Install Linux on it and use it to (insert anything here)
ah... a physical pr0n server.... well whatever floats your boat...... make sure you have full medical insurance......
I'm sure it does, but I doubt it fetches partial cache lines so once it starts a fetch the bus and (more importantly, because before it was slower) the memory subsystem is busy untill a full cache line has been fetched.
Also if the bus doesn't have some form of flow control then the memory controller will have to wait until it has a full line almost ready to go before it can start sending data - this will add a lot to latency - but now with a bus/memory combination that run at the same rate these sorts of latencies will also go away.
A prediction - as 266MHz (2x133) DDRAMS start to appear the k7 bus will be pumped up to 133MHz (266 on both edges) and k7 speed grades will start to appear in 66MHz increments rather than the current 50MHz ones.
Athlon has a 200MHz system bus (really 100 but they move data on both edges of the clock). On the other hand the PC100 memory systems only make ata at 1/2 this rate - this means that great bus is starved for performance and stuff must wait - even with PC133 it has to wait for the memory - to make matters worse K7 has cache lines that are twice as long - it has to wait twice as long as an Intel part - now with a 2x memory system and bus the latency for k7 is the same as that for a Pentium (latency is king for reads) and twice as much data is transfered to boot
In other words K7's been crippled by its memory subsystem from day 1 -it's only now you will start to see it running at full speed....
If you possibly can you should provide the company's lawyer with all the prior art you can.... after all you are going to sign this patent application under penalty of perjury saying that it's a new unique invention. You are not only legally but morally obligated to tell the truth:-)
The lawyer on the other hand will be trying to push the envelope as much as possible (and pull the wool over the patent office's eyes by citing as little as possible).
Make sure that as many existing patents as possible and previously published articles are cited in the application - well chosen these
I bet that you are far more capable at researching your patent than the patent attorney is - if you are especially carefull you can find existing patents that have expired as prior art:-)
KDE/QT will never be able to get away from this minefield, every time they make a gesture in the GPL direction RMS and his cabal throw up another roadblock... what will be next? I think they're doing the right thing and saying they wont play this silly game any more.
Besides how RMS can possibly think that someone who put a GPL header in a source file AND included the QT include files in their KDE program didn't intend their program to be used with QT is beyond me
If you have some razor/razorblades backend business model then you had beter make sure you have your Linux software ready the same day you ship.... or pissed potential customers will write their own software and in effect hijack your platform....
Logging - this sort of process is often called a 'clean room' or 'chinese wall' technique (I have no idea why).
What's very important is that you log every transmission between the 2 groups so that you can prove in court that no illicit information passes between the groups.
"Your Honor - I kept telling Mr Jobs - the roof - it leaks over my cube... right onto the computer - but he didn't do anything other than screaming at me to stop complaining and work harder..... then one day there was this blinding flash...."
Actually A/UX was never Mach based - it came from another whole family tree that's missing from the chart - 'UniPlus' which was an early System 3/5/5r2/5r3 variant with BSD networking and utilities on top - A/UX came from the system 5r2 branch.
UniSoft did over 100 ports of UniPlus to mostly 68k based platforms in the early to mid 80s.
Come on - $1500/programmer (no royalties or runtime licenses when you sell your product) - or $2400 for dual platform. If you're doing a real commercial product this is chickenfeed - maybe 2% of what you're paying the person who's using it.
I'm a chip-designer - the cost/seat for tools in our biz can be close to $100k for each engineer (for a simulation and synthesis license - more if you include a router) - so high we have to share in all but the largest companies - and they give QT away to people who aren't planning on making money with it - what more can you ask?
the KDE 2.0 panel is a hell of a lot like the Gnome 1.0 panel.
Well to my untrained eye the Gnome panel looks a hell of a lot like the KDE 1.0 panel....
Which is a GOOD thing - it shows that people from both groups are stealing good ideas from each other - we're all better off from that sort of cross-fertilization - it's why having 2 GUI platforms is a great thing in an open source world
However as a (smalltime) KDE developer myself I do feel Kurt's annoyance - we're all focused on writing cool code - it's what's we work really hard at - and it's the results of this that we want to be judged by.
The whole licensing thing is a non-issue to most of us that's been beaten to death over and over - if you want to do a line-by-line review of something - look at my CODE, make it better - then I'll respect you.
Remember: "those who code do, the rest become lawyers"
VPNs are typically encrypted.... if a large proportion of the packets going to/from the 'net go thru @home maybe it's in the Fed's interest to make sure they can be sniffed by the Carnivore box back at the ISP.....
Either that or they want to be able to charge 'business rates' just like the phone company does....
Personally I think it should be none of their business what's in your packets - after all it's YOU who are paying THEM to move the packets for YOU...
