| Thursday, September 5, 2002 |
12:26 - Silicon Implants
http://www.powerlogix.com/press/releases/2002/020904.html
|
(top)  |
PowerLogix, a leading maker of CPU upgrade cards for Macs, just announced a new set of dual-CPU upgrade units in speeds up to 1GHz (with higher speeds coming soon), in both 100MHz and 133MHz bus speeds, for just about all Power Mac G4 systems. Considering that DDR RAM doesn't seem to do these machines all that much good (as we've seen, crestfallen, recently), this kind of upgrade would boost the performance of a nearly three-year-old machine like my 450MHz G4 to nearly that of the newest models. That plus a new video card would make it, in Marcus' words, quite l33t.
These upgrade cards look like a nice package. They slot in to the existing CPU daughterboard slot, and they come with a much nicer three-phase power supply than the one the machine already has. Besides, it would be a dualie, with all that implies. Not a bad deal for about a thousand bucks.
Well, there is this little tiny weensy problem. Apparently, a certain early revision of the logic board on the six-month-long-lived "AGP Graphics" generation of G4 tower machines has a bridge chip that's incompatible with dual processors. This logic board was succeeded in quick order by one that was dual-compatible, but PowerLogix provides a link with details on how to find out if your machine supports dual CPUs or not. You have to boot into OS9, run the System Profiler, generate a certain kind of report, and read off an obscure number that's supposed to read "uni-n: 7" or higher. Lower numbers won't work with the dual cards.
Guess what mine says? uni-n: 3. Aaauugh!
Ah well. I can hold out for a bit longer. (And technically I could spring for a single-CPU upgrade card to 1GHz or so, but that's just not as alluring.) And I'd have to get a new GeForce4, and a' that, and a' that.
I'm planning to wait for the Next Big Thing before I upgrade my main home machine. IBM is planning to give its dissertation on the new PPC64 in October, and sources say that they've been working on it in secret for a good long time now-- by the time they give the speech, the chip will be in live testing, and slated for release in consumer machines made by companies whose names begin with A and rhyme with "Snapple" in about July of next year-- possibly earlier, if everybody claps their hands really reaaallly loudly, and belieeeeves...
I can hold out that long. I want to spend about $3000 on a new machine, which means getting the top-end of what's available at any given time. (I plan to get more mileage out of my Cinema Display than out of my computer.) And if I'm going to be spending that within a year, I don't want to put down more than $1000 on stopgap upgrades. I can wait.
Meanwhile, 10.2.1 is rapidly on its way, according to Think Secret. This release is set to come so soon after 10.2 because from what I hear, Jaguar was rushed out the door sooner than it really deserved to be-- they cut a few spit-polish features like minimize-in-place and Rendezvous playlist-sharing/streaming in iTunes and complete Quartz Extreme optimization, so they could get it out the door and into people's hands before they were accused of promising "Summer" and then releasing on September 20, as has happened before. (And pushing it out ahead of the 9/11 season could only have been good for sales.) In software, the only way to keep from submerging under endless validation and integration testing is to cut features with brutal prejudice-- early on in the process if possible, but later and right up to release day if necessary. It's a shame to cut things you've promised, even things that are really really cool. But software is unique in its ability to surprise you with its susceptibility to the butterfly effect; wiggle this bit up here in TCP/IP, and up pops a bug down there in filesystem browsing. Den Beste has a recent piece which describes this in action better than I can.
But the upshot is that we'll be getting those features now in a point release, probably within the next week or two. This will address stability as well-- I've had very good luck so far with Jaguar, and most of the stories I've been hearing from people are extremely positive, but there are the occasional horror stories. (This one draws the gleeful finger of Glenn Reynolds-- I can certainly understand the temptation to respond to numerous people's insufferable recommendations of Macs to seek out reasons why such recommendations are horse manure; but honestly, an isolated and temporary unstable state of the OS is hardly a deal-killing indictment of a platform. Or it shouldn't be, anyway.)
That said, Think Secret notes the following:
Informants now say that one of the areas Apple is addressing in 10.2.1 is the SCSI I/O kit, which stands to provide benefits primarily to developers. "There will be major changes to it," one source said. "These changes are very welcome ... SCSI support in OS X to date has been abominable."
Good... I'm glad I'm not the only one who noticed. Maybe now we'll be able to boot our machines without having to have our connected SCSI devices powered-on for it to load the proper stubs.
My SCSI scanner died again when I installed Jaguar. (Search above for "scanner" to find the tales of banshee-wailing woe surrounding my scanner over the past year.) Turns out that this wasn't because the OS became incompatible with its driver for my SCSI card or anything, because technically I can still scan using <shudder> VueScan... but now that I've been spoiled by the lusciousness that is SilverFast, I refuse to subject myself to that torment. And SilverFast's software is what's become incompatible with 10.2, through what mechanism I do not know.
Silverfast is reported to have responded to customers' complaints already on this issue, and they say they will have a fix "in a few weeks".
That's perfectly fine. I can wait.
|
|
