64 bit vs 32 bit

So, i am not sure i get the difference between 64bit and 32bit. I have a Sony Vaio Z, VGN-Z540, can i install 64bit version on that? And can someone quickly ecplain what the diff is between the two?

 

main difference: 64bit can use more than 4gb ram.

the laptop has a core 2 duo, right? if so, yes, it can.

 

if your laptop has a 64bit cpu and 2gig+ ram
then its 64bit all the way, 32bit should have died years ago
i have been 64bit since march 07

 

there's no gain for him, though. which is why i haven't moved to 64bit yet. there's no need to if you're <4gb ram. 32bit, in some cases, has more compatibility in the driver department. i will move to 64bit, but i guess i wait till win8 for it.


edit: and before you come up with some random placebo-gains. 64bit is a bit bigger on disk, and needs a bit more ram for the same tasks. so on the same environment, espencially on ~2GB ram situations, 32bit is better.

 

Feel the POWER with 64bit - the Smart choice:

1. 64bit can run much faster than 32bit since the CPU can process twice as many bits in a single operation.

2. 64bit offers better native mathematical precision.

3. 32bit can only address a total of 232 memory locations (4GB RAM). However not all 4GB RAM is available since some addresses are reserved for system resources such as video memory.

The computer industry went through a smilar transition 10 years ago (16bit to 32bit). The days are now numbered for 32bit. It has been rumoured that Office 2010 will only be available as two separate products - 32bit or 64bit. If this is true then it would make sense to go for the 64bit version since it will last longer and be portable to any new computer.

Most companies are now producing new 64bit drivers and are compatible. I wont buy software unless it is 64bit or at the very least 64bit compatible.

Note - to use 64bit Windows you need a 64bit CPU - Core 2 is 64bit.

Feel the POWER with 64bit - the Smart choice.

 

1) untrue as, for the most part, 64bit and 128bit operations exist even in 32bit chips. there are only tiny improvements.

2) no, as this is in the 32bit chips in as well.

3) no clue how you get to those 232 memory locations, but yeah, up to 4gb.

the only thing that matters about 64bit is that it can access more memory.

there are TINY improvements in the basic spec of x64 compared to x86. those allow for SOMETIMES a bit faster computational performance.

 

Depends on operation if the data size exceed 32bit if you were to use 32bit instruction set it may need to run twice compared to once on 64bit.

Nope FPU is still the same, depends on the processor anyway
Note:64bit refers to general purpose register size not floating point registers.

 

If you want to get Technical then OK

64bit operation is 2 times more than 32bit operation. However the reality is more complex because 32bit CPU have various enhancements that allow them to extend their data precision - For 32bit SIMD registers are 128bit so yes you are correct SIMD registers which are 128bit. Again, However, you will see the main benefit with 64bit when using encrypted data.

64bit binary value can represent a decimal to around 14 significant digits, while 32 bit can only represent around 7.

SO YES FEEL THE POWER WITH 64bit - The Future

 

I find the truth to be somewhere in the middle

x64 OS allows you to use more ram yes. Because it has more memory address space. However even in x86 you can use all 4GB of RAM as long as your motherboard supports PAE (address extension)

As for x64 vs x86 speed/improvements. Well first of all not anything really going to happen unless you are not only in a 64bit OS environment but also using a 64bit binary.

From my personal testing I find x64 vs x86 binaries to be a mixed bag, usually they are faster in one thing but then slower on another. x64 is taking the lead now overall however BUT its very situation dependent.

One of the most cpu intensive tasks out there is video encoding, it maxes all 4 cores on my cpu to 100% from start to finish. All the UI and precompiled encoders out there for H264 are using a x86 binary.

A x64 compiled binary actually is about 15% faster for me, thats pretty significant and its all just because it can harness the extra processing power of the x64 architecture.

So I say in most cases (97%) x86 and x64 do not have a large enough difference in performance to even notice, but there is a time and place where x64 has a clear advantage.

