There was a discussion recently on comp.arch about the "Memory Wall", which is a similar phenomenon to the hard disk latency problem. Processors are getting faster faster than other parts of the system.
One of the more insigtful comments is that we need to go back to Knuth's Fundamental Algorithms and see how he handles the memory hierarchy. In this case, tape drives. Tape drives in those days had reasonable bandwidth (for the time) and miserable latency. Sounds like hard drives now. So we need to reread the old algorithms and do a little translation:
Tape drives -> disk drives
Disk drives -> DRAM main memory
Core (RAM) memory -> 2nd level cache
Cache -> 1st level cache
This isn't simple, of course, because things have changed, amongst other things there is no detailled control over what is in which cache level, but the gist of it is that we need to drop the assumption that accessing random addresses in RAM is just as fast as accessing RAM in such a way that cacheing and readahead are possible. This is still a universal assumption in algorithms textbooks.
Similarly, we need to drop the idea that we can get to any part of the disk within a reasonable time. If we treat modern disks like a tape drive with really good mechanics for fast-forward and rewind, then we are closer to the reality, where you can transfer over a megabyte in the time it takes you to read a random bit from the other end of the disk, (about 10ms).
This implies our 'hard disks' should be RAM disks, and the current hard disks should be relegated to some sort of backup role.
The mainframe I used at Cambridge (Phoenix, a heavily hacked IBM 370 - this is only 10 years ago!) would move your files out to tape if you didn't access them for a few weeks. Next time you accessed them it took a few minutes for the tape robot to find your tape and put it back on the disks.
RDS works fine in most of Europe. You can keep the same radio station as you drive across the country, and it can do a few other useful things like change channels or pause the casette player when traffic news comes on.
Of course, this only works for stations that are broadcast all over the country you are in, but in my experience these are the good ones anyway. After all, if a radio station doesn't broadcast over a wide area, how are they going to be able to spend any money on production. And if all they do is inane chatter and mainstream rock music, then who cares whether you are listening to one radio station or another.
Colour screen means your batteries don't last long. The Psion is supposed to last 30 days on 2 AA batteries.
Win CE machines need colour because the user interface sucks without it.
Does this thing just hiss or what?
on
Digital VCRs
·
· Score: 1
It's got a big disk.
It's in your living room, connected to your sound system.
To me this says "MP3!"
To me this says noise. I never heard a hard disk I would want to have in my living room. I have enough trouble with my monitor, but that's because unfortunately it's an old Eizo with a fan.
What's wrong with it is that he has one. Seems very childish. None of the really big free software developers have one. All of the script kiddies have one.
Rubbish! The IJG library which is part of SPECint95 is integer-only and that is the software that everyone uses, since it has a very liberal license.
Jpeg compression is part of SPEC
on
K7 Benchmarking
·
· Score: 2
Actually, JPEG compression is already part of the SPEC suite so when the real SPEC results are released you just have to look at the breakdown. Another of the SPEC marks is a run of gcc (v 1.38 I think).
Unfortunately the SPEC marks are never compiled with gcc because it isn't as fast as Intel's compilers, which I presume AMD will use:-). I hope this list will be updated when the K7 is out, since it is probably a good indicator of Linux integer performance.
For me, VIM is the killer app for a palmtop. That's why I want an 80 character display, a keyboard and a long battery life (ie monochrome).
For others it could be uEmacs. Combined with mutt, lynx and a miniature SMTP server that can talk over infrared or mobile phone it would be an ideal mobile communication centre for a geek.
Some crazy guys are porting vi to EPOC actually, so perhaps Linux isn't necessary.
Digital looked at MMX and decided they didn't really need that stuff since they were already fast enough at that sort of thing. So for the 21264 they did the MVI Motion Video Instructions instead, which are supposed to speed up MPEG compression. Don't know of any benchmarks though,
The point is that other companies aren't doing what intel is telling them to. And intel really doesn't like that.
Indeed. Add to this that Intel has a large stake in Rambus and that every single Rambus module sold will result in a royalty payment to Rambus and you can see why they don't want PC133 and PC266 to succeed.
The K7 may have a 200 mhz bus between the chipset and the processor
According to this great article on K7 rumours Slot A will be able to run up to about 250MHz, but Slot B will go up to 400MHz. Yum yum!