Lots of stuff has to come together for a new process to be viable:
the FAB equipment is wont be available in commercial quantities to do this new process for a long time
the FABs have to decide that they need to include that equipment in their capital expenditure (a chicken and egg problem - no one wants it so we wont spend the money - but we don't offer it yet so no one bets their next chip on the process)
it will probably take them years to make a process that will yield in volume
the CAD tools aren't here yet (today's are limping just coming up speed with 3d extraction - what happens at 0.01u? I don't know - do they have to include quantum effects? the hall effect? something else that's in the noise today the way RC effects were at 1u?
and maybe there aren't enough applications today to make it viable people are used to targetting their next design to next year's process and may not think that far in the future (though somehow I suspect DRAM would be the first obvious application here - at.2u today we have 100Mb DIMMs - at.01u we'l have 4Gb DIMMs.... oops time to move to the 64-bit CPU real soon now - actually I really like the performance jump we'll get by moving the DRAM on with the CPU....)
nor attempt in any other manner to obtain the source code.
So adovocating for open source (as a general pricipal or for this driver in particular) would put you in violation of the agreement... and nvdia's lawyers will come and take your 1st born child (of graphics card whichever is more precious to you:-)
OK, you obviously know nothing about OS design but thanks for showing it up front.
Here's a "good reason". To load a DOS session, you need to load DOS drivers. DOS drivers are non-reentrant real-mode 16-bit. Supporting them destabilizes a multi-tasking protected-mode 32-bit OS by requiring all kinds of thunking from the kernel.
You're being a bit hard and way too nasty - no one in their right mind who knew anything about OS design would imagine that anyone would get themselves into a state where they have an OS that needs to 'thunk into real mode' - M$ should have bit the bullet and got rid of this crap years ago. BTW you really think that ME doesn't still have this crap underlying it? of course it does.
And while I'm here I have to point out that anyone who has a reasonably broad background in computers just has to grimace whenever they hear the M$ use of the word 'thunk' it's been so perverted from it's original meaning (check out references to Algol 60 and call-by-name and the Burroughs B6700 [now the Unisys A-series] from the 60s) for it's true meaning
IBM has a long-time patent exchange contract with Intel from back in the days when they used to build x86s.
Intel has a large number of patents on the x86 architecture - some of them pretty bogus with prior art up the wazoo..... but they also have a lot of lawyers.... probably more that a small company like TM has employees of any kind....
What they are saying is that TM is making test tapeouts to a second foundry - one that has a lower cost per die than IBM. TSMC is know for it's very competitive pricing. As far as I know IBM is still making them.
This has a couple of advantages to them - the obvious one, the the chips get cheaper - it also provides competition to encourage IBM to get their yields up and their cost down too - and it provides a 2nd source (customers like that - if IBM's FAB burns down someone else will still be making them) and it does provide a minor exchange rate hedge (since a lot of their customers will be in Asia).
As the article mentions the downside is that it leaves them outside the legal umbrella of IBMs patent exchange with Intel....
In the real world chips don't suddenly get made and sell - after the release of something new there's a long slow adoption period - and chip production lags untill you have orders - it can take a foundry up to 3 months to respond to a sudden rise in orders (unless you've reserved space in the fab queue, pay extra, or just made chips on spec) - this can leave you in a chicken and egg situation - we can't use your chip because we can't get a lot of them next month - we can't ramp production until someone orders them.....
Not according to the Columbians I met - they claimed to be "Americans" - that's how their culture sees themselves - but then I don't see many people running around claiming to be "North Americans" either.....
Slashdot Mirror
until after it's a standard (and it's too late to change) before we break it ...... (evil grin :-)
Install Linux on it and use it to (insert anything here) ah ... a physical pr0n server .... well whatever floats your boat ...... make sure you have full medical insurance ......
Also if the bus doesn't have some form of flow control then the memory controller will have to wait until it has a full line almost ready to go before it can start sending data - this will add a lot to latency - but now with a bus/memory combination that run at the same rate these sorts of latencies will also go away.
A prediction - as 266MHz (2x133) DDRAMS start to appear the k7 bus will be pumped up to 133MHz (266 on both edges) and k7 speed grades will start to appear in 66MHz increments rather than the current 50MHz ones.
In other words K7's been crippled by its memory subsystem from day 1 -it's only now you will start to see it running at full speed ....
Dave's also taken on Scientology and Amway
The lawyer on the other hand will be trying to push the envelope as much as possible (and pull the wool over the patent office's eyes by citing as little as possible).
Make sure that as many existing patents as possible and previously published articles are cited in the application - well chosen these I bet that you are far more capable at researching your patent than the patent attorney is - if you are especially carefull you can find existing patents that have expired as prior art :-)
Besides how RMS can possibly think that someone who put a GPL header in a source file AND included the QT include files in their KDE program didn't intend their program to be used with QT is beyond me
In short fuck with their business model ....
