Are 64 bits really faster than 32?

[PavkaGuru]

I have heard that 64 bit architecture does not outperform 32 bit because 32-bit applications do not take advantage of the oversized processors since they run under a backward compatibility mode. To take advantage of the 64-bit the old 32-bit apps should be rewritten or at least recompiled.

In other words, buying a 64-bit laptop, configured with Win 7-64, is just a waste of money if I don't buy 64-bit apps. Is that true?

 

[JFAMD]

*Generally* speaking 64-bit is about larger memory addressability and not about 64-bit instructions.

 

[Fayd]

the main advantage of 64 bit is the greater available addressable memory space.

that's available to you regardless of whether you use 32 bit or 64 bit programs.

where on windows XP or 32 bit 7, you would cap out your memory somewhere around 3.5 GB (depending on video card RAM), on 64 bit windows 7 you would have the full amount of whatever ram you installed in the machine. on a modern i3 based laptop, i think you can install up to 8 gb, for example.

if you only install 3 gb or less of ram, and only use 32 bit programs, then no, there's no benefit to 64 bit.

however i dont see it as a waste of money, because 32 bit and 64 bit OS's cost the same. and going with 64 bit gives you greater compatibility (with 64 bit programs), and expandability should you choose to install more than 3 gb.

 

[JAG87]

The main advantage is that you can map and address much more than 4GB in memory.

 

[sandorski]

Speed isn't why 64bit is better than 32bit. It's mostly, at least in practical terms, the extra RAM capacity. There's really no reason to use 32bit anymore, it is being phased out.

 

[Scali]

32-bit apps don't run any slower either, so you're not exactly losing anything by going 64-bits.
At the very least you can make better use of your memory (most computers have 4 GB now, many even more, you can only use ~3.3 GB in a 32-bit version of Windows). Each 32-bit application can get up to 4 GB of memory.
And you also get the benefit of running 64-bit applications. Things like .NET and Java will generally already run in 64-bit by default.

 

[Harvey]

Except for custom ordered/built machines, virtually all laptops made by major manufacturers and sold through commercial chains come with a 64 bit CPU, and all but the most RAM crippled come with the 64 bit edition of Win 7. That makes those machines the most available and affordable.

They may be overpowered for running your older apps, but as long as they run well, a machine with the expanded capability won't hurt, and it will handle newer, more demanding apps and remain useful longer.

When you can buy a decent laptop with at least a dual core, 64 bit CPU, 4 GB of RAM, > 320 GB HD and an LED backlit screen for aroun $400, you're spending more time over-thinking a problem that doesn't exist, and you probably won't save any money buying a lesser machine.

Hope that helps.

 

[VirtualLarry]

64-bit is faster, but apps have to be recompiled to take advantage of 64-bit mode, which supplies extra registers to use.

 

[aigomorla]

64 bit os allows greater then 3gigs.

You will have some compatibility issues on 32bit aps.

There is also a lot less viruses you can get on a 64bit OS.

 

[VirtualLarry]

There is also a lot less viruses you can get on a 64bit OS.

That's because of certain features that MS added to the 64-bit OS kernel, not because of 64-bitness in general.

 

[aigomorla]

That's because of certain features that MS added to the 64-bit OS kernel, not because of 64-bitness in general.

thanks for that info.

Not really into software... my friend told me, 64 bit was less suspectable to viruses.

 

[Absolution75]

It's called PatchGuard
http://www.microsoft.com/servers/64bit/x64/benefits.mspx

It basically keeps rootkits from installing. This to me, is just as important as the ability to use more memory.

 

[fluffmonster]

Not any more. there are now 64-bit rootkits in the wild that can penetrate the kernel.

 

[sandorski]

One downside is the twofold increase in Child Choking Hazard.

 

[mv2devnull]

One downside is the twofold increase in Child Choking Hazard.

In a sense yes.

Addressing larger memory requires larger address-numbers. Storing larger numbers takes more space. Therefore, when you compile same source to 32-bit and 64-bit binaries, the latter probably needs a bit more memory (but has other benefits due to it).

Whether such miniscule change in memory usage actually changes performance ... I would not worry about that at all.

 

[Absolution75]

Not any more. there are now 64-bit rootkits in the wild that can penetrate the kernel.

Its not to say its impossible to break - but it does add a significant amount of security.

 

[Schmide]

A couple things about 64bit OS/programs.