VIA made K7 chipsets [will support PC133]
Here in time for Christmas by the looks of things. This sabre rattling by Intel might even make takeup faster. If people are worried about whether VIA has the rights to the GTL+ bus VIA might advise them to use the EV6 stuff for the 21264 and K7 instead. If only AMD would second-source the K7 so people could really believe that supplies will be reliable. You don't piss off Intel unless you are very sure you won't have to come crawling back. Actually I did see some rumours of a second source for the K7. IBM and Samsung would be the obvious candidates.
While future K7 chipsets will support RAMBUS
It would be ironic if high end K7 chipsets were delayed because they decided to invest a lot of effort getting Rambus to work, and then the RAM modules don't turn up. I think for the high end, with huge 2nd level caches and enormous bandwidth requirements Rambus may have the edge if the caches take the top off the latency problems, and AMD may have thought the same way. And who would have guessed that an Intel-sponsored technology could fail in the PC space?
If Intel is refusing to lisence GTL+, then VIA can't produce a chipset which works with Slot 1. I'm not a lawyer, but I think that would be considered using monopoly power to gain a monopoly in the chipset market
This is more a patent issue than anything else. If there were no patents involved, why would VIA neeed a GTL+ license at all? I don't know the precise rules here, is there some sort of obligation on the part of a patent holder to license the patent to other manufacturers at reasonable conditions? I feel there should be, but I don't know. A clause that more or less insists that VIA can't use the license to make a chipset that is better than Intel's doesn't sound reasonable.
Whatever your view on patents (and hardware patents obviously have a lot less damaging effects than software patents) they are a sort of government-granted monopoly, so you cannot be surprised when they are used to maintain monopolies.
You won't be able to buy one until after you got the full specs. Intel aren't stupid, they are funding gcc and by extension Linux development on IA64, and they will release all the docs before the hardware is on sale.
The idea is, you have two or three characters that look very similar. One is Chinese, one is Japanese, one may be Korean. They look similar, but usage differs (ie, they have different meanings). Because of the ridiculous Unicode proposal, they are all unified into the same code point
According to the Unicode web page and everything I have ever read on Unicode the unification only takes place if the characters have the same meaning. Can you name an exception where that rule hasn't been followed?
depending on the Unicode font used, you might get the Chinese character, you might get the Japanese character, or maybe the Korean character
That's just font management and/or language management. Every decent DTP system needs font management anyway, and if it is going to get hyphenation right it needs language tagging, even if you are only using Latin-1.
The whole point of Unicode was that code points were supposed to be kept separate between languages
What on earth makes you think that was the whole point of Unicode?
One of the big points in Unicode is that it should be possible to convert from any character set encoding to Unicode and then back again. That has caused some compromises for example the fact that the Greek capital letter Alpha has a different encoding to the Latin capital letter A although you could argue that they are the same character.
But you can't make that 2-way conversion guarantee for encoding systems that let you switch from character set to character set with escape codes. Amongst other problems that would make that impossible is the fact that you can use the same escape codes to switch into Unicode, so you would get an infinte recursion.
If you want to convert to and from escape-code switching encoding systems you will have to extract the implicit language and font information and make it explicit in the Unicode version of your data. That is probably a good idea anyway, and is possible in HTML and any other serious text format.
If it's a 'plain text file' then you can't embed the font or the language information, but that's why plain text sucks, and the same problem appears in Latin-1 plain text files.
Presumably in order not to lose any market share to 3com they have now made the most important parts of the SDK fre. See Epoc World.
There's also a Linux port underway. With a real keyboard and an 80 character screen this is a real hackers palmtop. Forget cheezy colour that will eat your batteries or wimpish pen based entry systems for people who can't type. You can get a flash disk up to about 100MB for your ext2fs and 8 MB RAM is available now, 16 and 24 soon. (Btw. I don't actually own one yet!)
If you want the mobile stuff wait for the next mobile phones with basically a built in Psion from Ericsson.
Just curious, what are the chances that MS would get a hold of a majority of Redhat's shares?
There's a limit (20%?) to how much they would be allowed to buy up anonymously. After that they would have to make a bid. I think the price would be very high if people knew they were selling to Microsoft. Also the anti-trust (monopoly) authorities would be likely to take an interest, even ban it.
You want to underwrite it yourself? I don't think your pockets are quite that deep, Mr. AC. Or are you a rich Republican?
If the IPO is oversubscribed, which doesn't seem unlikely, the usual procedure, at least in the UK, is that small stockholders are allocated the full amount, and people who asked for more are only given a certain percentage. Anyone know what the plans are, or whether they are even fixed in advance?
