Snow Leopard - Your Benchmarks vs Leopard

[t0mat0]

CNET:
iTunes encoding - 10.5 just better
Photoshop - 10.6 just better
QT multimediate multitasking test - 10.5 much quicker
Cinebench - 10.5 just better

CNET results -

MacWorld.com - Looks at 10.6 performance here


10.6 was Faster:

• 10-15 minutes faster in Initial Time Machine backup to an external FW800 Average 32% faster with Time Machine backups across the 3 systems (remember it's a smaller footprint too)
• Faster during shutdown (3 seconds vs 6 seconds - not insignificant for laptop users)
• Just faster on a compressor test, (encoding a .mov file using Apple’s H.264 for Video Podcasting presets) - expect OpenCL to boost this Dramatically.
• Zipping a 2GB folder in the Finder was 6-11% faster
• Importing 150 photos from the HD into iPhoto was 8-21% faster
• Quicker to scroll through a large pdf in Preview mode
• Safari was benchmarked "more zippy" - JavaScript performance was average of 33% better (using Sunspider benchmark results)

Snow Leopard = speed to Leopard:

• Start up times
• Photoshop test suite times

Snow Leopard was slower than Leopard:

• Finder doing a duplication of a 1GB file (" took 2-3seconds under 10.6
• A wake from sleep & to the network test that took 1-3 seconds longer

The other results show as previously guessed - The Benefits of Snow Leopard (speedwise) is dependent on the machine

• Unzipping a compressed file in the Finder and importing a camera archive into iMovie both slower under 10.6 on iMac, but faster on MBP & Mac Pro
• Exporting a project from iMovie using that application’s iTunes: 10.6 faster on iMac, slower on MBP = on the Mac Pro
• Converting 2 CDs of AAC-encoded song files to 256Kbps MP3s in iTunes - quicker with 10/6 on iMac & MBP, but longer with Mac Pro

Their chart shows "the percentage speed boosts (above 100%) or speed reductions (below 100%) we found when upgrading a system from Leopard to Snow Leopard. A score of 100% indicates that the test took the same time on both systems."

Apple will, no doubt, continue to refine and improve performance of the OS in the weeks and months to come. And applications, including Apple’s own, will be optimized to take advantage of Snow Leopard’s processor technologies like Grand Central Dispatch and Open CL.

MacWorld lab benchmarks:
Chart shows % speed boosts (above 100%) or speed reductions (below 100%) found when upgrading a system from Leopard to Snow Leopard. A score of 100% indicates that the test took the same time on both systems.

 

[MikhailT]

The benchmarks don't honestly tell the whole story. It's all about the user's perspective. GUI can be much more responsive, cleaner and faster from a user's perspective even if the numbers shows no difference at all.

It's like comparing Vista to W7, they both shows same performance but from a user perspective W7 is much lighter, responsive and just clean overall. That's what SL is suppose to be, lighter, not faster overall.

 

[t0mat0]

I'd agree - it's going to be easy and hard to go on both sides for this, and this is just with the limited range of hardware Apple has over the last few years. Imagine how Microsoft is going to swing Windows 7 benchmarks over the broad range...

First hand having a go on a Snow Leopard machine will help, or just try it and see

 

[TheSpaz]

The benchmarks don't honestly tell the whole story. It's all about the user's perspective. GUI can be much more responsive, cleaner and faster from a user's perspective even if the numbers shows no difference at all.

It's like comparing Vista to W7, they both shows same performance but from a user perspective W7 is much lighter, responsive and just clean overall. That's what SL is suppose to be, lighter, not faster overall.

No, I'm pretty sure Apple is advertising faster speeds... not just lighter.

 

[MikhailT]

No, I'm pretty sure Apple is advertising faster speeds... not just lighter.

The thread isn't about what Apple did say or not. Marketing isn't reality, we're talking in reality here.

 

[rwilliams]

10.6 was Faster:

• 10-15 minutes faster in Initial Time Machine backup to an external FW800 Average 32% faster with Time Machine backups across the 3 systems (remember it's a smaller footprint too)
• Faster during shutdown (3 seconds vs 6 seconds - not insignificant for laptop users)
• Just faster on a compressor test, (encoding a .mov file using Apple’s H.264 for Video Podcasting presets) - expect OpenCL to boost this Dramatically.
• Zipping a 2GB folder in the Finder was 6-11% faster
• Importing 150 photos from the HD into iPhoto was 8-21% faster
• Quicker to scroll through a large pdf in Preview mode
• Safari was benchmarked "more zippy" - JavaScript performance was average of 33% better (using Sunspider benchmark results)

I don't know - maybe it's just me, but my life isn't so hectic that a 3 second improvement on shutdown time, or scrolling through a PDF a little faster, or shaving 10 seconds off of a photo import is going to make much of a difference to me. I appreciate the improvements, but I just can't champion Snow Leopard like so many people here do. The changes will be fine and will be good for Mac users in the coming years, but it just feels like so many people are trying to find anything to boast about when it comes to Snow Leopard, and it all feels very forced.

