[alain_30_alain]

Hi, I just bought a MacBook Pro (Version 2,2, not with the Nvidia video card, no 2,2 or 2,4 Ghz processor)..

Here are my specs:
2.16 Ghz Intel Core 2 Duo
120 Gb Hard Drive
ATI Radeon X1600 with 128 Mb GDDR3
2Gb of DDR2 (2x1Gb)

...and I was wondering if I could run Mac OS X Leopard 10.5 in 64-bit mode?... and what is the difference between the 32-bit and 64-biit mode?

Thanks for helping me!!

[chabig]

Congratulations! You have a 64 bit processor. As far as you're concerned, there are no 32 or 64 bit "modes." Apps just run.

Whether you get 32 or 64 bit goodness depends upon how the developer wrote and compiled the application.

[slpdLoad]

Yes, you can. You don't really have to run it in a certain "mode", either. The OS is 32 and 64 bit compatible, so if the application is 64 bit, the OS will take advantage of that. Just install Leopard and you're all set.

edit: what he said ^

[TETENAL]

The main difference of 64 bit is that applications can adress more than 4 GB of RAM. Since your machine has only 2 anyway, it doesn't really matter much.

[ibookuser2]

They can also do basic arithmetic operations on larger numbers. It depends on the application how much this affects performance. On one extreme, if you're doing scientific computing, this makes a big difference. At the other end of the spectrum, browsing around the Finder or the web isn't going to be affected much by this. Computationally-intensive applications like games and video encoding fall somewhere in the middle; not huge gains but not insignificant either.

You also don't have to have more than 4 GB of RAM to make use of a 64-bit address space. For example, say you're randomly accessing a 4 GB file. On a 32-bit machine, that's too big to fit into a user program's address space, so you can't just mmap it into memory; you'd have to manually map it a piece at a time, or use sequential access methods and seek to the parts of the file you wanted to read. On a 64-bit machine, you can just mmap the entire file into memory in one contiguous block and let the virtual memory system handle reading the relevant parts of the file into physical memory and cache as you access different parts just as you'd access memory.

[Tomchu]

Registers, people. When will some of you quit bringing up the "If you don't have >4 GB of RAM, 64-bit is irrelevant to you" argument?

It's all about having more GP registers. THAT is why many 64-bit applications enjoy a performance boost.

[Chuckit]

Originally Posted by Tomchu 

Registers, people. When will some of you quit bringing up the "If you don't have >4 GB of RAM, 64-bit is irrelevant to you" argument?

It's all about having more GP registers. THAT is why many 64-bit applications enjoy a performance boost.

And having to push around more data slows them down. Overall, there's not much benefit just to being 64-bit.

[ibookuser2]

Originally Posted by Chuckit 

And having to push around more data slows them down. Overall, there's not much benefit just to being 64-bit.

It's not the fact that the registers are wider so much as the fact that there are more of them. The 32-bit x86 architecture is on the low side in terms of number of GPRs. Adding just a few more (x86-64 doubles the number) registers can cut down on a lot of situations where register thrashing is occurring and you're continuously having to access cache to execute a critical section of code.

And yes, it's true that 64-bit registers obviously require more hardware and can be slightly slower than 32-bit wide registers, but as long as the chip clock speed is the same and the pipeline isn't extended significantly, that's something that's of concern to chip architects trying to meet a particular clock speed, not to the end user.

Any time you can keep important values in the processor instead of having to go out to the L1 cache and wait for data to come back, you're going to see performance increase (given, of course, that the applications have been compiled to take advantage of the extra registers provided by the architecture.)

[TETENAL]

Originally Posted by ibookuser2 

Any time you can keep important values in the processor instead of having to go out to the L1 cache and wait for data to come back, you're going to see performance increase.

It's just that the L1 cache is "half" as large since integers are twice as wide under 64 bit. So while applications that need to handle a lot of memory may see a large performance increase, many applications that don't have such large memory requirements don't or might even be slower.

[OreoCookie]

64 bit apps can be slower than their 32 bit counterparts, especially on x64. One of the reasons is that binaries are larger. The increase in speed between x86 and x64 is mainly due to the additional registers that are accessible in 64 bit mode.

Average users shouldn't really care, unless they need more than 4 GB of RAM.