Over on Newegg some of the feedback on the WinXP Pro X64 states that it's root is Server 2003 and not NT, and that's why the XP X64 works so well (stable, fast, etc.). Can anyone on here confirm or deny any of this?
ah ha! "...[edit] Windows XP Professional x64 Edition This edition supports the x86-64 extension of the Intel IA-32 architecture. x86-64 is implemented by AMD as "AMD64", found in AMD's Opteron and Athlon 64 chips, and implemented by Intel as "Intel 64" (formerly known as IA-32e and EM64T), found in some of Intel's Pentium 4 and all of Intel's later chips. Windows XP Professional x64 Edition uses version 5.2.3790.1830 of core operating system binaries, [21] the same version used by Windows Server 2003 SP1 (and Windows XP 64-bit Edition version 2003) as they were the latest versions during the operating system's development. Even service packs and updates for Windows XP x64 and Windows Server 2003 x64 are distributed in unified packages, [22] much in the manner as Windows 2000 Professional and Server editions for x86. During the initial development phases (2003â2004), Windows XP Professional x64 Edition was named Windows XP 64-bit Edition for x86 Extended systems and later, Windows XP 64-bit Edition for Extended systems, as opposed to 64-bit Edition for Itanium systems. ..." from http://en.wikipedia.org/wiki/Windows_XP_Pro_x64#Windows_XP_Professional_x64_Edition So are we not getting Server 2003 by purchasing XP Pro X64?
As I understand it, a program has to be specifically written to take advantage of memory using physical address system and only some database programs are, ordinary programs such as Tradestation cannot benefit, right?
In general that is not correct. Java software doesn't care if it's 32 bit or 64 bit - just run the correct JVM. C/C++ software in general SHOULD be portable between 32 bit and 64 bit. But it must be compiled for the appropriate target architectue. This is just a matter of giving a target directive to the compiler. Physical addresses are not used (and should not be used) - the operating system takes care of that. Vast amounts of free software is available for linux distros such as Ubuntu in both 32 and 64 bit flavours and in most cases the source code is the same and it is just compiled with appropriate compiler C/C++ compiler command lime switch. If a program wants lots of memory, then it just keeps asking the OS for more. Eventually the OS will refuse but the limit will be higher on a 64 bit box. The application programmer does not need to make any special changes to their code for each of the environments.
I believe a single process can only address 2GB of memory with or without PAE. If it's loaded with the /3GB switch then it can address 3GB. PAE is only advantageous when you run multiple processes.
Do you want to know what the difference between a 64bit processor and 32bit processor is? 32 bit processors can process 2^32 bits per second. 64 bit processors can process 2^64 bits per second. To use a 64 bit processor for a system that will be purely used for trading is like killing a bug with a nuclear bomb. So unless you are performing complex mathematical calculations that require hours or even days to process, Donât get drawn into the hype of a 64 bit processor and pay extra for something that you will never use to even its 20% potential.
Is the dude looking at saving a few bob on 32 bit OS even done any seriuos price checks? I find it idiotic that this question of cost should be an issue.
I'm assume you're joking with respect to processing speeds of 2^32 and 2^64 above. Processing speed and throughput is a function of clock speed, memory speed, cache and memory configuration, etc, and not the size of the addressable memory space of the CPU/MMU. The biggest benefit of 64 bits is the larger address space assuming you're on a 64 bit OS (Linux or W2003 for instance). 64 bit also helps the virtual memory subsystem significantly.