The memory thing comes into play as well. Photoshop CS4 has a 64bit binary now (for windows not mac ) and being able to give photoshop 4+ GB of ram to work with is really going to help when working with large files.

Maybe you guys didnt know it, but even if you have 20GB of ram and a x64 OS and your SYSTEM is using 20GB of ram your programs will still be limited to <4GB unless the binary too is 64bit.

Hope I enlightened somebody today

 

if you want to get technical, then the net result is, 64bit processors allow more ram, and have some tiny improvements. anything else is simple nonsense.

if you want, you can read the x64 spec trough to find out any of the differences. hint: there are nearly none.

we got some more registers, that allows more simple compiler optimisations, thus a bit better optimized code.

and the 64bit integers we have now, we had them before trough mmx, and they are not often needed or used anyways. how often do you need numbers > 4 billions? especially where you don't need them on 32bit? that happens about never => you have to implement 64bit math on 32bit systems anyways, too, if you want your app to run there. so you use mmx, then (or even sse3 which can do 128bit integers.. or is it sse4?), and can reuse that code directly on the 64bit processor.

64bit has NOTHING to do with 64bit math. that was implemented in one way or another since pentium 2. and got used.

floatingpoint is 100% independent of the bitness of the cpu anyways, as it's always 80bits, or 32 and 64bit in the sse case.

integer always had options for 64bit math.

the only thing that had no 64bit option yet was memory addressing.

that was the LAST and ONLY thing that needed the change. and it needed it for >4gb memory addressing.

 

16 bit vs 32 bit is alive again 10 years later with the 32 bit vs 64 bit. Where is 16 bit Windows now? Goodbye 32 bit. The FUTURE - 64 Bit

 

exactly. but more on my side.. *smile*

actually, pae allows more than 4gb ram, not? without it, i can adress exactly 4gb ram. except if some of it is reserved for hw remapping.

indeed. and a direct ported binary, direct ported to 64bit will be bigger, use more memory, and more bandwith. thus it will be (marginally) slower, just from the start.

mainly due to the more registers. they allow compilers to work more simply optimize complex code.

see above

which is the estimated average gain you can have, based on the new extensions that are default in x64.

x64 has some gains. 64bit not. x64 is a new architecture with some new features. 64bit memory accessing is one of those features, and in itself would not have any further gains.

this is where the real power is.

except if it's a multiprocess-app, like the new ie8, or chrome. there, each tab can have 4gb ram. no clue if that will ever be a limit => no need for browsers to go 64bit, really.

jup, as always, your posts are in general enlightening.

 

so much for "going technical" you have no clue about 64bit, but just love it cause it's hip

nah, just kidding. but your 2x as fast is 100% nonsense, that is true..

 

You need to re-read all my posts on page 1 (this is not what I have said). You should see the statement, 'However the reality is more complex' - then an explanation about the SIMD register.

 

page one? i have only one page..

but yeah, you haven't said that, it just sounded a bit like that.

then again, i allready replied that it's nonsense that it computes faster because it has more bits. that doesn't by default reduce the instruction count, thus the workload, thus the computational speed. it can, in some times, help. but not in general.

thus the most gain you can get in a general app is about 20%.

 

I have 2 pages.

Don't worry about it - it's all good fun When Windows 8 comes out you will have completely forgotten about 32 bit - because Technology MOVES ON - 64 bit is the future in my opinion and I would rather move to 64 bit NOW (especially if you have to make a choice of 32 or 64 when buying Office 2010 - if rumours are correct).

 

well, i have both. and i have 0 gain from 64bit. and i'm a developer, and worked myself into 64bit. <4gb ram, there is no real gain except for being hip.

but it sure is great to have 64bit available, and it will be the future. on the other hand, it's great that win7 doesn't require it, and we can run it just well on old p3/p4 based systems.

and i love that i can still use some of my special hw that by now gets it's first 64bit drivers (a.k.a. still buggy compared to the years of practice 32bit drivers).

so i sure support going 64bit. but it's nothing magic.

