| View previous topic :: View next topic |
| Author |
Message |
Re-JeeP
Apprentice
Joined: 10 Oct 2006
Posts: 294
Location: Sweden
|
 Posted: Thu Aug 09, 2007 1:58 pm Post subject: Intel 32-bit VS. AMD 64-bit processors |
 |
|
Hi!
I have read some articles about Intel 32-bit VS. AMD 64-bit processors. The only articles / forum discussions are all pretty old. So I would like a status report of the situation today when the new Intel Dual Core has been presented.
And I am of course talking about the performance on a GNU/Linux system.
What are the ups and downs with these different processors today? Which would you choose if you should buy a new computer, and why?
Spit it out! 
_________________
Dig where you stand! |
|
| Back to top |
|
 |
didymos
Advocate
Joined: 10 Oct 2005
Posts: 4798
Location: California
|
| Posted: Thu Aug 09, 2007 2:12 pm Post subject: |
|
|
Umm, the Intel processors are just as 64-bit as the AMD ones. Have been for awhile now. Besides which, the dual core thing isn't really new. The Core Duo line is relatively new, but Intel had dual core CPUs well before those were introduced.
_________________
Thomas S. Howard |
|
| Back to top |
|
|
Re-JeeP
Apprentice
Joined: 10 Oct 2006
Posts: 294
Location: Sweden
|
| Posted: Thu Aug 09, 2007 2:24 pm Post subject: |
|
|
| didymos wrote: | | Umm, the Intel processors are just as 64-bit as the AMD ones. Have been for awhile now. Besides which, the dual core thing isn't really new. The Core Duo line is relatively new, but Intel had dual core CPUs well before those were introduced. |
So intel doesn't make 32bit processors anymore?
Not long ago I bough a laptop with a 32bit pentium-m processor.
But isn't the new dual cores 2 x XMHz 32-bits?
_________________
Dig where you stand! |
|
| Back to top |
|
|
vivi131313
Tux's lil' helper
Joined: 11 May 2004
Posts: 108
Location: Folsom, CA
|
| Posted: Thu Aug 09, 2007 10:43 pm Post subject: 32-bit vs 64-bit |
|
|
Here's good start to answering your question: http://en.wikipedia.org/wiki/X86-64
Long story short, neither architecture is truly 64-bit. There are onyl a few truly 64-bit processors, those are the Itanium and other higher-end server processors like SPARC's etc.
These processors are actually 32-bit processors that support 64-bit memory addressing. Simply allowing you to have more system memory then the 2^32= 4 Gigabytes. Which is all a 32-bit processor is capable of addressing. Other thans that there are some other extensions to the processors, and differences between them, but at the core they still operate primarily on the 32-bit instruction set. So yes Intel does still make 32-bit processors as does AMD.
_________________
Now I make the stuff you get pissed off at. |
|
| Back to top |
|
|
Genone
Retired Dev
Joined: 14 Mar 2003
Posts: 9625
Location: beyond the rim
|
| Posted: Thu Aug 09, 2007 11:29 pm Post subject: Re: 32-bit vs 64-bit |
|
|
| vivi131313 wrote: | Here's good start to answering your question: http://en.wikipedia.org/wiki/X86-64
Long story short, neither architecture is truly 64-bit. There are onyl a few truly 64-bit processors, those are the Itanium and other higher-end server processors like SPARC's etc. |
Out of curiosity, how do you define if a processor is "truly 64-bit" ? |
|
| Back to top |
|
|
barophobia
Apprentice
Joined: 27 Apr 2004
Posts: 229
Location: somewhere
|
| Posted: Fri Aug 10, 2007 12:36 am Post subject: Re: 32-bit vs 64-bit |
|
|
| Genone wrote: | | Out of curiosity, how do you define if a processor is "truly 64-bit" ? |
Yes what is a "truly 64-bit" processor? One that will only run 64bit instructions?
All new intel and amd processors have the 64bit extensions. |
|
| Back to top |
|
|
didymos
Advocate
Joined: 10 Oct 2005
Posts: 4798
Location: California
|
| Posted: Fri Aug 10, 2007 2:05 am Post subject: Re: 32-bit vs 64-bit |
|
|
| vivi131313 wrote: | | Long story short, neither architecture is truly 64-bit. There are onyl a few truly 64-bit processors, those are the Itanium and other higher-end server processors like SPARC's etc. |
You may want to read that article a bit more closely:
| Quote: |
The primary defining characteristic of AMD64 is its support for 64-bit general purpose registers, 64-bit integer arithmetic and logical operations, and 64-bit virtual addresses. The designers took the opportunity to make other improvements as well.
|
That right there, my friend, makes it a "real" 64-bit processor. It's just that it can also use a 32-bit mode and run standard x86 code. Also, the Itanium and SPARCS have 32-bit modes as well, though not the same implementation of x86 in the Itanium's case, and for SPARCs, it's running SPARC32 code (a close analogue to the x86_64, in fact, only on a RISC architecture).
