Slashdot Mirror
Slashback: Hardware, Lexis, Free
More RAM than I can afford. RyanT5000 writes "The article referenced in "Getting Rid of the Disks" incorrectly states that the maximum RAM for a 32-bit Intel compatible system is 4 GB. This was true up to and including the original Pentium processor, and it makes a lot of sense (after all, 2^32 = 4 294 967 296). However, with the Pentium Pro, Intel added 4 pins to the address bus, expanding the maximum physical address space to 64 GB (using paging, since it still uses 32 bit addressing). I would assume AMD has a similar feature. If you're on Windows, you'll need a server version to get above 4 GB, but most major Linux/BSD/etc. OSes support it. This would probably be cheaper (and definitely faster) than SCSI SSDs. If you need more than 64 GB of solid state storage, you probably shouldn't be running on an Intel."
"Free" always makes people suspicious. imevil writes "A while ago slashdotters asked some questions to the GNUWin II team. Well, here are the answers. In the meantime, more people joined the team, and more languages were added (this one looks pretty cool)." There's also a short article about GNUWin running at NewsForge.
On a related note, cos(0) writes "According to this(1), this(2), and other stories, many people are interested in running open source, high-quality software on MS Windows. The author of this site provides an up-to-date CD image of the latest versions of numerous high-quality OSS applications (complete list on the site), updated monthly, downloadable via BitTorrent. (The same site also offers a web-based Code Beautifier.)"
Cool distribution method! (And if you're on dialup, $5 seems like a great bargain -- Are you listening, Cheapbytes?)
Toys are so tempting to the wallet ... OrenWolf writes "Ars Technica Has a review up of BroadQ's QCast Tuner software. Unlike the earlier /. review, this article goes into great detail about the technical capabilities of the software. A must read for PS2 owners looking for a PC-PVR-esque solution."
Yeah, but does this review include any original software? ;) And david_adams writes "Slashdot linked to an article I wrote last month about my experience with a CDMA2000 1x wireless network from SprintPCS. It sparked quite a bit of controversy, but not for the reasons I expected. Because I called Sprint's service 3G in the title, but admitted it was 2.5G in the first paragraph, I heard from people on both sides, chiding me on the one hand for calling it 3G, and on the other for calling it 2.5G. I decided to research and write a new article to get to the bottom of it. What is truly 3G? Where is the line between 2.5G and 3G?"
The time to wait is now! ThunderDawg writes "Intel resumed Canterwood Pentium 4 3 GHz 800FSB shipments yesterday. TAFKAEFKAF (The Anomaly Formerly Known as Errata Formerly Known as Flaw) was corrected with a software patch.
Intel is again shipping its new 3GHz Pentium 4 processor, a week after it halted shipments due to the discovery of an "anomaly," an Intel spokesman said Monday. PC makers that use the chip in their systems have been supplied with a software update to fix the issue, George Alfs, an Intel spokesman said. Vendors including Hewlett-Packard, Dell Computer, and Gateway introduced desktop systems based on the chip when it was released on Monday last week. The issue with the 3GHz Pentium 4 with support for an 800MHz system bus occurs only in rare circumstances and users are unlikely to be affected, according to Alfs."
I'd take google and a strong AI any day. hondo77 writes "A bit of a followup to this article from back in February, LexisNexis has been named the publisher of official reports by the California Supreme Court, according to this press release. "The public will have free access to the official text of the opinions at a Web site hosted by LexisNexis linked to the court's Web site." IANAL but it doesn't sound ominous to me."
While theoretically you can put up to 64 gigs into a recent IA32 machine, my understanding is that in Linux at least the practical limit is 16 gigabytes. After that the page tables won't fit in kernel space.
I'd expect that most other OSes have limits like that due to architectures designed when nobody was close to using a full 4 gigs.
I was suspicious about their P1 floating point bug, but what really got me to avoid them forever was their PIII.
And now P4 is buggy, just like P1....
Does that mean a horrendous sucking privacy violation is to come on the P6?
..."The issue with the 3GHz Pentium 4 with support for an 800MHz system bus occurs only in rare circumstances and users are unlikely to be affected, according to Alfs."
