Future Frostbite 2 Games Will Require 64-Bit OS

[Shinanigans]

My sole point was that Microsoft needs to cater for everyone and they would lose sales by only offering 64-bit versions of their OS.

That's the problem, no one wants to be the guinea pig who drives tech and alienates markets/money.

Good on em for taking the plunge. One of the first big titles to drop XP, now going to be one of the first to drop 32-bit. Huzzah!

[vcatkiller]

Guess no one else remembers the good old 16bit days before WinXP, which was 32bit only

Uh, Windows 95 was 32-bit (as were Windows 98 and Me).

Sure, it was backwardly compatible with 16-bit apps (to ease the migration from Windows 3.1), but still... hardly the "good old 16-bit days".

Actually think you'll find Windows 95/98/ME was a 16 bit OS with a 32 bit shell stuck over the top. Hence why it was still compatible with 16-bit apps. And why you had lovely oddities like how Windows 98 would **** itself if you ran it on a system with more than 512MB RAM.

[realOne]

Well Microsoft way back in 2010 did have plans to make Win8 OS 64-bit only but for the obvious reasons they added 32-bit for the mass market.
If we ain't ready in 2015 for Win9 64-bit only OS then we are in trouble seriously suck in the past with old tech, Both AMD and Intel are on the brink of 128-bit processors in the years to come.

Also Win8 is rumored to be the last 32-bit OS and making Win9 the forefront in advancement in 64-bit only OS.

[Mekon]

we are on the brink of 128-bit processors in the years to come.

I'm failing to see what advantage 128-bit processing will bring to the table, TBH.

SIMD instruction sets already allow access to pseudo-128-bit registers. Beyond that, what's the point? We don't have memory addressing issues under 64-bit, so short of a massive increase in RAM... I see little benefit beyond cryptography.

[cyclobs]

we are on the brink of 128-bit processors in the years to come.

I'm failing to see what advantage 128-bit processing will bring to the table, TBH.

SIMD instruction sets already allow access to pseudo-128-bit registers. Beyond that, what's the point? We don't have memory addressing issues under 64-bit, so short of a massive increase in RAM... I see little benefit beyond cryptography.

probably the same thing people said about 64bit, yet here we are.

why are people so against advancing technology?

[Mekon]

probably the same thing people said about 64bit, yet here we are.

Except it's not. There is a current clear benefit from 64-bit registers - the amount of RAM commonly used in desktop computers has exceeded the number of bits addressable with 32-bit registers. It might be a while before we need to go to 128-bit memory addresses, given that (according to wikipedia), the estimated total amount of digital data stored worldwide in 2010 was 2^70 (~1.2 zettabytes).

Do you have an actual reason for needing 128-bit arithmetic on your CPU? Or do you just think larger numbers must be superior, because, well, biggerer is betterer, duh!?

edit: 128-bit technology already exists, of course. GPUs commonly use 128 or 256 buses and have done for years. The Cell processor also has a number of 128-bit registers, for that matter.

why are people so against advancing technology?

You need a good business case for architectural changes - "just because" doesn't tend to cut it.

[Sathias]

If Moore's law holds for memory capacity (which is debatable), we won't need 128-bit CPUs until sometime around 2025.

http://www.zdnet.com/blog/storage/the-1 ... c-when/333

[Mekon]

If Moore's law holds for memory capacity (which is debatable), we won't need 128-bit CPUs until sometime around 2025.

http://www.zdnet.com/blog/storage/the-1 ... c-when/333

Not quite. The article is referring to the upper memory limit set by the new OSX.

16TB is easily handled with 64-bit registers. The upper limit of addressable memory in a 64-bit number is 18 exabytes (or 16 exbibytes)... assuming the whole register is used for adressing.

That's a looong way off.

[DarthLondar]

Guess no one else remembers the good old 16bit days before WinXP, which was 32bit only

Uh, Windows 95 was 32-bit (as were Windows 98 and Me).

Sure, it was backwardly compatible with 16-bit apps (to ease the migration from Windows 3.1), but still... hardly the "good old 16-bit days".

Actually think you'll find Windows 95/98/ME was a 16 bit OS with a 32 bit shell stuck over the top. Hence why it was still compatible with 16-bit apps. And why you had lovely oddities like how Windows 98 would **** itself if you ran it on a system with more than 512MB RAM.

Yeah, WinXP was the 1st true 32bit OS, 95/98/ME used a hybrid 16/32bit kernel.

[PalZer0]

Wasn't Windows 2000 a 32-bit only OS as well?

[Otto-matic]

Main reason for the 32-bit compatibility is business apps. Plenty of 32-bit business apps (and 16-bit with 32-bit wrapper) don't work in 64-bit Windows even though they potentially should through emulation.

[realOne]

Wasn't Windows 2000 a 32-bit only OS as well?

There was a Windows 2000 64-bit but it was very buggy at the time then first service pack came out and fixed a lot of issues.

I know i've tried them all from Windows NT 64-bit alpha, Windows 2000 64-bit, Windows XP 64-bit, Windows Server 2003 64-bit, Vista 64-bit, Windows Server 2008 64-bit to now Windows 7 64-bit.

The perks of working in a computer store in my younger years in the late 90's/early 2000's.

[PinothyJ]

I remember being a bit against it being an option with Vista. I think part of the slow change may have been (and correct me if i've heard wrong) that the only real world difference is that you can access more RAM (maybe slightly faster load speeds).

This is true. The memory ceiling is 2.95GB (2.75GiB) for 32bit and 18.45EB (16EiB) for 64Bit.


Of course, you need a CPU and motherboard with support for the memory amount as well…

[cyclobs]

I remember being a bit against it being an option with Vista. I think part of the slow change may have been (and correct me if i've heard wrong) that the only real world difference is that you can access more RAM (maybe slightly faster load speeds).

Kind of, you also get a bit more processing speed out of it as well. Since you are now processing 64bit instructions for every cycle rather then 32bit instructions. So basically you're processing your data in larger chunks over 32 bit which 'could' lead to an performance increase but not really one you'd see in a games i believe, other then being able to access more then 3.4GB of memory.

This is why 32bit programs die with an error of 'out of memory' because they've used up all the memory they can be allocated even though you have 12 - 16GB of ram installed.

Things like encoders, decoders and encryption software will benefit because of how they process the data.

I had to do some research on it back in tafe.

This is true. The memory ceiling is 2.95GB (2.75GiB) for 32bit and 18.45EB (16EiB) for 64Bit.

i think i read somewhere that the limit of 3.4 (or 3.6GB) was an artificial limit that they hard coded into windows. Using tricky techniques they could get much more ram allocation out of 32bit but it became very unstable or something.

[Ash_Williams]

Server 2008/2003 in 32bit will load upto about 12-16GB of RAM, the same goes for my Mac Pro which runs a 32bit EFI, Kernel and OS and it accesses all 16GB of the RAM in it.

32bit can access upto 32GB of RAM from memory but it comes down to how things are coded to access such amounts without becoming unstable.