This is probably some deadlock keeping a vital datastructure locked (eg memory allocation). That can quickly cause a total lockup and it's the sort of bug that's difficult to pin down the responsibility for - it results from an unclear deadlock avoidance policy, which I could imagine you easily get when you put a BSD kernel on a Mach kernel.
It's not a big deal in the overall scheme of things, though obviously it's a PITA if you were trying to use MacOS X as a web server (brave thing to do on such a new OS!).
Slashdot Mirror
One of the more insigtful comments is that we need to go back to Knuth's Fundamental Algorithms and see how he handles the memory hierarchy. In this case, tape drives. Tape drives in those days had reasonable bandwidth (for the time) and miserable latency. Sounds like hard drives now. So we need to reread the old algorithms and do a little translation:
Tape drives -> disk drives
Disk drives -> DRAM main memory
Core (RAM) memory -> 2nd level cache
Cache -> 1st level cache
This isn't simple, of course, because things have changed, amongst other things there is no detailled control over what is in which cache level, but the gist of it is that we need to drop the assumption that accessing random addresses in RAM is just as fast as accessing RAM in such a way that cacheing and readahead are possible. This is still a universal assumption in algorithms textbooks.
Similarly, we need to drop the idea that we can get to any part of the disk within a reasonable time. If we treat modern disks like a tape drive with really good mechanics for fast-forward and rewind, then we are closer to the reality, where you can transfer over a megabyte in the time it takes you to read a random bit from the other end of the disk, (about 10ms).
This implies our 'hard disks' should be RAM disks, and the current hard disks should be relegated to some sort of backup role.
The mainframe I used at Cambridge (Phoenix, a heavily hacked IBM 370 - this is only 10 years ago!) would move your files out to tape if you didn't access them for a few weeks. Next time you accessed them it took a few minutes for the tape robot to find your tape and put it back on the disks.
Am I reading too much into it when I note that the website we all love to hate is down. Will it be back up on Monday with some content?
Unless they are contracting for the NSA they will have to tell the public what they are up to at some point, otherwise how can anyone buy the product.
At the moment the best bet seems to be that it is a project rather like the Elbrus E2K, but who knows?
When it comes online they are going to need the 4-way SMP and Apache tweaks to stop being slashdotted...
Of course, this only works for stations that are broadcast all over the country you are in, but in my experience these are the good ones anyway. After all, if a radio station doesn't broadcast over a wide area, how are they going to be able to spend any money on production. And if all they do is inane chatter and mainstream rock music, then who cares whether you are listening to one radio station or another.
Win CE machines need colour because the user interface sucks without it.
It's got a big disk.
It's in your living room, connected to your sound system.
To me this says "MP3!"
To me this says noise. I never heard a hard disk I would want to have in my living room. I have enough trouble with my monitor, but that's because unfortunately it's an old Eizo with a fan.
What's wrong with it is that he has one. Seems very childish. None of the really big free software developers have one. All of the script kiddies have one.
Rubbish! The IJG library which is part of SPECint95 is integer-only and that is the software that everyone uses, since it has a very liberal license.
Unfortunately the SPEC marks are never compiled with gcc because it isn't as fast as Intel's compilers, which I presume AMD will use :-). I hope this list will be updated when the K7 is out, since it is probably a good indicator of Linux integer performance.
For others it could be uEmacs. Combined with mutt, lynx and a miniature SMTP server that can talk over infrared or mobile phone it would be an ideal mobile communication centre for a geek.
Some crazy guys are porting vi to EPOC actually, so perhaps Linux isn't necessary.
Digital looked at MMX and decided they didn't really need that stuff since they were already fast enough at that sort of thing. So for the 21264 they did the MVI Motion Video Instructions instead, which are supposed to speed up MPEG compression. Don't know of any benchmarks though,
The point is that other companies aren't doing what intel is telling them to. And intel really doesn't like that.
Indeed. Add to this that Intel has a large stake in Rambus and that every single Rambus module sold will result in a royalty payment to Rambus and you can see why they don't want PC133 and PC266 to succeed.
The K7 may have a 200 mhz bus between the chipset and the processor
According to this great article on K7 rumours Slot A will be able to run up to about 250MHz, but Slot B will go up to 400MHz. Yum yum!
VIA made K7 chipsets [will support PC133]
Here in time for Christmas by the looks of things. This sabre rattling by Intel might even make takeup faster. If people are worried about whether VIA has the rights to the GTL+ bus VIA might advise them to use the EV6 stuff for the 21264 and K7 instead. If only AMD would second-source the K7 so people could really believe that supplies will be reliable. You don't piss off Intel unless you are very sure you won't have to come crawling back. Actually I did see some rumours of a second source for the K7. IBM and Samsung would be the obvious candidates.
