I've butted dual GXP60's (one per channel of course) on/dev/ataraid against a MegaRaid1500 with four optimally striped UltraWide 7200 rpm scsi drives after copying "/" from one set of drives to the other. Given these identical configurations I ran both bonnie++ and hdparm to find that udma100 beat scsi-raid by a factor of nearly 3 in (IIRC) sequential reads and seemed equivalent or superior in other respects.
Cheap Performance!
Nope.
Some fantastic benchmark numbers didn't add up when I timed kernel compiles between the two. The scsi setup performed consistently better. What's more is that putting my desktop to work on the ide drives (websurfing in particular) is comparatively sluggish compared to the scsi system. Scsi drives feel like a well designed stick-shift auto. Ide is like a minivan with an automatic transmission -- you hit the accellerator and pick up steam seconds later. I think it's more than the "higher quality" of scsi drives. The GXP's hail from a younger, more sophisticated generation.
it works on self-deforming flat screen technology. Touch-screens with the ability to make "buttons" could be as revolutionary to the UI as the mouse was. Just don't ask me how it would work.
His point isn't valid if berlin allows one to scale that 6 inch wide dialog box down to 3 inches. That is the intent of resolution independence. You've missed j7953's point just as Salamander missed berlin's point. j7953 didn't miss anything.
IIRC, a gcc developer has pointed out (on this forum) that gpl code optimization is legally hindered by patents in a way analogous to mp3 encoding. They might want use (even after simply discovering...) a certain algorithmn in gcc but can't because the intellectual property is privately owned.
GNU's site has no mention of this and claims the current version is 2.2.3. Freshmeat had an announcement for glibc-2.2.4-pre4 on Wednesday. Who exactly is releasing this?
Slashdot Mirror
My PIII cooled with a slow fan is running at 31.9 / 89.4 -- at boot it's 23.5 / 74.3.
Didn't the article claim rpm affects latency?
Cheap Performance!
Nope.
Some fantastic benchmark numbers didn't add up when I timed kernel compiles between the two. The scsi setup performed consistently better. What's more is that putting my desktop to work on the ide drives (websurfing in particular) is comparatively sluggish compared to the scsi system. Scsi drives feel like a well designed stick-shift auto. Ide is like a minivan with an automatic transmission -- you hit the accellerator and pick up steam seconds later. I think it's more than the "higher quality" of scsi drives. The GXP's hail from a younger, more sophisticated generation.
It's the interface.
it works on self-deforming flat screen technology. Touch-screens with the ability to make "buttons" could be as revolutionary to the UI as the mouse was. Just don't ask me how it would work.
Fix that flicker with xvidtune.
His point isn't valid if berlin allows one to scale that 6 inch wide dialog box down to 3 inches. That is the intent of resolution independence. You've missed j7953's point just as Salamander missed berlin's point. j7953 didn't miss anything.
IIRC, a gcc developer has pointed out (on this forum) that gpl code optimization is legally hindered by patents in a way analogous to mp3 encoding. They might want use (even after simply discovering...) a certain algorithmn in gcc but can't because the intellectual property is privately owned.
GNU's site has no mention of this and claims the current version is 2.2.3. Freshmeat had an announcement for glibc-2.2.4-pre4 on Wednesday. Who exactly is releasing this?