If you have some razor/razorblades backend business model then you had beter make sure you have your Linux software ready the same day you ship .... or pissed potential customers will write their own software and in effect hijack your platform ....
What's very important is that you log every transmission between the 2 groups so that you can prove in court that no illicit information passes between the groups.
"Your Honor - I kept telling Mr Jobs - the roof - it leaks over my cube ... right onto the computer - but he didn't do anything other than screaming at me to stop complaining and work harder ..... then one day there was this blinding flash ...."
Kind of like before the invention of the radio .....
Actually A/UX was never Mach based - it came from another whole family tree that's missing from the chart - 'UniPlus' which was an early System 3/5/5r2/5r3 variant with BSD networking and utilities on top - A/UX came from the system 5r2 branch.
UniSoft did over 100 ports of UniPlus to mostly 68k based platforms in the early to mid 80s.
All this in the same deposition where they tried to find out who was running the FTP site at 127.0.0.1 that had all their files on it (YHBT-HAND).
For more antics of those wacky lawyers check out www.xenu.net
I'm a chip-designer - the cost/seat for tools in our biz can be close to $100k for each engineer (for a simulation and synthesis license - more if you include a router) - so high we have to share in all but the largest companies - and they give QT away to people who aren't planning on making money with it - what more can you ask?
Well to my untrained eye the Gnome panel looks a hell of a lot like the KDE 1.0 panel ....
Which is a GOOD thing - it shows that people from both groups are stealing good ideas from each other - we're all better off from that sort of cross-fertilization - it's why having 2 GUI platforms is a great thing in an open source world
However as a (smalltime) KDE developer myself I do feel Kurt's annoyance - we're all focused on writing cool code - it's what's we work really hard at - and it's the results of this that we want to be judged by.
The whole licensing thing is a non-issue to most of us that's been beaten to death over and over - if you want to do a line-by-line review of something - look at my CODE, make it better - then I'll respect you.
Remember: "those who code do, the rest become lawyers"
Either that or they want to be able to charge 'business rates' just like the phone company does ....
Personally I think it should be none of their business what's in your packets - after all it's YOU who are paying THEM to move the packets for YOU...
So adovocating for open source (as a general pricipal or for this driver in particular) would put you in violation of the agreement ... and nvdia's lawyers will come and take your 1st born child (of graphics card whichever is more precious to you :-)
Here's a "good reason". To load a DOS session, you need to load DOS drivers. DOS drivers are non-reentrant real-mode 16-bit. Supporting them destabilizes a multi-tasking protected-mode 32-bit OS by requiring all kinds of thunking from the kernel.
You're being a bit hard and way too nasty - no one in their right mind who knew anything about OS design would imagine that anyone would get themselves into a state where they have an OS that needs to 'thunk into real mode' - M$ should have bit the bullet and got rid of this crap years ago. BTW you really think that ME doesn't still have this crap underlying it? of course it does.
And while I'm here I have to point out that anyone who has a reasonably broad background in computers just has to grimace whenever they hear the M$ use of the word 'thunk' it's been so perverted from it's original meaning (check out references to Algol 60 and call-by-name and the Burroughs B6700 [now the Unisys A-series] from the 60s) for it's true meaning
Intel has a large number of patents on the x86 architecture - some of them pretty bogus with prior art up the wazoo ..... but they also have a lot of lawyers .... probably more that a small company like TM has employees of any kind ....
This has a couple of advantages to them - the obvious one, the the chips get cheaper - it also provides competition to encourage IBM to get their yields up and their cost down too - and it provides a 2nd source (customers like that - if IBM's FAB burns down someone else will still be making them) and it does provide a minor exchange rate hedge (since a lot of their customers will be in Asia).
As the article mentions the downside is that it leaves them outside the legal umbrella of IBMs patent exchange with Intel ....
In the real world chips don't suddenly get made and sell - after the release of something new there's a long slow adoption period - and chip production lags untill you have orders - it can take a foundry up to 3 months to respond to a sudden rise in orders (unless you've reserved space in the fab queue, pay extra, or just made chips on spec) - this can leave you in a chicken and egg situation - we can't use your chip because we can't get a lot of them next month - we can't ramp production until someone orders them .....
So is IBM - stuff it manufactures is covered by it's patent exchange with Intel
Who aren't covered are small upstart companies that haven't had the past clout to get these sort of contracts in place ....
yeah well - the rest of the world calls us 'yanks' but that doesn't sit so well with the southern redneck contingent
Not according to the Columbians I met - they claimed to be "Americans" - that's how their culture sees themselves - but then I don't see many people running around claiming to be "North Americans" either .....