ is recommended, because of C++ binary compatibility.
Debian Sid has gcc 3.3.2 at the moment, and Redhat switched to the 3 series a year or so ago.
Actually that doesn't make it a "flat-out lie" - it could easily be interpreted as Sontag says. Think of it as a explanatory diagram rather than an example of alledged infringing code.
It's a stretch, and even then you're not out of the woulds. How would you explain the slide being labeled "System V Code"?
This is what a judge means when he says "I see a problem with your argument".
Sloppy math but... $8,250,000 / $700 = 12,000 licenses bought?! That number is probly much much lower. The only way that they could have come up with this kind of money is from their stock, or they playing more games to get their stock back up.
Sloppy brain. Do you not remember that Microsoft bought millions of dollars worth of license (they didn't need) and that Sun did some kind of stock flip as well?
Apparently with GNU make, to get best performance, you should use -j(N+1) for a machine with N CPUs, to ensure that the CPU always has work to do. I suspect this doesn't scale well to very large numbers of CPUs, but for N
Indeed, it's more efficient and it's what I do when I'm compiling for real. I suppose it's the right thing to do when comparing CPU performance as well, since we're not very interested in how efficiently the processor idles.
Eek, I meant two. Though I usually test with -j4 on dual proc and -j2 on single proc, on the theory it focuses the work more on the CPU and less on the disk.
That should have been "make clean". Anyway, here's a more useful version that is actually a pretty good general benchmark for a dual processor system:
make mrproper && cp../config-2.4.21.config && make oldconfig && make dep && time make -j2 bzImage
The "mrproper" makes sure that a benchmark run can't take advantage of work done by the preceding one. The "j2" runs four make threads in parallel. To be fair, this should be "j1" when testing the single processor system, since there may be a little unnecessary overhead for running two threads on a single CPU.
There should be a reboot between each run, to clear out the file cache. When comparing between machines, it's essential to use the same kernel and kernel source.
it's not the price that's holding me back (I'm considering a dual xeon for some serious SGML crunching) but rather the question of "what's the added value of 64bit for me here ?"
I'm doing fine with my 3GB Ram...
On a 32 bit architecture, the kernel has to use nasty, expensive tricks to address memory over 3/4GB (for the default 1:3 userspace/kernel memory split). These tricks require lots of TLB invalidations, which are painfully slow, since the cache has to be reloaded from main memory. 64 bit architectures can just directly address that high memory, so you get a speed boost. How much, I don't know yet, since I don't have an Opteron at hand to benchmark. It will be quite measurable.
Slashdot Mirror
MEP McCarthy said in a June analysis of the proposed directive that there were links between the patentability of computer-related inventions and the growth of IT industries in the United States. Such patents aided "in particular the growth of small and medium enterprises and independent software developers," she wrote, citing a study on the issue carried out for the European Parliament by London's Intellectual Property Institute.
Roughly analogous to the tobacco industry carrying out a study on the risk of cancer from smoking.
Does anybody have a copy of this study, on which Ms. McCarthy relies?
America has a strong software business yes
but thas because software patents benefit large corporations
That has nothing to do with U.S. software companies being big and rich. For the most part, they are big and rich because of copyright. No other protection is needed for the kind of software that allows companies to grow big and rich, its complexity ensures that.
It has been suggested that the Parliament's report will for the first time allow the patentability of computer-implemented inventions. This is simply not true. The patenting of computer-implemented inventions is not a new phenomenon. Patents involving the use of software have been applied for and granted since the earliest days of the European Patent Office (EPO).
Is he intentionally twisting the truth by implying that software patents are valid in Europe, when they aren't, or that the EPO has followed the letter and spirt of the law in granting those patents, when it clearly hasn't? One way to approach this is to give him the benefit of the doubt, and explain to him why his statement is incorrect. Give him a chance to prove his honesty. If he proves otherwise, at least we know where he stands, and perhaps his constituents would like to know as well.
In any event, perhaps it would be possible to find out something about his affiliations, does he have any obvious vested interest that would prompt him to distort the facts in this particular way? Is there any such thing as a public record of campaign contributions for an MEP? Do you know anybody who knows him personally?