On a 64bit OS, the environment seen by 32bit apps and generally for 64bit apps looks exactly the same as the 2gb(upper 2gb os) address space used in a 32bit OS. Advantages come because of extended functionality of page tables to physical memory. In actuality much of this functionality can be mimicked by PAE on a 32bit processor, it is much cleaner/easier on a 64bit processor.

When running in 64bit Long Mode, 64bit register usage or must be done with a REX prefix instruction (i.e. default data type is still 32bits) This provides the ability for 32bit compiled programs to run in a 64bit address aware environment. This is called compatibility mode.

The only thing I could ever see causing problems with a 64bit OS is at the driver level as they must be written to handle both legacy and long mode.

 

[PavkaGuru]

The main advantage is that you can map and address much more than 4GB in memory.

Ok, I realize that you can address memory beyond the 4GB barrier. But the 32-bit apps would take advantage of that extra addressable memory? I have an US$15,000 engineering application which makes an extensive use of the 4GB installed memory. When running, it causes the OS (Win Vista Pro) to perform an intensive memory pagination. If I buy a new 64-bit laptop configured with 64-bit Win 7 and 8GB RAM (or more), is there a chance that memory pagination could be lowered? Take into account that the engineering software is a 32-bit app.

Thank you for your help

 

[RavenSEAL]

Nope, George W. Bush just needed something new to lower the taxes on.

This is not P&N. Try your comments there, not here
Markfw900
Anandtech Moderator

 

[Schmide]

If I buy a new 64-bit laptop configured with 64-bit Win 7 and 8GB RAM (or more), is there a chance that memory pagination could be lowered? Take in account that the engineering software is a 32-bit app.

Thank you for your help

No chance without a 64bit compile.

 

[Scali]

Ok, I realize that you can address memory beyond the 4GB barrier. But the 32-bit apps would take advantage of that extra addressable memory? I have an US$15,000 engineering application which makes an extensive use of the 4GB installed memory. When running, it causes the OS (Win Vista Pro) to perform an intensive memory pagination. If I buy a new 64-bit laptop configured with 64-bit Win 7 and 8GB RAM (or more), is there a chance that memory pagination could be lowered? Take into account that the engineering software is a 32-bit app.

Thank you for your help

It might make a difference.
In a 32-bit version of Windows (server versions excluded), everything has to be mapped into the 4 GB address space. That includes your applications, drivers, kernel, memory-mapped I/O etc.
As a result, you'll never be able to use more than ~3.3 GB of memory in total... and of that, only about 3 GB can be used by a single application at most.
In a 64-bit OS, the drivers, kernel, memory-mapped I/O and such can be mapped outside of the range of physical memory (just like in the old days in 32-bit, when you didn't have enough physical memory to run into the address space shortage issues). As a result, a single 32-bit application can use the full 4 GB addressing for physical memory without modifications.

 

[RavenSEAL]

Nope, George W. Bush just needed something new to lower the taxes on.

This is not P&N. Try your comments there, not here
Markfw900
Anandtech Moderator

That was supposed to be a joke...

 

[Schmide]

As a result, a single 32-bit application can use the full 4 GB addressing for physical memory without modifications.

A 32-bit app has to be compiled with /LARGEADDRESSAWARE to take advantage of the full 4gb space. The default virtual page environment is 2gb.

http://msdn.microsoft.com/en-us/library/aa366778%28VS.85%29.aspx

 

[Scali]

A 32-bit app has to be compiled with /LARGEADDRESSAWARE to take advantage of the full 4gb space. The default virtual page environment is 2gb.

http://msdn.microsoft.com/en-us/library/aa366778%28VS.85%29.aspx

Yes, but I took the liberty of assuming that a $15000 engineering application would be linked (not compiled) with that flag.

 

[Voo]

Yes, but I took the liberty of assuming that a $15000 engineering application would be linked (not compiled) with that flag.

You could also take the liberty to assume that a $15000 engineering application could possibly run something other than Windows - PAE support on Windows is extremely anemic without some (rather trivial, but still) Kernel hacks.. just to add that as well

On the performance side: The doubled size for pointers isn't so much a problem for the memory, but you waste also 50% of you cache and cache is expensive and rare (well you can use more arrays than pointers to reduce that) and depending on the data/algorithms the 64bit ISA can speed up things quite a bit. Oh and more registers are always nice.