Microsoft Working on 128-bit Windows

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
They are researching 128bit as it applies to future products. Ten years ago if you told me I'd have a PC with 8 GIGS of RAM I'd punched you in the face. Ten to fifteen years is the time frame they are planning for which is an incredible step for any business. Today I say in 15 years we as PC users wont need it, but servers are already running 64+ gigs of RAM. How much RAM will they be running in 5 years? 10 years?

Also, a lot of are thinking: higher bit = higher speed. The bits are referring to the size of the memory bus. The only time you'll see an improvement in speed is when you address a file/program that is larger than the old bus can address in the RAM. So you're speed boost comes from being able to use more RAM and access the page file less often.

So the actual thinking should be: Higher bit implies more RAM = higher speed (when working with large files)

I think Tom's did 32bit Vs. 64bit review a long time ago and the 32bit came out on top with most applications because there was less bus to calculate. I'd like to see an updated article for this since we can test 4gigs (32bit high speed) Vs. 8gigs (64bit high speed) and have programs that actually take advantage of the extra bus size.
 
And we need 128bit because????

Wait, how much memory access does Win7 64bit have? Lets see...
16GB for home... fine. Most Win7 users won't need more than 2GB.
192GB for Pro+... anyone in the market for such a motherboard?

Okay, Tyan has a 256GB mobo for $800 (nVidia chipset)
Tyan and SUPERMICRO has 128GB mobos for $300~500 (AMD and nVidia chipsets)

Hmmm... 128GB of DDR2-EC is about $6400+ The 8GB DIMMs aren't out yet. So current 4GB x 16 = 64GB = $3200... Crysis had better run good.

And so consumers need 128bit for what?
 
[citation][nom]izliecies[/nom]What is the point of going higher than 64 bit?[/citation]

If I remember correctly the term 64 bit refers to the largest single piece of information the processor can handle. Inside the processor there are things called registers. You put a piece of data in these registers and then you can perform operations on them. An example would be putting the number 4 in the AX register, and the number 5 in the BX register and calling an ADD operation to get the answer placed in the CX register. Then you can do whatever you want to with the result in the CX register, such as move it into system memory.

That is the basic gist-of-it, I think. The need for larger registers comes from the need to use multiple operation on a single piece of data. An example would be working with a double precision float data type (64 bits of memory are required to work with this data type). On a 32 bit system the ability to add two doubles together would require multiple operations at the register level. On a 64 bit system it would only require one. Thus leading to smaller and faster code.

Of course there is also the need to address larger amounts of memory, but as you can see in other posts that is no longer a problem. This is an issue with register size because the memory address you are trying to access (read or write) must fit into a single cpu register. Thus the maximum addressable memory on a 32bit system 4GB, and 16EX on a 64bit system.

But like I say, "If I remember correctly". It have been years since school, and even longer since my assembly days on a C64.
 
first off, most of you guys are misunderstanding the basics of computer architecture. IA64 is not AMD64 aka x64. IA64 is the platform for Intel's Itanium RISC processors used in huge super computers and high end servers. IA128 I would assume would follow suite, which means Microsoft let the cat out of the bag before Intel's official announcement of the next generation Itanium. However, rumor had it that Sun's next generation OpenSparc/UltraSparc processors will be 128bit.

So to answer your questions, "NO, it will not run Crysis. It will not be a consumer level Windows." The purpose of having a bus that wide will significantly help database and other cloud applications.
 
[citation][nom]jasperjones[/nom]Merely two years ago, if you wanted to run a 64-bit OS, your best bet was Linux x86-64. 64-bit Windows Vista had driver issues. And not just with uncommon hardware. Of course, hardware support in XP Pro 64-bit is pretty bad, even today.So I'm glad to hear they're working on 128-bit now...[/citation]

Yeah but nobody really supports XP directly anymore anyhow, considering Vista is around, and is now being replaced with Windows 7. Sure lots of people are stilling using XP, but it's not the current or future OS, therefore companies don't have as much reason to offer direct support for XP in new products (or even old products which they themselves don't support anymore).
 
No he didn't, meat81 ruined it for you in the first post. You just didn't see it because it was voted down. It's a dumb thing to say, and it's not funny.
 
It wouldn't surprise me if 64bit goes the way of DX10... slight benefits, more real trouble than it was worth and its lifespan cut short by a future development... Mass adoption to hit 128bit...

I can see it happening to be honest.
 