At a time when many of our traditional industries are migrating to Asia and when Europe needs increasingly to rely on its inventiveness to reap rewards, it is important to have the option of the revenue secured by patents and the licensing out of computer-implemented technologies.
Glibly overlooking the fact that the net cost to the public of patents, in the form of lack of availability of free software, will certainly exceeds the revenue to be earned by the owners of patents. Furthermore software development in the EU will slow down, denying European software companies an advantage they could have had. If he wants to hand the advantage to Asia, he has certainly set out on the right path to accomplish it.
Perhaps he needs some help understanding the fact that there is no shortage of evidence that copyright provides all the protection nessary for a software company to survive and prosper.
ISDN - very expensive to start with (+- R2000-00 initial startup @ R 7.50 / $1) then you still pay for the call charges. If the config goes haywire you can end up with a bill of R 4000-00/month.
ADSL is only available in certain areas - but there is a 3gig monthly cap. some guys can go through that in a day if they wanted to, and the service is being oversubscribed so quickly that the transfer rates are becoming dysmal. The only advantage is the 24x7 online connectivity (although they say that this is not guaranteed)
So most subscribers pay for 56k access (and we do pay for every local call made)
That pretty much describes the situation in Germany, up until two years ago. Barbaric. The weird thing is, the locals just never understood how badly Deutsche Telekom was abusing them until broadband came along and made it perfectly obvious how far the country was going to fall behind if things didn't change fast.
Local calls still carry toll charges, it's so stupid, as if the bandwidth needed to carry a voice call actually cost anything measurable these days. It's also still pretty much impossible to get flat rate 24/7 dialup access, as Canada has had since about 1995. So if you aren't in an urban area, you are still a digital hillbilly, you connect to the internet only on special occasions.
2. uClinux is not readily hackable, at least until you drift of it, and also know how to recover when this thing freezes. You can not just dive into it as if it was a linux PC.
You have never worked on any embedded system, for pay or otherwise.
are almost pointless, a 56k connection is bad enough without it being shared across several computers.
Oh, you won't be able to surf your pr0N, but it's good enough for email and irc, when your broadband goes down (it happens, and chances are, your POTS will still be there).
But that's not the point. The point is, you can run your own code on this thing. One application I've always wanted: a fanless unit sits on my dsl connection, always on, waits for me to connect from overseas, say, and powers up my server with wake-on-lan.
Since you have falsely accused me of plagiarism I have a hard time taking lessons from you on 'courtesy', but I will be a good sport and I will try to rephrase my future submissions to Slashdot if others agree with your position.
His position is about as nonexistent as a position gets. Please save your valuable time for something useful.
They do cheat, yes. There is no need to break backwards compatability, in fact the protocol already has what's needed, it's mostly a matter of XFree engineering and getting it effecient enough to not kill performance.
It will be pitifully slow without major changes to the rendering model. Window buffers are one approach, but there's a better way to do it: retained mode graphics where each window buffers a list of graphics primitives that are re-rendered on window expose or any visibility event, such as lying underneath a translucent window that changed. The primitives can be sorted into, e.g., a quadtree, to make partial exposes of complex windows efficient. This can be done with minimal impact on applications.
There are big advantages to this approach:
- Memory efficient
- Relatively processor-efficient, especially if accelerated with a 3D render card
- Subpixel rendering - buffered window bitmaps don't support that
- Easily extended to use standard 3D visibility and culling algorithms
- Double buffering is an easy extension
As much as I love Open-Source/Free Software, it IS very very hard to make a living making it. I'm not talking about writing code for an employer, and then getting permission to release it; I'm talking about actually making your living DIRECTLY off of making, releasing and "selling" open source/free software, a la Red Hat (who just recently turned a profit for the first time). I'd love to hear some more stories from people who've actually made money by coding OSS/FS.
Practically all the core Linux and BSD contributors are paid full-time to hack on Free, open source code, sometimes with proprietary coding/design work also as a job duty, in many cases not.
This trend really started at the height of the Dot-bubble, and many thought it would end with it as well, but surprise, it didn't. There was a lot of job-changing to be sure, but nobody who could hack the kernel stayed without a patron for long. Reason: this stuff is strategic, and getting more so.