While future K7 chipsets will support RAMBUS
It would be ironic if high end K7 chipsets were delayed because they decided to invest a lot of effort getting Rambus to work, and then the RAM modules don't turn up. I think for the high end, with huge 2nd level caches and enormous bandwidth requirements Rambus may have the edge if the caches take the top off the latency problems, and AMD may have thought the same way. And who would have guessed that an Intel-sponsored technology could fail in the PC space?
If taking memos is too slow, perhaps you need to overclock your hand, or perhaps get something with a real keyboard.
Looks slashdotted already. If you like me can't get through, try JC's page for more K7 info and rumours.
This is more a patent issue than anything else. If there were no patents involved, why would VIA neeed a GTL+ license at all? I don't know the precise rules here, is there some sort of obligation on the part of a patent holder to license the patent to other manufacturers at reasonable conditions? I feel there should be, but I don't know. A clause that more or less insists that VIA can't use the license to make a chipset that is better than Intel's doesn't sound reasonable.
Whatever your view on patents (and hardware patents obviously have a lot less damaging effects than software patents) they are a sort of government-granted monopoly, so you cannot be surprised when they are used to maintain monopolies.
You won't be able to buy one until after you got the full specs. Intel aren't stupid, they are funding gcc and by extension Linux development on IA64, and they will release all the docs before the hardware is on sale.
According to the Unicode web page and everything I have ever read on Unicode the unification only takes place if the characters have the same meaning. Can you name an exception where that rule hasn't been followed?
depending on the Unicode font used, you might get the Chinese character, you might get the Japanese character, or maybe the Korean character
That's just font management and/or language management. Every decent DTP system needs font management anyway, and if it is going to get hyphenation right it needs language tagging, even if you are only using Latin-1.
The whole point of Unicode was that code points were supposed to be kept separate between languages
What on earth makes you think that was the whole point of Unicode?
One of the big points in Unicode is that it should be possible to convert from any character set encoding to Unicode and then back again. That has caused some compromises for example the fact that the Greek capital letter Alpha has a different encoding to the Latin capital letter A although you could argue that they are the same character.
But you can't make that 2-way conversion guarantee for encoding systems that let you switch from character set to character set with escape codes. Amongst other problems that would make that impossible is the fact that you can use the same escape codes to switch into Unicode, so you would get an infinte recursion.
If you want to convert to and from escape-code switching encoding systems you will have to extract the implicit language and font information and make it explicit in the Unicode version of your data. That is probably a good idea anyway, and is possible in HTML and any other serious text format.
If it's a 'plain text file' then you can't embed the font or the language information, but that's why plain text sucks, and the same problem appears in Latin-1 plain text files.
There's also a Linux port underway. With a real keyboard and an 80 character screen this is a real hackers palmtop. Forget cheezy colour that will eat your batteries or wimpish pen based entry systems for people who can't type. You can get a flash disk up to about 100MB for your ext2fs and 8 MB RAM is available now, 16 and 24 soon. (Btw. I don't actually own one yet!)
If you want the mobile stuff wait for the next mobile phones with basically a built in Psion from Ericsson.
Yes, but the stock that was bought back can't be used afterwards to vote with or to prevent a hostile bid. It just disappears.
OK, that link isn't working for some reason. Netscape or something is inserting spaces in it. Try just going here and searching for Red Hat.
IANAL, but I don't think that's allowed in the US.
and sell a chunk of it to their own staff.
Now that they are certain to do.
I guess we will soon know how much of Red Hat is owned by Intel and Netscape/AOL.
There's a limit (20%?) to how much they would be allowed to buy up anonymously. After that they would have to make a bid. I think the price would be very high if people knew they were selling to Microsoft. Also the anti-trust (monopoly) authorities would be likely to take an interest, even ban it.
He left the company, but are you sure he sold the stock (options)?
You want to underwrite it yourself? I don't think your pockets are quite that deep, Mr. AC. Or are you a rich Republican?
If the IPO is oversubscribed, which doesn't seem unlikely, the usual procedure, at least in the UK, is that small stockholders are allocated the full amount, and people who asked for more are only given a certain percentage. Anyone know what the plans are, or whether they are even fixed in advance?
It's not a big deal in the overall scheme of things, though obviously it's a PITA if you were trying to use MacOS X as a web server (brave thing to do on such a new OS!).