 

[TheSpaz]

The thread isn't about what Apple did say or not. Marketing isn't reality, we're talking in reality here.

You think 64-bit and thinner apps aren't going to be faster? It's not marketing. Snow Leopard *IS* faster than Leopard all around.

 

[TheSpaz]

I don't know - maybe it's just me, but my life isn't so hectic that a 3 second improvement on shutdown time, or scrolling through a PDF a little faster, or shaving 10 seconds off of a photo import is going to make much of a difference to me. I appreciate the improvements, but I just can't champion Snow Leopard like so many people here do. The changes will be fine and will be good for Mac users in the coming years, but it just feels like so many people are trying to find anything to boast about when it comes to Snow Leopard, and it all feels very forced.

I heard that long operations won't see a great improvement, but the OS as a WHOLE is more responsive. Apps launch faster, sites load faster and the Finder is a nice breath of fresh air. The UI is more responsive too.

Trust me, you'll most likely notice a difference. It will feel as if your computer got a speed boost (like buying a newer computer).

 

[Richard1028]

The benchmarks don't honestly tell the whole story. It's all about the user's perspective. GUI can be much more responsive, cleaner and faster from a user's perspective even if the numbers shows no difference at all.

This is the dumbest thing I've heard here in a while. Benchmarks are "reality". They have nothing to do with a user's "perspective". If some idiot wants to think his processes are running faster because it "feels" that way when they really aren't well then... he's an idiot with his head in the sand (or a fanboy).

That's what SL is suppose to be, lighter, not faster overall.

Then Apple needs to find another word for, "snappier".

 

[dukebound85]

You think 64-bit and thinner apps aren't going to be faster? It's not marketing. Snow Leopard *IS* faster than Leopard all around.

did you even read the results?

 

[MikhailT]

You think 64-bit and thinner apps aren't going to be faster? It's not marketing. Snow Leopard *IS* faster than Leopard all around.

I never said it isn't faster or slower. All I said was that benchmarks doesn't tell the whole story. I also said that whatever Apple says about the speed, that's just marketing, they won't post benchmarks showing slower or same speed.

The results shows that it isn't faster all around, it's faster in some area but not all.

Thinner apps? It has nothing to do with speed, thinner apps are just space on the disk, it won't make the app faster, the OS will load the right data. A few millisecond of initial loading to determine which arch isn't noticeable by anybody.

64bit Applications isn't exclusive to Snow Leopard, it may be a couple of millisecond faster due to the pure 64bit kernel not needing to convert messages between 32bit kernel and the CPU but that means in 64 bit kernel, the 32 bit applications will have some milliseconds latency in order to work in the 32bit submode of the CPU. Regardless, they aren't noticeable by users.

This is the dumbest thing I've heard here in a while. Benchmarks are "reality". They have nothing to do with a user's "perspective". If some idiot wants to think his processes are running faster because it "feels" that way when they really aren't well then... he's an idiot with his head in the sand (or a fanboy).

You obliviously have no idea what I was talking about. Read my post again from the beginning. The benchmarks doesn't tell the whole story, they never do, they are just as bad as statistics, they tell you a number, not conclusions.

User's perspective or experience with the product is far more important than some numbers ever will, that's a fact. If a user doesn't like UAC prompting him every 5 seconds, it does not matter to him that Vista can run Word 10X faster than XP. MS, Apple and all companies in the world spends millions if not billions on user focus groups to test the experience of their OS. Even Mozilla just marked their FF.next (FF4) project focusing on the user's experience and perspective of speed of Firefox. It is not about how fast Firefox runs, but how fast it shows up and render to the user.

Think about Finder for a second. If Finder's copy operations are exactly the same in both Leopard and Snow Leopard, almost same speed, for some people, they'll say that it is not any faster or Apple did not do anything with Finder in Snow Leopard. Now suppose Finder's ability to keep working while network drives disconnect randomly in Snow Leopard and the lack of responsiveness of Finder in Leopard, that's a user perspective thing, benchmarks will never tell you that Finder sucks with network shares on Leopard compared with Snow Leopard.

 

[t0mat0]

This is the dumbest thing I've heard here in a while. Benchmarks are "reality". They have nothing to do with a user's "perspective". If some idiot wants to think his processes are running faster because it "feels" that way when they really aren't well then... he's an idiot with his head in the sand (or a fanboy).

Then Apple needs to find another word for, "snappier".

Boomtasticly quick? The perspective does count. If the users using programs and parts of the OS that are faster than Leopard, then they'll feel that. We're talking about processes feeling faster, because they are, right?

 

[jgbhardy]

Benchmark programs unreliable?

I don't think we can rely on these benchmark programs, have they been programmed to take into consideration the way that snow leopard handles processes in a different way through Grand Central. So it may seem like its slower in the benchmark because its still using old Leopard methods to check speeds. To the end user we get to see and feel the effect of the multi-core processing etc. Could this explain some of the drops in performance?