Technical Discussion
  >> Linux Issues


Register (or login) on our website and you will not see this ad.


Pages in this thread: 1 | [2] | 3 | 4 | 5 | >> (show all)   Print Thread
Standard User AEP
(knowledge is power) Fri 01-Apr-11 16:17:43
Print Post

Re: 32- and 64-Bit


[re: awoodland] [link to this post]
 
SSE2 isn't exclusive to 64-bit processors, not does it require the processor to be in 64-bit mode so I can't count it as an advantage of 64-bit OSs. It is true that when in 64-bit mode there are 16 SSE registers available, as opposed to the 8 in 32-bit mode - I covered this under the registers part of my OP.
Standard User awoodland
(regular) Fri 01-Apr-11 16:21:54
Print Post

Re: 32- and 64-Bit


[re: AEP] [link to this post]
 
There are no 64-bit processors without SSE though, which means no need for extra branch statements (and hence branch prediction failure) to pick if (SSE2) { ... } else { ... } code paths, which definitely is a feature unique to x64.
Standard User AEP
(knowledge is power) Fri 01-Apr-11 16:30:29
Print Post

Re: 32- and 64-Bit


[re: awoodland] [link to this post]
 
There are precious few processors nowadays without 64-bit instructions. But the discussion that I was raising was whether those 64-bit instructions were used or not. (I.e whether to use 32-bit or 64-bit programs and OSs.) In that context, SSE2 is not exclusive to 64-bit, so can't really be listed as an advantage of 64-bit programs.


Register (or login) on our website and you will not see this ad.

Standard User camieabz
(legend) Fri 01-Apr-11 16:36:02
Print Post

Re: 32- and 64-Bit


[re: AEP] [link to this post]
 
http://www.phoronix.com/scan.php?page=article&item=u...

Interesting comparison of 32-bit, 32-bit PAE and 64-bit.

~~~~~~~~~~



© Camieabz 2002-2011 - All rights and lefts reserved.

report this link
Standard User 12eason
(eat-sleep-adslguide) Fri 01-Apr-11 16:47:57
Print Post

Re: 32- and 64-Bit


[re: AEP] [link to this post]
 
I think the main problem is the huge amount of extra HD space you need, since the HD is the biggest bottleneck. Real world use of 64bit is pretty stupid unless you have SSDs. The artificial tests that are often done ignore the HD usage down sides.
Standard User AEP
(knowledge is power) Fri 01-Apr-11 17:36:17
Print Post

Re: 32- and 64-Bit


[re: camieabz] [link to this post]
 
That's very interesting, and certainly demonstrates my belief that 64-bit is not just about more memory (although PAE clearly is). Most of the results show the performance increase that I would expect, but I am amazed by the Apache and Disk Transaction benchmarks (which are no doubt connected). If you want to run a web server I think you can see which way to go!
Standard User AEP
(knowledge is power) Fri 01-Apr-11 17:39:17
Print Post

Re: 32- and 64-Bit


[re: 12eason] [link to this post]
 
I'd suggest you have a look at the benchmarks that Cammie linked to, one of which specifically demonstrates disk transactions. I'm not quite sure what you mean by "huge amount of extra HD space" - you don't need any more HD space with 64-bit programs than 32-bit ones. Why on earth would you?

Ask web serving companies which OS they use and whether they use SSDs.
Standard User 12eason
(eat-sleep-adslguide) Fri 01-Apr-11 17:46:03
Print Post

Re: 32- and 64-Bit


[re: AEP] [link to this post]
 
Obviously because they are twice the size. Web serving companies all use dedicated storage boxes with raid.
Standard User AEP
(knowledge is power) Fri 01-Apr-11 18:35:23
Print Post

Re: 32- and 64-Bit


[re: 12eason] [link to this post]
 
I take it you are not very familiar with the architecture and programming of X86_64 processors. That is the only explanation I can find for that unbelievably ignorant statement.

Stick to topics you know something about.
Standard User 12eason
(eat-sleep-adslguide) Fri 01-Apr-11 19:12:33
Print Post

Re: 32- and 64-Bit


[re: AEP] [link to this post]
 
Dude, you only have to check the disk space. It's easy to prove.
Pages in this thread: 1 | [2] | 3 | 4 | 5 | >> (show all)   Print Thread

Jump to