Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

trancepriest

macrumors 6502
Original poster
Nov 9, 2007
270
1
South Florida
Diglloyd tested Snow Leopard in both 32 bit and 64 bit kernel modes and noticed that the 64 bit kernel was faster in many photo applications such as lightroom, aperture, photoshop and nikon capture. Anyone else notice any speed differences between 32 and 64 bit kernel modes?

http://macperformanceguide.com/SnowLeopard-Performance.html#NikonCaptureNX2

Many users on the forums have stated their is no difference between the two modes. If Diglloyd is correct then there is a noticeable difference.
 
Many users on the forums have stated their is no difference between the two modes.

Thats the dark side of free internet forums. Anyone can say anything they want, even if they don't have a clue of what they're talking about.
 
If you read the post at the linked URL it deals with facts and not opinions, unless we assume the author faked the test or twisted it.

I do not agree with Apple's attitude to cut the corners on after sales service and issue no firmware upgrades. It is simply unacceptable that one cannot even buy an EFI upgrade for a three year old machine.

I do not know how often I have wasted hours and days because I have no EFI64.
 
I've ran the 64-bit kernel on my Nehalem mac pro and it was noticeably much faster and more buttery smooth.. Sadly it wont run some of my 32-bit apps that I need to use under preferences.

I cant wait for all my apps to become 64-bit.
 
It is simply unacceptable that one cannot even buy an EFI upgrade for a three year old machine.

I do not know how often I have wasted hours and days because I have no EFI64.

I wasn't aware it was common practice to upgrade motherboard BIOS or sell upgrades for BIOS or firmware from 32-bit to 64-bit. I think there's either no precedent for that so they didn't bother, or some technological issue precluding it.
 
I believe gugucom's referring to the attempts with RAID0 of a pair of SSD's on a bootable card. It's long and full of dissappointments (system ports, RR2642, before going with an EFI boot capable RAID card). L-o-n-g story. :rolleyes: ;)

It might also include issues with graphics cards, and the ability to run the 64bit Kernel in SL. Not sure if this was encountered or not, but will be if it hasn't, given some other posts on EFI32 & future usefullness. ;)

I wasn't aware it was common practice to upgrade motherboard BIOS or sell upgrades for BIOS or firmware from 32-bit to 64-bit. I think there's either no precedent for that so they didn't bother, or some technological issue precluding it.
For PC's, the firmware was upgraded to 64 bit when the chips went to 64 bit.

The MP's didn't. Technically speaking, there's no reason at all that would prevent it. Intel used the same CPU's and chipset, and installed 64 bit firmware on their own products, not just other board makers. ;)

It would seem there are other reasons. Developmental resources are one possible reason (lack of), and the lack of any concern towards after sales support is another. Oddly enough, both can easily go hand-in-hand (one justifying the other, from either direction). :eek: :p
 
I've been using the 64 bit kernel for a few days now and I haven't noticed any great differences. I've noticed micro speed differences when doing tests. If someone has seen significant speed increases of say 5% or more in an application please let me know what apps those are. Generally what I'm seeing is say on hard disk transfers the 64 bit kernel will always beat the 32 bit kernel by like a few seconds. In FCP 7 (yeah I know its 32 bit) Log & Transfer by a few seconds too. These are tests i've done over and over again to verify the speed differences. As I stated they aren't significant speed differences.
 
good topic. is there anyway to identify if programs are running as 32-but or 64-bit? or should we just assume that every product apart from apple's main ones (Finder, iLife etc) are NOT 64-bit?

i would have thought the speed increases would have been more noticeable, having the extra room for instructions and whatnot - maybe it will become more apparent with 64-bit video editing processes?
 
I wasn't aware it was common practice to upgrade motherboard BIOS or sell upgrades for BIOS or firmware from 32-bit to 64-bit. I think there's either no precedent for that so they didn't bother, or some technological issue precluding it.

Mobo firmware upgrades are common practise in the computer industry and all mobo manufacturers do them until the development on the particular socket ends. Practically any development on a 771 socket would be covered if you would use a Supermicro, Tyan or Asus mobo. Expensive 500$ dual socket products would be meticulously covered. They do this for free and there is no question that any compatible CPU, operating system and firmware development technique would be covered. All manufacturers did upgrades some years ago when 64-bit processors required a wave of new drivers. When Vista allowed you to load drivers by USB they upped a lot of 2-3 year old BIOS files to make those compatible and they do it even for processors not made by Intel if they fit the socket.

Apple in contrast will not touch anything superseded in their current sales program. They will not do it even for money.

I had issues from day one with EFI32 when I bought my MacPro1,1 and I know a lot of other people who do. The list is long as nanofrog points out. He even forgot one of the most annoying issues, which Apple quickly fixed with the EFI64 on the next generation and did not care to give to MP1,1 and MP2,1 owners. It is the failure to boot most 64-bit operating system installation disks or support 64 bit operating systems for the propriatory devices.

