Bigger blocks means you can smear in a bit more redundancy so that an error in a single byte can be recovered instead of being fatal. CDs for example are highly resistant to bit errors and their block size for data is 2048 bytes.
I can't believe they let the world waste all those CPU cycles between mid August and the end of September. I figure they've poured $50,000 of user's electricity down the drain - and that's being conservative, if you'll excuse the pun.
If the user never agreed to the GPL, then she is using unlicensed software. Could you not then countersue both for the infringement of your copyright, and for any damages incurred through the misuse of your software - including the original law suit?
256 colors? I was very happy playing it on monochrome (yes, simple black or white, no grey in between) monitors of Sun ELCs and IBM RS/6000s. Many a mis-spent hour of youth...
It's great how they increased the monthly rate from E110 to E132 (more like US120) before they even released the service. If only they'd actually release the damn thing. I could take the E228 (US$200) per month they're suggesting for uncapped 1Mbit down, 256Kbit up. Right now the alternatives are dialup and/or several times more expensive.
Then there is the issue that N64 and PS2 both use Rambus; why, oh why
It's simple. With RAMBUS they get high bandwidth from very few chips and a very low pin count. Playstation 2 has only two RDRAM ICs. Have a look at a 32MB GeForce DDR and tell me how many RAM chips there are on it... In a guaranteed high volume product with a relatively long life expectancy it makes sense to go for fewer components and fewer tracks for a simpler PCB and simpler manufacturing.
There are a lot of really good questions here. It's going to be difficult to choose the 10 best. Any chance you could short-list say 20 of them and post it as a slashdot poll?
Your argument is not a convincing one. I see no impending break point in the history of Earth. In fact the most powerful common source of radio waves here is radar. A technology that was invented by British military science.
No country in it's right mind will freeze technological advancement - the risk of another country doing so, and then beating you to a bloody pulp, is too great.
Actually, the 75GB drive was out in SCSI first (it might have been 72GB, but the same tech otherwise)
That's an illusion. Those SCSI drives are older technology. It's a larger package with more platters - TWELVE (!) versus five - more noise, and more heat than the 75GXP from IBM.
For the price of one of those 72GB SCSI disks, excluding the U2W controller you'd need, you could have two 75GB IDE disks and a shiny new ATA/100 controller offering a dedicated channel for each. I know which would sooner earn a home in my PC.
SCSI has its place. That place is not outside the super-high-end.
3. Did I mention that a UW SCSI bus can have 15 targets
The IDE approach has always been to have multiple cheap channels rather than an expensive single do-everything channel. A cheapo i815 chipset comes with two 100Mbyte/sec channels - that's already more aggregate bandwidth than your modern 160MByte/sec SCSI channel which likely cost more than that whole i815 motherboard. Look for >2 channels to become common when SerialATA hits the settles.
4. Disconnect.
AFAIK IDE has supported SCSI style disconnection since ATA/66. Operating system support is the problem.
plus, the better drives are always out for SCSI first
/me glances at www.storage.ibm.com. Not from where I'm standing lad. OK. You got me on the 10,000rpm drives, but I'll take 75GB 7200rpm any day - the data transfer rates seem comparable.
Read the story and the background. RAMBUS are claiming pretty damned wide patents on anything vaguely approaching SDRAM technology. They were part of the JEDEC standards organization when SDRAM was being standardized. According to the rules of that organization, companies are required to disclose any patent interests they have in a technology that's before the organization. They failed to do so. They allowed the standard to progress and become all-pervasive. Then they popped up exclaiming "Oh, look what I've just found in my back pocket!"
The other way to look at it is RAMBUS are effectively claiming a monopoly on the worldwide RAM market. They get to set the price of a given technology. They get to say that their madcap RAMBUS technology will be licensed for a slightly less extortionate rate than SDRAM. If that's not in breach of the Sherman Act I don't know what is.
The DRAM chip is still being put into a power saving state - just a different one. There's no significant difference in the performance of the two from what I can gather. It's nothing like the performance hit of an FDIV hack.
They're fixing it in software because it can be fixed in software. FDIV was a hardware problem that was practically impossible to work around in software without a huge performance hit.
This problem is entirely different. The DRAM chip is going into a power saving mode that doesn't quite work and loses information. All the software patch does is prevent the device from using that particular mode.
What did the world do before Microsoft invented the 3D acceleration API? Umm. We got along fairly well with OpenGL thank you. OpenGL would undoubtedly be in a better position today had Microsoft chosen to cooperate with all the other OS vendors to make OpenGL the single viable API.
What did the world do before Microsoft invented the TCP/IP stack? Dunno about the rest of you, but I was pretty happy with BSD sockets. The real question is why it took so damned long for Microsoft to put that functionality into their OS. Oh wait. I know why - they were busy maintaining their monopoly through slash-and-burn acquisitions, exclusive OEM licensing agreements, glossy marketing, and back-stabbing short-lived 'alliances' with other companies.
Slashdot Mirror
We just have to invent some kind of "computational" device to automate the process...
Spatial localization of bit errors.
Bigger blocks means you can smear in a bit more redundancy so that an error in a single byte can be recovered instead of being fatal. CDs for example are highly resistant to bit errors and their block size for data is 2048 bytes.
/dev/random you say? Those random bits aren't free you know...
What you need is a random accelerator card.
In the same way gunshot victims who don't wear body armour are at fault.
If the animal gets loose, how much good will it really do to be able to track down the owner based the tag embedded in him?