 

I don't think anybody is saying it is magic, are they? However if Office 2010 is only going to be available in 2 separate products (32 or 64 bit only), then I will be buying Office 2010 64 bit. (again assuming rumours are correct).

Hence I am going 64 bit now.

 

it won't matter for office. how often does office use 4gb ram? i guess never. how often does it use much cpu? i guess, again, never. so having it 64bit will be zero gains.

 

I know.

But if Windows 8 is 64bit only, then obviously you will need Office 64bit. If MS make us choose between buying Office 32bit or buying Office 64bit, then I will be buying Office 64bit (since this will be required for future Windows/Windows 8).

All my current applications all work on 64 bit, so I am OK. So it is Windows 7-64bit all the way for me.

 

well, you don't buy 64bit. you buy a license, and if it is anything like.. anything else microsoft does currently, that is bit-agnostic anyways. you can install 64bit if you want (on 64bit capable systems. not all you buy today are. some atoms aren't).

and win8 won't be 64bit only, i guess. at least not the starter edition. it would be incapable of getting used on all those old p4, or even p3 system. which can handle it just fine, and might get put to use in some non-rich environments. and microsoft doesn't want that piracy wins there. so chance is high, they let their 32bit codepath coexist. but go mainly 64bit next time. maybe win9?

 

FOR original Poster cstassen

Anyway this is what I think

But Remember

However the reality is more complex because 32bit CPU have various enhancements that allow them to extend their data precision - For 32bit SIMD registers are 128bit so the speed differential (between 32bit and 64bit) at present is marginal. You will see the main benefit with 64bit when using encrypted data.

Remember also 32bit will begin to fade out soon.

Also Windows 7-64bit also has a 2GB RAM requirement.

 

there is no speed difference between different data precision type. they are identical in speed. they only have different precicion.

and i would reorder your points to have 3 at the top. because it's the one major feature, the reason we wen't 64bit at all.

and espencially 2 is not true. there is no new more precise type on a 64bit processor available that wasn't before. the only change is, now a memory adress is a 64bit integer, before, it was a 32bit integer. it only changed for the memory adresses. all the others HAVE TO STAY IDENTICAL so apps don't compute different numbers on different typed cpus.

 

If you want to install a 64Bit OS on a Vaio, search the Sony forum else you may end up missing a couple of Fn keys.

And see my sig for a little outline on what the difference is...

 

It all depends on how the application is written. Even in a 32 bit CPU apps that need high precision or fast mathematics tend to be written specially to use SIMD registers (128bit wide) rather than the general purpose 32 bit register.

 

yes, and even those use the same simd on 64bit.

an app has to be REWRITTEN for 64bit to benefit from the larger registers there. and by default, it won't. none of them. because the gains are so minimal, it doesn't matter. most apps don't need integers > 32bit, that didn't on 32bit systems.

that's what i wanted to say. no application suddenly needs higher precision on a 64bit system than on a 32bit system, so NONE will change any of it's logic. the only thing that changes are the memory addresses. those will get expanded from 32bit to 64bit.

it will take years till some special application will be 64bit only AND developed from new. only by then, some will maybe benefit from those additional high precision registers for anything else, except for storing bigger memory addresses in it.

 

64 bit does have better NATIVE mathematical precision. A 64 bit binary value can represent a decimal value to around 14 significant digits. Whereas 32 bits can only represent around 7. if a 32 bit application needs 64 bit accuracy, it must spread a " double precision" value across two CPU registers, which can be slower.

 

For Reference my use of 232 memory locations (4GB RAM) is just sloppy for 2 (to the power) 32 or 2^32 = 4,294,967,296 bytes = 4GB

 

Its not just "sloppy" - its near disastrous...

How is any sane person expected to understand 232 = 2^32 - no one ever will, because its not.
One is the number 232
the other 2 to the power of 32

 

That is why I used the (4GB RAM) statement so it was obvious. 232 memory locations is still valid speak though. It's just that the arm chair computer user generally talks using the 4GB RAM limit. That is were the 4GB limit comes from (2^32). I suppose it is actually the other way round, in that the word 'sloppy' was probably the wrong choice of words.