[citation][nom]JasonAkkerman[/nom]If I remember correctly the term 64 bit refers to the largest single piece of information the processor can handle. Inside the processor there are things called registers. You put a piece of data in these registers and then you can perform operations on them. An example would be putting the number 4 in the AX register, and the number 5 in the BX register and calling an ADD operation to get the answer placed in the CX register. Then you can do whatever you want to with the result in the CX register, such as move it into system memory.That is the basic gist-of-it, I think. The need for larger registers comes from the need to use multiple operation on a single piece of data. An example would be working with a double precision float data type (64 bits of memory are required to work with this data type). On a 32 bit system the ability to add two doubles together would require multiple operations at the register level. On a 64 bit system it would only require one. Thus leading to smaller and faster code.Of course there is also the need to address larger amounts of memory, but as you can see in other posts that is no longer a problem. This is an issue with register size because the memory address you are trying to access (read or write) must fit into a single cpu register. Thus the maximum addressable memory on a 32bit system 4GB, and 16EX on a 64bit system.But like I say, "If I remember correctly". It have been years since school, and even longer since my assembly days on a C64.[/citation]

Common assumption (in this thread too), and DEAD WRONG!
In any talk of 16-, 32-, 64-bit processors, software and operating systems, It's always about the length of adress pointers. This is the crucial issue.
In terms of width of registers, data, data paths and processing, we've already been at 128-bit for a long time, with 32-bit processors/software/Windows. That's essentially what all these SSE extentions are about, since the Pentium days.
 
[citation][nom]liquidsnake718[/nom]That was the first thing I thought of. If 16gb for Win 7 is maxed out, then 32gb would be the estimated Gb amount of Ram since 128 double the bit. Wow I remember when whole console systems were 128bit and THAT seemed revolutionary in 1999! This is a damn OS which is virtual!I can imagine 32 GB or even 64gb worh of RAM. By then our Ram might even be 32nm SSD type of RAM or something.... But I suspect we wonot be seeing this until 2015 or WIN9 by 2018... Still a long way to go.[/citation]

Win7 isn't maxed out at 16Gig, that's your Motherboard. Win7 will support up to 192Gig in the client versions.
 
P.S. Also, am I right in thinking that because 'Robert Morgan is working to get IA-128 working backwards with full binary compatibility on the existing IA-64 instructions'... does that mean that they have a working model of Win 8 in 128bit mode and that they are just focusing on compatibility?
 
And there is absolutely no point to a 128-bit OS.
This "IA-128" is either a complete fraud, or it's a name for something else. What I'm saying is that the "128" in that case does NOT concern the adress width of an ISA.
 
There's no reason why 128-bit would ever be needed. People say: "Well, in 1990, I never thought I would need more than..", BUT, computers are much more mature now than they were in the 16bit and early 32 bit days. 64bit vs. RAM limitations is a bit of a misnomer anyways, Windows Server Datacenter edition has supported huge amounts of RAM(32/64/128GB) even as a 32bit OS, thanks to NUMA. Developing a 128bit OS before AMD or Intel has devised a x86_128 standard makes even less sense...


Believe it or not, computers will eventually be 100% mature, with no more room to improve. How much can you possibly discover before you've discovered everything?
 
-FFS! Are you all so clueless about how a computer works!
64-bit adress pointers are good for 16 ExaByte. And that's:
16 X 1024 X 1024 TeraBytes.
Even with Moore's law going on, 64-bit will last for decades.
 
This is not for you guys... at least for near term future.
this porbably will be for NASA or weather center, where they need to analyze huge data in very small period of time

So maybe google will have it for it's search engine... where it will be able to analyze 1mln photos in 1 second and give you a picture with a small hair fluctuation....

it is good thing to have all of this.

but do nto forget, soon, maybe HDDs will be able to work on the memory speed, then, to have so much memery will not make that much difference.

So relax, let see how it will turn out
 
WTF is everyone babbling about. 64bit is capable of addressing 2^64 bits or 2,147,483,648 Gigabytes. Windows memory limitations have nothing to do with 64bit.

128bit will have a practical application when they make 1 exabyte ddr35 modules.

Furthermore 64bit doesn't make code faster than 32bit, so 128bit doesn't have a direct performance impact.

128bit will not happen for Windows 8. There is no reason for it, 128bit hardware is far far away for PCs.
 
[citation][nom]hellwig[/nom]They said IA-128, which means Itanium.[/citation]

You sure? I doubt it. IA-32 means x86 after all. IA-64 is Itanium whereas Intel 64 is Intel's implementation of x86-64.

But it's doubtful that Intel will continue to develop Itanium longer than it is contractually required to do so. Itanium is not a money-maker, contracts with HP are the primary reason Intel doesn't dump Itanium (just yet).
 
[citation][nom]liquidsnake718[/nom]That was the first thing I thought of. If 16gb for Win 7 is maxed out, then 32gb would be the estimated Gb amount of Ram since 128 double the bit. [/citation]

That is not at all how the limitation is calculated, that's just Windows 7 boundary as it stands now, the actual memory limit of 64-bit is 2^64 Bytes, which is somewhat larger. If you want to know the maximum memory supported by xx-bit hardware, then take 2 and raise it to the xx power.
 
Im sure there has to be to the 128-bit architecture than just memory access. Granted Im sure the supercomputers will have the need to access more than what a 64-bit operating system can access but there has to be something they know about the direction hardware is headed that is causing them to do this. Every piece of silicon is getting smaller and smaller every year and we are on the verge of CPU and GPU integration. Maybe that has something to do with it.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom