64-bit question

[akadmon]

Just to share, 1P 3.0 is sooo nice. I haven't had an issue with logins or anything. I'm not the strongest user of it but it hasn't given me an issue so far.

I'm glad you like it. What I don't like is that Agile is forcing long time users of 1P to buy 1P 3 in order to take advantage of the faster, more stable 64-bit Safari . They should fix 1P 2 for those of us who do not need all the new features that 1P 3 supposedly has.

There is a rant on this subject on the 1P support forum:


http://support.agilewebsolutions.com/showthread.php?t=19189

 

[Amdahl]

No, a 32-bit kernel, PAE or not, cannot address more than 4GB of memory. No single process can address more than 32-bits of address space. The kernel is a single process. The OS, as a whole, can utilize more than 4GB because of PAE.

S=

There is no such thing as 'an OS, as a whole.' The kernel can do it, or nobody can.

The processor can address it, the kernel runs the processor, the kernel can address it. Q.E.D.

Your argument, which is attempting to split a hair that can't be split, is wrong. Take, for instance, real mode x86. A 16-bit pointer may only be able to address 64k, but when paired with a 16-bit segment/selector, 1MB was possible on 8086. PAE basically lets the page table act as a segment/selector for 32-bit pointers. The only difference is that only the KERNEL can manipulate the page table. So while an app still can't address more than 4GB at any particular instant (sans a call to the KERNEL), the kernel can by simply flipping bits in the page table, the equivalent of a segment/selector.