save2600 wrote:
Thanks da9000, lots of good info there.
You're welcome! Glad I could provide it.
save2600 wrote:
About the ram thing though and my observations. I wouldn't dream of complaining going from one computer to the other with different OS's. My observations are based on the same configuration. Just going from 10.4 to 10.5. My 2.1ghz iMac has 2.5gb of ram, so ram should not be an issue. That's plenty for what I use my machine for. And I'm not "spotlighting" while working either. lol Tinkertool and I are old friends to be sure. What a great program. I'll have to look at those other recommendations of yours soon though.
Right. I meant in general. More specifically I meant people who complain when going from a 10.3.9 based PowerMac G4 @ 933Mhz with 512MB to 10.4 or 10.5. All I can think of is: "well, duh". 10.5 was never intended for those machines. In fact Apple forces you to upgrade by checking the minimum speed of the machine. If it's under 887Mhz (I think), it won't install. (of course the installer is simply using JavaScript, which means you can copy the installation DVD into a disk image that's writable, modify the JS by removing the check, and burn a new disk or blast it onto a hard drive partition and install from there - hint, hint)
Anyhow, that's what I meant. However, as I pointed out before, I do agree with your general observation: 10.5 seems slower on older (and usually non-Intel) hardware.
save2600 wrote:
Oh, one last point. I make the comment that PeeCee HD's fail more often than one used in a Mac because that has been my experience over the course of 25+ years. Yes, they may share the same technology now, but when you consider how the Mac's file system is so much different than the typical PeeCee's file system, I feel that comment IS more than substantiated. One only has to look at ScanDisk & defragging to understand Winblows is throwing data illogically all over the platter. That's a lot more wear and tear going on the entire mech of the HD compared to how the Mac reads/writes. Less wear and tear = reliability and fewer HD related problems. Again, in my experience.
I see what you mean. I think it's a valid point and very plausible even though I also don't have data. The reason I think it's plausible is from my experience too, but I would not necessarily blame just the filesystem. I mean FAT is crap, for sure, but the entire OS is a culprit. Very simple explanation: if an OS doesn't have a caching mechanism or simply bad caching, it will cause a lot more disk reading (and of course also lower performance). This will make the drive do more work, as you say. That will, theoretically (i don't have evidence) increase wear and tear. From my experience Windows has been typically really bad at disk IO. One of the worse parts of XP was how swap-happy it is. Even though you might have oodles of RAM, it will prefer to page out data/code before getting close to the RAM high watermark. It's always "on disk" as we say in operating system talk.
To take it one more step further, even bad architectural implementations at the core OS level will cause bad disk access patterns. For example, many modern OSs like Linux and Mac OS X feature a unified buffer cache (
http://utcc.utoronto.ca/~cks/space/blog/unix/UnifiedBufferCache). (Not sure if Windows does have one today and if so which was the first that had it. My *guess* is that Vista might, Windows 7 must, and XP probably didn't) Meaning that all IO, including disk, but also VM and mmapped accesses go through a single cache. This utilizes memory much much better (since there are no duplicate pages all over RAM) and therefore better caching, less swapping!
So in general, now that I understood what you meant, I agree based on my experience, but without having empiric proof. I would love to have empiric proof though, so if anyone knows of any research in the field...
Here are a couple of studies that are related and very interesting, albeit very long and involved:
labs.google.com/papers/disk_failures.pdf
http://www.techpowerup.com/?25708http://hardware.slashdot.org/hardware/07/02/21/004233.shtmlMore here stuff:
http://en.wikipedia.org/wiki/Hard_driveOK, so since we're in the deep side of things by now, I suggest reading the following. There are some very good explanations as to OS X's performance. As always, it comes down to choices. What's more important? Powerful programming interfaces and interoperability or stupidly fast performance, yet braindead APIs and hard to program interfaces?
Search for "Some parts of OS X are much slower than others." in the following, and read the first couple of replies to that. But be forewarned, it's quite geeky :-)
http://hardware.slashdot.org/article.pl?sid=06/03/23/1717259Cheers