You may as well say that, e.g., the original Pentium wasn't really a 32-bit processor since it could also execute code in real mode.
[edit] OK, newer Itanium's have ditched IA-32 support in hardware, so the above only applies to earlier revisions.
_________________
Thomas S. Howard |
|
| Back to top |
|
|
vivi131313
Tux's lil' helper
Joined: 11 May 2004
Posts: 108
Location: Folsom, CA
|
| Posted: Fri Aug 10, 2007 5:00 pm Post subject: |
|
|
For some reason I didn't make the connection that the extended registers are exactly what I was looking for. So since both processors ALU's would be using these 64-bit registers then I guess they are technically 64-bit processirs. I hadn't considered them pure 64-bit because they are simply extended versions of processors that were and still fully support,without emulation, 32-bit. They still fully support 32-bit applications because they use the same logic and almost an identical instruction set, obviusly with the exception of some improved address decode logic and a few more instructions to handle full 64-bit operations. As you said with the Itanium and sparc having a 32-bit mode, which from my understanding only emulates 32-bit mode. So I guess I was looking for an overhauled architecture defining the jump to 64-bit, like the Itanium, and not just an extended 32-bit processor.
Take that as you will, I guess I haven't studied the current architectures enough, I've had IA-32 and some other less popular architectures, but x86_64 hadn't been developed when I had my computer architecture courses. And now I just do front-end work on chipsets and graphics.
_________________
Now I make the stuff you get pissed off at. |
|
| Back to top |
|
|
ShinyThings
Tux's lil' helper
Joined: 03 Jul 2007
Posts: 89
Location: Ontario
|
| Posted: Fri Aug 10, 2007 5:59 pm Post subject: |
|
|
| I was under the impression that they aren't "truly" 64-bit yet because they don't really use 64-bit addressing. They use 48-bit addressing, because 64 bits would be overkill for today's computers. |
|
| Back to top |
|
|
didymos
Advocate
Joined: 10 Oct 2005
Posts: 4798
Location: California
|
| Posted: Fri Aug 10, 2007 6:36 pm Post subject: |
|
|
Yeah, but I still think it's the native word size, register width, and instruction size that matter the most. Memory addressing can get pretty strange, esp. on x86 descended systems, so even if you don't have a full 64-bit addressing scheme, with memory management tricks and extensions, the max amount of memory supported can actually exceed that of a flat 64-bit address space. And don't forget floating point, which uses entirely different bit widths. Really, when people talk about 64-bit, they typically mean integer size.
Wait, wasn't there also some point when i686 chips started to use a 36-bit memory model?
_________________
Thomas S. Howard |
|
| Back to top |
|
|
eccerr0r
Watchman
Joined: 01 Jul 2004
Posts: 9891
Location: almost Mile High in the USA
|
| Posted: Fri Aug 10, 2007 7:16 pm Post subject: |
|
|
I think the standard definition of an "x" bit computer is how many bits of data it can compute in one instruction. Which is sort of confusing for some instruction sets that deal with more bits over multiple cycles. Hence all these wars for definitions.
I think the weirdest example is the HP48 Saturn CPU which instructions dealt with 64 bits but only handled 4 per cycle...
Then the other end of the spectrum is all these SIMD instructions...
Memory management has always been weird. There's "directly mappable" memory and "indirectly mappable" memory and I think for the most part "directly mappable" is what's considered "physical address space" (else going back into time, is EMS on 8086's really full address space? How about segmentation issues, which is what's needed to deal with >32 bits of memory on ia32 (not to mention all the 64K segments on 8086)?)