My orginal post (first post) was only intented to be a quick 1, 2, 3 of the main difference of 32 bit vs 64 bit. The problem comes when you have to start explaining every little detail, it then starts ramping up the intensity of the thread. There are some other differences also.

I suppose in reality, most people want 64 bit because:

1. To use 4GB/more than 4GB of RAM.
2. Technology moves on and the future is 64 bit.
3. Encrypting Data/Cryptography (64bit speed increase on this application).

might be able to make a case for better security on the 64 bit kernel too. Possibly robustness too

I suppose the main problem with 64 bit:

1. Possible compatibility issues with Drivers and Software.
2. Possibly because Windows 7-64bit has a 2GB Minimum RAM Requirement.
3. Possible because you require a 64bit CPU.

Me personally going 64 bit now. Collecting/Only buying 64 bit compatible software now. If you out grow your laptop/pc during 2010 then it is best to have a 64 bit version of your program applications software (especially if you have paid good money for it). Although as someone previously mentioned, you are OK with Photoshop CS4 since it installs both 32 & 64 bit (on typical default install) on Windows 7-64bit. This is not the case for all software though.

Maybe someone else might be able to add some other ideas?

232, 264 See ya later.

 

Cinebench R10 CPU Render test gives much better values to 64bits (means less time to render the motorcycle image) than the 32bits one.

For Cinebench http://www.maxon.net/index.php?id=162&L=0

No need to install just run 32bits and 64bits. It has also a OpenGL test.

 

I think Cinebench does utilise SSE or the XMMS registers.
So sometimes data has to be transferred from the SSE registers to GP registers.
64bit makes up for some of the precision I think so it doesn't have to take more clock cycles to transfer the same width of data.

 

did you have researched this, or just took it out of your ... somewhere..

because nothing of this is true. significant digits only matter in floating point, which, since their first appearance in p1 or p2, are completely UNRELATED TO MEMORY ADDRESS BITNESS.

floatingpoint since the first day are 80bit.


and since mmx, no cpu ever had to care about spreading a 64bit value into two registers. none. and even compilers spit out mmx automatically.

and this is only an issue for integer (whole numbers) and integers emulating fixed point values. the only place i ever met 64bit integers for SOME useful tasks was when writing software renderers. and even there, you could work around it easily.

you just mix and match how you like, it doesn't make your statements right..

i do like your effort, though.


another info: floatingpoints are never stored in the integer registers. just in case, they have their own registers. and they are 80bits. so they won't ever care if the cpu itself has 32bit of 64bit registers. not at any single moment.

 

no. never. i never heard that weird use, ever before. and i'm with a lot of geeks, in a lot of different computer scenes. i know tons of strange namings that are "valid speak" for them.

never heard yours, and it makes no sense. ^32 would make sense, f.e. and would be the same amount of keys. or simply 32bit why not?

 

Please don't argue.
MMX is 64bit however since the advent of SSE developers ought to be coding with SSE registers which is 128bit.
It is true with regards to what Sienna said

However you shouldn't be using MMX in the first place instead you should be using the higher precision SSE Registers because SSE was suppose to replace MMX.

 

sse for the most part was floatingpoint only. mmx by default is integer/fixedpoint.

don't mix those two. they have nothing to do with each other.

and the register size of sse components, yes, that's 128bit. but the actual value sizes that the computer can work with in there are 32bit and 64bit floatingpoint (and since not so long, integers, too).

but at close to no point in modern software development, the following quote is true anymore:

that's what i'm saying, and that's why i'm still saying, the only real important thing about going to 64bit means >4gb ram. all the rest are minimal spec changes that won't matter really.

 

Microsoft has a free download for Vitural PC. With this program running in your Windows Vista 64 bit environment, you can run a Windows XP 32 bit environment, and launch an MS DOS environment from there. Additionally, you can create as many of these virtual environments as you want.