It's not just kernel work either. For example, being a member of the Samba team is pretty much a ticket to full employment. There are dozens of high profile projects in the same category. And finally, any FOSS credits on your resume are money in the bank.
I was thinking the same thing. The top prize is $2000. Over four months that's less than $3 an hour. And that's if you win. Screw that.
I estimate that for over half the world's population, it's over half a year's per-capita income. (Per capita GDP by country here. )
While you are unlikely to do this purely for the money you live in a rich, industrialized country, for you, the resume item is easily worth the effort. And what the heck, even if you don't win, you are going to learn enough to make it worthwhile.
The reason the stock has been going up is because a company known as Integral Capital Partners has apparently been purchasing a shitload of stock and filed with the SEC that they had over 5% interest in SCO as of 8/22. Interestingly enough, ICP owns the majority of shares in Drugstore.Com, and Melinda Gates (Bill's wife) sits on the board of Drugstore.com.
Sorry for replying twice, but this is important. Are you able to substantiate your comments on who's buying the SCO stock? I see that Integral Capital Partners is based in Seattle, what a coincidence.
The reason the stock has been going up is because a company known as Integral Capital Partners has apparently been purchasing a shitload of stock and filed with the SEC that they had over 5% interest in SCO as of 8/22. Interestingly enough, ICP owns the majority of shares in Drugstore.Com, and Melinda Gates (Bill's wife) sits on the board of Drugstore.com.
Nice research. So, assuming that Integral Capital Partners is actually a proxy for BillG himself, it's just another way of pumping the stock to finance SCO and its officers. How blatant. I'm still wondering how Darl McBride is getting his payoff, since he himself isn't dumping stock like the others.
I would say that EISA was a lot LESS common then the Vesa Local Bus slots. I have a lot of PC Magazines of the time - they all advertised VLB slots, not EISA slots.
But Eisa outlived the VESA bus. After PCI came in the VESA bus wasn't needed any more because PCI was fast enough[1]. But it didn't cost much to throw an Eisa slot or two onto a motherboard beside the PCI slots since they also handle ISA cards. Lots of manufacturers did, and I'm still running a machine today with PCI and Eisa slots in it. I never did see an Eisa card in one of my machines though.
[1] When graphics bandwith outgrew the PCI bus, AGP was devised, filling much the same role as VESA did.
I think that this is the perfect environment for an anti-worm. If the spread of such a worm was limited to the college's netblock, it could be easily controlled (luckily computer viruses don't spontaneously mutate) and it could be set to download all needed patches from a campus server, and destroy itself on command from the same server. Something like this could also be worthwhile on corporate networks. Why haven't antivirus companies caught on to this?
Once the machine is owned by a virus, patching it and pretending everything is ok is just plain stupid. You have no idea how many trojans the virus installed. Once an infected machine is found, it should be blocked from the net immediately, physically disconnected, shut down, and reinstalled from scratch, including all applications. Basically, the only safe approach is to boot from a CD and wipe the disk.
Even with all of the above, you're still not 100% safe, your BIOS may have been trojaned (i.e., reflashed). The best approach is prevention: just don't run an OS that leaves you wide open like that. The second time it happens to you, you might as well put Linux on the machine. You're obviously going to save time in the long run, not to mention keeping your valuable data safe from snooping or perhaps total loss.
Kernel 2.4.20 fails to compile on my alpha using Gcc 3.3, 2.95 works though. I'm sure it's fine for x86, but I find the ports lacking in refinement.
This would be because (some of) the 2.4 ports haven't been cleaned up, and possibly never will be. I should have mentioned, I was talking about 2.6. Though to be sure, it's important that 2.4/x86 just works, so the vast majority of users can use gcc 3.2+ without worrying. Since it's installed by default on most distributions, you have to go out of your way not to use it anyway.
To transition to a new compiler, it's essential that the new compiler support both the most current release in the previous stable series as well as all the releases in the new stable series, at least for x86 which is, like it or not, "mainstream". Where goes x86, so go all the ports, eventually, even if it may be slightly less smooth to make the transition.
The fact is, you can make windows as secure as any other OS out there, as long as you know what you're doing.
What turns that glib claim into a lie is, with closed source it's impossible to know what you're doing.