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
| Back to top |
|
|
Homer512
n00b
Joined: 04 Feb 2007
Posts: 13
|
| Posted: Fri Aug 10, 2007 9:40 pm Post subject: |
|
|
There has been an interesting discussion the gentoo-amd64 mailing list about Intel 64bit vs. AMD 64bit some time ago. Instead of linking it, I just quote the relevant parts. If you want to read the whole discussion, its name was "Is there any difference with 4 core?" You can find it on gmane.org
Duncan:
| Quote: |
As for Intel vs. AMD, with the hypertransport interconnect system AMD has, with AMD's hardware IOMMU for DMA access above the 4 gig boundary
(Intel has to emulate it with software bounce-buffers on most of its hardware, kinda nullifying the point of DMA), and with the vaunted Intel
performance lead mostly a 32-bit phenomenon, dual dual-core AMD is often more efficient on 64-bit than quad-core Intel.
|
Bob Sanders:
| Quote: |
[referring to the PentiumD Dual Core] And it's a crappy cpu. Don't waste you're money. Any AMD cpu running at half it's speed will eat it alive.
[...]
Typically, when compute power is more important to the mix, the Intel Core2 platform will perform the best. When memory and i/o bandwidth are important - database transactions, the AMD platform delivers a higher sustained bandwidth. And if the job mix is very peaky - lots of variations with long periods of lull time, the AMD platform will deliver overall power savings and require less cooling for a decrese in peak performance.
|
A bit nearer to the topic of this thread: "[gentoo-amd64] Re: 32 or 64 for web server and mysql"
Duncan:
| Quote: |
Generically, 64-bit can at times be worse, because some things (integers, memory addresses, etc) take 64-bits rather than 32, meaning bigger
binaries on disk, more memory used, more bandwidth necessary between disk and memory and memory and CPU, larger cache required for the same effect, etc. Thus, on many archs other than x86, it's quite common to have the kernel as 64-bit to allow addressing larger memory and etc, but run a 32-bit userland -- basically everything but the kernel.
On x86, it's a bit different. 32-bit x86 was always severely register constrained, among other things, and one of the improvements AMD made
with the 64-bit extensions was that the spec required more registers. As someone else already posted, this is often a big win on x86_64 as
compared to x86 (32), especially when apps are optimized to efficiently use all those extra registers. The difference the extra registers make
is generally way more than the cost associated with the larger integers, so on x86, 64-bit is generally better than 32-bit, even with the
additional expense of the larger integers and as a result binaries and etc.
There are additional considerations, however. The biggest one is whether you'll be running any closed source software. Often, that's not
available for 64-bit, or is available but with less testing and support. Of course, if it's Oracle or the like, they should support 64-bit no
problem.
On a server, most of the typical 32-bit only binary-only stuff isn't an issue, and if you ARE running any binary-only stuff, it's far more likely
to have native Linux 64-bit binaries available and well supported (xref Oracle as already mentioned). Be sure to look before you jump, however.
If you are going to be running 100% FLOSS on your server, as with the desktop, things lean rather more 64-bit.
How much memory are you going to be running? If >3 gig, you almost certainly want 64-bit if you can, as 32-bit gets rather more inefficient
at addressing >4 gig.
Also what sort of CPUs are you running? True AMD64 or Intel em64t? True AMD64 CPUs tend to be better at 64-bit than Intel, which still optimizes
for 32-bit even on their em64t stuff. If you are running true AMD64 and there's no closed-source-ware preventing it, 64-bit will almost certainly
be your better choice. If you are running em64t, you just might be better on 32-bit, depending on your exact app and load profile.
Finally... 64-bit /can/ be more secure from a hardware perspective. There's certain features built into the 64-bit extensions that improve
resistance to buffer overflows and the like, or more precisely, compiling a hardened profile, as you may be doing on a server, doesn't cause the
performance penalty on amd64 (generically, so em64t also) that it does on x86. If you are going to be using a hardened profile, I'd strongly
recommend going 64-bit for that reason.
|
|
|
| Back to top |
|
|
didymos
Advocate
Joined: 10 Oct 2005
Posts: 4798
Location: California
|
| Posted: Sat Aug 11, 2007 12:33 am Post subject: |
|
|
When did the discussion occur?
_________________
Thomas S. Howard |
|
| Back to top |
|
|
Corona688
Veteran
Joined: 10 Jan 2004
Posts: 1204
|
| Posted: Sat Aug 11, 2007 4:59 am Post subject: |
|
|
I don't think any of those answers have become obsolete since whenever they were posted.
_________________
Petition for Better 64-bit ATI Drivers - Sign Here
http://www.petitiononline.com/atipet/petition.html |
|
| Back to top |
|
|
didymos
Advocate
Joined: 10 Oct 2005
Posts: 4798
Location: California
|
| Posted: Sat Aug 11, 2007 6:04 am Post subject: |
|
|
No, I don't necessarily think so either, I just wanted to know when.
_________________
Thomas S. Howard |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Kernel & Hardware