Try to run XP64, which is on the market since 2005 and you will realise it isn't supported. If you try Vista 64 you find that the EFI32 boot loader stalls on multiple images on one DVD or boot loading variables that are commonly used. Hence you have to single out an induvidual image and delete all load variables with a program like vLite. It took me days to sort this out.

Thorougly annoying is the failure of all Apple EFIs to deal properly with SATA AHCI drivers for other than Apple's own OS X. The user has no means to tell the firmware he will be using AHCI instead of legacy. So the OS will crash unless the registry and boot loader is manipulated before rebooting.

I will not bother with the sad story of the lack of support for graphic cards. Only the most ardent fan boy can be happy with Apple's policy there. I understand that it helps streamline development resources but surely Apple go to extremes compared to any other industry participant.

Back to the topic I would once again stress that it isn't surprising that properly supported apps will run faster on a 64bit kernel than on a 32bit kernel. It could be observed with MS OSes as well.
 
Bios upgrades are common practise in the computer industry and all mobo manufacturers do them until the development on the particular socket ends. Practically any development on a 771 socket would be covered if you would use a Supermicro, Tyan or Asus mobo. Expensive 500$ dual socket products would be meticulously covered. They do this for free and there is no question that any compatible CPU, operating system and firmware development technique would be covered. All manufacturers did upgrades some years ago when 64-bit processors required a wave of new drivers. When Vista allowed you to load drivers by USB they upped a lot of 2-3 year old BIOS files to make those compatible and they do it even for processors not made by Intel if they fit the socket.

Apple in contrast will not touch anything superseded in their current sales program. They will not do it even for money.

I had issues from day one with EFI32 when I bought my MacPro1,1 and I know a lot of other people who do. The list is long as nanofrog points out. He even forgot one of the most annoying issues, which Apple quickly fixed with the EFI64 on the next generation and did not care to give to MP1,1 and MP2,1 owners. It is the failure to boot most 64-bit operating system installation disks or support 64 bit operating systems for the propriatory devices.

Try to run XP64, which is on the market since 2005 and you will realise it isn't supported. If you try Vista 64 you find that the EFI32 boot loader stalls on multiple images on one DVD or boot loading variables that are commonly used. Hence you have to single out an induvidual image and delete all load variables with a program like vLite. It took me days to sort this out.

Thorougly annoying is the failure of all Apple EFIs to deal properly with SATA AHCI drivers for other than Apple's own OS X. The user has no means to tell the firmware he will be using AHCI instead of legacy. So the OS will crash unless the registry and boot loader is manipulated before rebooting.

I will not bother with the sad story of the lack of support for graphic cards. Only the most ardent fan boy can be happy with Apple's policy there. I understand that it helps streamline development resources but surely Apple go to extremes compared to any other industry participant.

Back to the topic I would once again stress that it isn't surprising that properly supported apps will run faster on a 64bit kernel than on a 32bit kernel. It could be observed with MS OSes as well.

It is a shame that this is the case, especially since Apple seems to be rolling in $$s of late and could easily afford to support older MPs.
 
good topic. is there anyway to identify if programs are running as 32-but or 64-bit? or should we just assume that every product apart from apple's main ones (Finder, iLife etc) are NOT 64-bit?

i would have thought the speed increases would have been more noticeable, having the extra room for instructions and whatnot - maybe it will become more apparent with 64-bit video editing processes?

Click on Apple, About this Mac, More Info, Applications... the info should be displayed.
 
Mobo firmware upgrades are common practise in the computer industry and all mobo manufacturers do them until the development on the particular socket ends. Practically any development on a 771 socket would be covered if you would use a Supermicro, Tyan or Asus mobo. Expensive 500$ dual socket products would be meticulously covered. They do this for free and there is no question that any compatible CPU, operating system and firmware development technique would be covered. All manufacturers did upgrades some years ago when 64-bit processors required a wave of new drivers. When Vista allowed you to load drivers by USB they upped a lot of 2-3 year old BIOS files to make those compatible and they do it even for processors not made by Intel if they fit the socket.
Nice post for this. :)

I had issues from day one with EFI32 when I bought my MacPro1,1 and I know a lot of other people who do. The list is long as nanofrog points out. He even forgot one of the most annoying issues, which Apple quickly fixed with the EFI64 on the next generation and did not care to give to MP1,1 and MP2,1 owners. It is the failure to boot most 64-bit operating system installation disks or support 64 bit operating systems for the propriatory devices.
Oops. :eek: :D :p
 
I've been using the 64 bit kernel for a few days now and I haven't noticed any great differences. I've noticed micro speed differences when doing tests. If someone has seen significant speed increases of say 5% or more in an application please let me know what apps those are.
EM64T has an extended register set so isn't nearly as register starved as IA32. This in turn allows compilers to produce better optimized code. So it's not surprising that merely recompiling yields a 10% improvement.
 
From my experience, 64bit is faster. My Mac Pro loads to desktop basically instantly, whereas under 32bit it takes a while and loads piece-by-piece (ie: Menu bar, Dock, Icons). As for application performance...haven't noticed much of a change really, but I'm pretty sure the whole thing is "snappier".
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.