Never mind that security has never been an overriding concern in Windows' basic design. The end result speaks for itself, as any 13 year old can see.
And what happens when the DHS begins to use Linux/Solaris/et al and the attackers focus their attention on these products and find numerous and obvious vulnerabilities?
If they are obvious, then we already found them. Numerous... I don't think so, not in the core system. When a new Linux vulnerability comes out, it's big news and dozens of hackers descend on it immediately. Then when the fixes go out, they are *easy* to apply and highly unlikely to break anything unrelated in your system.
Any new features that go into core systems get heavily peer-reviewed for security impact. That's *proactive* security. This process has been going on for 30 years (long before Linux appeared) and you might say, it's reached a state of comparative maturity.
This is the difference between security as an afterthought and security as a process. Besides that, Linux 2.6 has a gleaming new plug-in security harness. This allows the user to tailor their own security system. For example, mandatory access controls allow the administrator to limit the actions of any process, even root. The impetus for this originally came from the NSA. You can bet that's interesting to government departments across the board.
Anyone know why they still require gcc 2.95? Or is this a minimum? Will it compile and run with gcc 3.3.x without problems? I was under the impression they tried to target the current stable version of gcc on each new major release.
There is still an architecture or two that requires gcc 2.95 to compile properly (unless you're running Sparc 32 you are probably OK) and there are some developers still fond of it because of 20% or so faster build speed. The cord will likely be cut in the next cycle.
Gcc 3.x has worked just fine for me for the past couple of years. I switched at 3.0.7 and didn't have any problems with kernel builds, though 3.2+ is recommended, because of C++ binary compatibility.
Debian Sid has gcc 3.3.2 at the moment, and Redhat switched to the 3 series a year or so ago.
Actually that doesn't make it a "flat-out lie" - it could easily be interpreted as Sontag says. Think of it as a explanatory diagram rather than an example of alledged infringing code.
It's a stretch, and even then you're not out of the woulds. How would you explain the slide being labeled "System V Code"?
This is what a judge means when he says "I see a problem with your argument".
Sloppy math but...
$8,250,000 / $700 = 12,000 licenses bought?!
That number is probly much much lower. The only way that they could have come up with this kind of money is from their stock, or they playing more games to get their stock back up.
Sloppy brain. Do you not remember that Microsoft bought millions of dollars worth of license (they didn't need) and that Sun did some kind of stock flip as well?
Apparently with GNU make, to get best performance, you should use -j(N+1) for a machine with N CPUs, to ensure that the CPU always has work to do. I suspect this doesn't scale well to very large numbers of CPUs, but for N
Indeed, it's more efficient and it's what I do when I'm compiling for real. I suppose it's the right thing to do when comparing CPU performance as well, since we're not very interested in how efficiently the processor idles.
The "j2" runs four make threads in parallel...
Eek, I meant two. Though I usually test with -j4 on dual proc and -j2 on single proc, on the theory it focuses the work more on the CPU and less on the disk.
That should have been "make clean". Anyway, here's a more useful version that is actually a pretty good general benchmark for a dual processor system:
../config-2.4.21 .config && make oldconfig && make dep && time make -j2 bzImage
make mrproper && cp
The "mrproper" makes sure that a benchmark run can't take advantage of work done by the preceding one. The "j2" runs four make threads in parallel. To be fair, this should be "j1" when testing the single processor system, since there may be a little unnecessary overhead for running two threads on a single CPU.
There should be a reboot between each run, to clear out the file cache. When comparing between machines, it's essential to use the same kernel and kernel source.
They've asked for help getting some dual proc benchmarking software...
make mrproper && time make bzImage
it's not the price that's holding me back (I'm considering a dual xeon for some serious SGML crunching) but rather the question of "what's the added value of 64bit for me here ?"
I'm doing fine with my 3GB Ram...
On a 32 bit architecture, the kernel has to use nasty, expensive tricks to address memory over 3/4GB (for the default 1:3 userspace/kernel memory split). These tricks require lots of TLB invalidations, which are painfully slow, since the cache has to be reloaded from main memory. 64 bit architectures can just directly address that high memory, so you get a speed boost. How much, I don't know yet, since I don't have an Opteron at hand to benchmark. It will be quite measurable.