:)
Wasn't that what they said about the floating point bug, too? Well, both of them?
Sounds like they are fixing it with a microcode patch, much as they fixed the PII FIST/FISTP bug, but the article was rather short on details.
"'Tis great confidence in a friend to tell him your faults, greater to tell him his." --Poor Richard's Almanac
Hmm, yes, but you're prolly looking at motherboards aimed at the individual computer users for personal uses. 4+gB amounts of memory are mianly needed in the server and clustering bit of computing, not for home/office use. (Well, with Win2k3 coming upon us, we might never know.) Although the video/photoediting branche of computing might have a use for that much memory, I doubt they are in favour of the x32 architecture anyways, considering Apple's strength and foothoold in the video/photo editing branche...
And like it's been stated before, if you really do need that much memory on a single machine, you might be better of with a different architecture anyways, possibly a whole different solution to the problem.
Hate me!
...or an apple XServe (unless they go to, say, a PowerPC 970 in future models).
...at any given time. An application could use a memory-mapping API such as mmap() to map pieces of a >4GB object into the address space as needed. Yes, that's rather ugly, but people have done it before, e.g. on PDP-11s back in the old days.
Another thing you can use lots of memory for is a disk cache; I know of one series of x86 machines that supports more than 4GB of memory, most of which is used for disk caching (given that all those machines do is file service).
The network infrastructure is very cool, in some ways. In other ways, it is really painful to do *anything* as this system has been an online and running 24x7 computer system for something like two *decades* now. Because of that, there are a lot of legacy links and layers on layers of stuff that may have made sense once, but now works *just* well enough that it is not yet cost effective to scrap it and rewrite it.
:)
They also have a very sophisticated OO environment, that permeates the system from top to bottom. API's layered on API's layered on API's... all OO (mostly C++, moving to Java).
They add tremendous value and do really complicated things, and do them fast and reliably. The people doing it are REALLY smart.... for the most part this is a real advantage.... but sometimes gets annoying (picture the two smartest people from every graduating class, all thrown together, trying to impress each other... they produce some pretty amazing stuff, but sometimes it would be nice to just have something simple and straightforward, even if it is only 90% as efficient).
Their network infrastructure is to die for. Backups on their redundancies, duplicates of their extras, connections upon connections. Complete with a REALLY cool "war room" where everything gets carefully watched. Not to mention that they have their own major air force base about 20 miles away (Wright Patt) with HEAVILY armed and highly motivated state of the art fighter aircraft... talk about your defense in depth
Anybody complaining about Lexis-Nexis and what they pay for information (wah! information wants to be free) is totally clueless. Information *does* want to be free, but it also wants to be useless. You pay people to add order and utility to information, and the bang for the buck you get from both Lexis and Nexus can't be beat.
That is why components should protect themselves. Overclocking is not generally a huge problem as long as you have sufficient cooling. But then the same problem occurs whn you place a CPU running at stock speed in a hot environment. That is why major components should take steps to protect themselves. This is the only reason why I currently prefer the P4 CPU over AMD CPUs (Even though for a "platform" it is the opposite).
Essentially parts should not be damaged by overheating, they should shutdown or work slower, but not be damaged. Current CPUs (Intel and AMD, though I'm not sure if either could be fooled into not working), and many GPU (eg: Nvidia) include this protection.
I personally would be more worried about a nasty person maxing out the vCore voltage in modern softBIOS motherboards. I personally would prefer the option to fall back to jumpers (via a jumper) if I wish. Some settings really shouldn't be adjustable mid-flight so to speak, and I question the wisdom of making such settings easy to change. (eg: vCore, multiplier, etc)
I would prefer an "adjust" jumper on the motherboard, which you can set to allow the settings to be changed, and then when everything is set up, set the jumper to "lock" and leave it.
LaTeX, derived from TeX, is irreplaceable when writing scientific papers. One benefit of TeX, or specifically LaTeX, is that is makes it easy to write nice-looking letters, conference papers, articles, and so on by taking care of most of the formatting for you. Its not big on flexibility, but it is big on looks.