Unfortunately, Microsoft requires you to have both a Windows XP licience iand a Vista Licence if you are going to run both Vista and XP (even with one in the virtual world). If you own a regerestered copy of XP, I'm told that a work around to this potential problem is to reinstall your virtual XP environment when the preceeding one experies it's 30 day registration requirement.

Good Luck!

 

Too much blablabla. 64 bit is the future, it more than obvious, if you can install your os 64, go 64. what are you going to do the day you decide to add some ram in that PC? or when 32 start getting phased out, reinstall?

Just install 64 and move on

 

Add RAM - that is if you want to pay that amount of money for RAM - at the moment 4GB "sticks" are rather expensive.
And 32Bit isn't phased out for a while.

In the long run, yes it is "the future" - but is doesn't have to be the best choice just yet.
Most software supports 64Bit fully - I'm not sure on the status of peripherals though - and if you have an older printer you might run into trouble.

 

most of the systems will happily run with never-more-than-4gb ram for the next few years, till they get phased out anyways. why should i actually reinstall at all?

 

I found Photoshop CS4 Extended (11.0.1 update) to be fine on 64 bit (the default installs both 32 and 64 bit versions). However, there are reports all over the forums of people having problems with Photoshop Elements 8 on 64 bit. Elements 8 product information does not mention 64 bit? I wonder how elements 7 & 6 fair? Some big Anti-Virus products don't fully support 64 bit either. There are big names out there that are not fully 64 bit yet.

Most current software is optimized for 32 bit. However this will soon start to change.

If I remember right don't some of the first/original Core i7 builds (allbeit pc builds) use triple memory modules (3 x 2GB), rather than pairs, and have 6GB RAM ? So you might not need 4GB sticks.

 

So, CS4 has 64Bit support, yes it is faster in 64Bit, but it works on 32Bit too.

And you just named some software that could benefit from 64Bit (Photoshop Elements) which apparently has problems.

The RAM - yepp, you are right However from what I heard this isn't yet available on laptops - they still use just two slots.

 

Yes I believe you are right.

If anyone is interested:
Windows 7-64 bit has both 32 bit & 64 bit versions of IE8. Some products require you to use the 32 bit version of IE8.

I also think I remember seeing somewhere that Flash 10.0.42.34 (or whatever) is only 32 bit as well?

 

I think I heard the same for Flash

64Bit has only become a full OS on Vista - on XP it was a test OS.

Win7 is the first Windows that aims to bring 64Bit "to the masses" - however, unless there is decent support it will be a fiasco just like Vista was due to lack of driver support...

I suppose in terms of pure processing power - Photoshop could easily use 128Bit and 256Bit once they arrive in a few years/decades whatever...

The average user needs something that works and at the moment this is still 32Bit - 64Bit works generally - until you find the few cases where it doesn't...

 

That is true.

I think what will kick start 64bit this time is the '4GB RAM' limit of 32 bit. The PC market will drive this. So in my opinion 64 bit has to take off this time. In order to get value for money, I only really want to buy new software that is 64 bit compatible.

 

If you can that's definitely a good idea

The only problem may occur if you want to use older devices out printer is... hmm, nearly 5 years old Canon inkjet, still works

Definitely works in XP, Vista, Win7, not sure if 64Bit though... but some printers wont...

 

True. Luckly all my external Hard Disks and TV tuner work fine with 64bit. I do have several old programs that work in XP but failed in Vista and 7. Sun's VirtualBox solved this problem in Vista 32bit. Luckly I have newer versions now anyway.

 

Some 64 bit Technologies to Improve Security & Stability (so I believe) on Windows 7-64bit, including:

1. 64 bit Kernal Patch Protection.
2. Hardware enforced Data Execution Prevention (on compatible CPUs).
3. Automatic rejection of drivers without a trusted electronic signature.

 

Yes, there are some security benefits

Although it can make life difficult too on the signed driver issue for example.

 

all possible in 32bit. they just haven't enabled it for compatibility reasons.