How long before a rejected student takes legal action against a school, ApplyYourself or both?
Get back to me when it uses electricity generated by the bike to power the PC.
I can't believe they let the world waste all those CPU cycles between mid August and the end of September. I figure they've poured $50,000 of user's electricity down the drain - and that's being conservative, if you'll excuse the pun.
If the user never agreed to the GPL, then she is using unlicensed software. Could you not then countersue both for the infringement of your copyright, and for any damages incurred through the misuse of your software - including the original law suit?
256 colors? I was very happy playing it on monochrome (yes, simple black or white, no grey in between) monitors of Sun ELCs and IBM RS/6000s. Many a mis-spent hour of youth...
It's great how they increased the monthly rate from E110 to E132 (more like US120) before they even released the service. If only they'd actually release the damn thing. I could take the E228 (US$200) per month they're suggesting for uncapped 1Mbit down, 256Kbit up. Right now the alternatives are dialup and/or several times more expensive.
Who's up for a bit of ROT13?
Paul.
Then there is the issue that N64 and PS2 both use Rambus; why, oh why
It's simple. With RAMBUS they get high bandwidth from very few chips and a very low pin count. Playstation 2 has only two RDRAM ICs. Have a look at a 32MB GeForce DDR and tell me how many RAM chips there are on it... In a guaranteed high volume product with a relatively long life expectancy it makes sense to go for fewer components and fewer tracks for a simpler PCB and simpler manufacturing.
Paul.
There are a lot of really good questions here. It's going to be difficult to choose the 10 best. Any chance you could short-list say 20 of them and post it as a slashdot poll?
Paul.
Your argument is not a convincing one. I see no impending break point in the history of Earth. In fact the most powerful common source of radio waves here is radar. A technology that was invented by British military science.
No country in it's right mind will freeze technological advancement - the risk of another country doing so, and then beating you to a bloody pulp, is too great.
Paul.
Actually, the 75GB drive was out in SCSI first (it might have been 72GB, but the same tech otherwise)
That's an illusion. Those SCSI drives are older technology. It's a larger package with more platters - TWELVE (!) versus five - more noise, and more heat than the 75GXP from IBM.
For the price of one of those 72GB SCSI disks, excluding the U2W controller you'd need, you could have two 75GB IDE disks and a shiny new ATA/100 controller offering a dedicated channel for each. I know which would sooner earn a home in my PC.
SCSI has its place. That place is not outside the super-high-end.
Paul.
3. Did I mention that a UW SCSI bus can have 15 targets
/me glances at www.storage.ibm.com. Not from where I'm standing lad. OK. You got me on the 10,000rpm drives, but I'll take 75GB 7200rpm any day - the data transfer rates seem comparable.
The IDE approach has always been to have multiple cheap channels rather than an expensive single do-everything channel. A cheapo i815 chipset comes with two 100Mbyte/sec channels - that's already more aggregate bandwidth than your modern 160MByte/sec SCSI channel which likely cost more than that whole i815 motherboard. Look for >2 channels to become common when SerialATA hits the settles.
4. Disconnect.
AFAIK IDE has supported SCSI style disconnection since ATA/66. Operating system support is the problem.
plus, the better drives are always out for SCSI first
Paul.
ls -l --color | less is ugly on the terminal and various xterms
Try ls -l --color | less -r instead. Maybe even alias it.
Paul.
Moderation has worked well for slashdot. I see a bright future for it in the Napster network.
Paul.
Read the story and the background. RAMBUS are claiming pretty damned wide patents on anything vaguely approaching SDRAM technology. They were part of the JEDEC standards organization when SDRAM was being standardized. According to the rules of that organization, companies are required to disclose any patent interests they have in a technology that's before the organization. They failed to do so. They allowed the standard to progress and become all-pervasive. Then they popped up exclaiming "Oh, look what I've just found in my back pocket!"
The other way to look at it is RAMBUS are effectively claiming a monopoly on the worldwide RAM market. They get to set the price of a given technology. They get to say that their madcap RAMBUS technology will be licensed for a slightly less extortionate rate than SDRAM. If that's not in breach of the Sherman Act I don't know what is.
Paul.
The DRAM chip is still being put into a power saving state - just a different one. There's no significant difference in the performance of the two from what I can gather. It's nothing like the performance hit of an FDIV hack.
Paul.
They're fixing it in software because it can be fixed in software. FDIV was a hardware problem that was practically impossible to work around in software without a huge performance hit.
This problem is entirely different. The DRAM chip is going into a power saving mode that doesn't quite work and loses information. All the software patch does is prevent the device from using that particular mode.
Paul.
"We intend to block French access to said auctions by posting them in the English language only."
It's about time their isolationist language policy did some *good*.
Paul.
I find Tim's comments deeply ironic.
What did the world do before Microsoft invented the 3D acceleration API? Umm. We got along fairly well with OpenGL thank you. OpenGL would undoubtedly be in a better position today had Microsoft chosen to cooperate with all the other OS vendors to make OpenGL the single viable API.
What did the world do before Microsoft invented the TCP/IP stack? Dunno about the rest of you, but I was pretty happy with BSD sockets. The real question is why it took so damned long for Microsoft to put that functionality into their OS. Oh wait. I know why - they were busy maintaining their monopoly through slash-and-burn acquisitions, exclusive OEM licensing agreements, glossy marketing, and back-stabbing short-lived 'alliances' with other companies.
Paul.