Filed under: Macbook Pro, MacBook, Leopard, Snow Leopard
Will Snow Leopard really make my computer any faster?
We've seen the benchmarks. We've heard from the techno-geeks. According to Apple, Snow Leopard should result in some impressive speed gains, and hefty hard drive space recapture. But does this speed bump actually result in tangible benefits for the average user? Do you really get back a functional amount of hard drive space? I undertook an intentionally low-tech approach to find out, looking at the space on the drive, and using my iPhone's stopwatch function to time various functions before and after upgrade. I took measurements on two computers: a low-end, bare bones white MacBook used lightly as a secondary computer, and a higher-end MacBook Pro used heavily as a primary computer. The white MacBook was generally speedy and efficient before the upgrade, due to the fact that it had very little installed on it. However, the MacBook Pro was bloated and slow due to lots of programs, with problems magnified by years of hard drive image flashes over various computer upgrades, typical of the non-technical business user.All start up times are true start up times. In other words, I didn't deem the computer to have "started up" until I had full, no-lag control of a fully-propagated desktop. Same went for the programs whose start up times I tested -- none were deemed started up until the program was responding to input and usable. I picked some common programs that I felt reflected typical use. I turned on auto-login to the primary account on both computers to gain timing accuracy. Not all the results are comparable between computers; certain programs were on one computer but not the other. But, in general, the task was instructive.
2 GHz Intel Core 2 Duo MacBook (white), with 2 GB 667 MHz DDR2 SDRAM
Leopard:
Start up chime to true, usable desktop: 35 seconds
Firefox start up: 11 seconds
Safari start up: 4 seconds
iPhoto start up to edit window on 1 GB photo library: 11 seconds
Microsoft Word 2004 start up to blinking cursor: 25 seconds
Snow Leopard:
Start up chime to true, usable desktop: 32 seconds
Firefox start up: 9.5 seconds
Safari start up: 1.5 seconds
iPhoto start up to edit window on 1 GB photo library: 9 seconds
Microsoft Word 2004 start up to blinking cursor: 26 seconds (after Rosetta install)
Upgrade time: 44 minutes
Hard drive recapture: 15 GB (see update below)
2.2 GHz Intel Core 2 Duo MacBook Pro with 2 GB 667 MHz DDR2 SDRAM
Leopard:
Start up chime to true, usable desktop: 153 seconds (ouch!)
Firefox start up to loaded homepage: 13 seconds
Safari start up to loaded homepage: 4 seconds
iPhoto start up to edit window on 9.45 GB photo library: 24 seconds
Microsoft Word 2008 start up to blinking cursor: 29 seconds
Adobe Acrobat 6.0 Standard start up: 120 seconds
Snow Leopard:
Start up chime to true, usable desktop: 57 seconds (ooh!)
Firefox start up to loaded homepage: 10 seconds
Safari start up to loaded homepage: 3 seconds
iPhoto start up to edit window: 12 seconds
Microsoft Word 2008 start up to blinking cursor: 25 seconds
Adobe Acrobat 6.0 Standard start up: 57 seconds (after Rosetta install)
Upgrade time: 66 minutes
Hard drive recapture: 10.5 GB (see update below)
My overall impression? The speed-improvement claims are fairly well-substantiated. Both computers are operating speedier than before, with the greatest overall speed gains in the power-user computer. The bare bones machine, while faster, is not really that much practically faster. For this computer, the 15 GB of extra space was the most dramatic improvement. As for the MacBook Pro, it achieved some stunning start up time improvements: start up time is now consistently forty percent of what it used to be. Even waking up from sleep (an event I did not time) both computers are now noticeably faster, locking on to a previously-used Wi-Fi network almost immediately.
In all, both computers were at least somewhat functionally enhanced from the upgrade, and I am glad to see an operating system upgrade that truly gave the user some real speed and space benefits rather than concentrating strictly on adding features.
UPDATE: As our commenters have noted (and Megan posted about last week) part of my space savings is due to the reworking of how Snow Leopard measures gigabytes.
UPDATE 2: Fellow blogger Joachim Bean has some of his own speed tests to share:
2 GHz Intel Core 2 Duo Mac mini with 1GB 1066 MHz DDR3 SDRAM (GeForce 9400M)
Leopard
Startup chime to true, usable desktop: 57 seconds
Shutdown: 8 seconds
Safari startup: 7 seconds
Microsoft Word 2008 startup: 17 seconds
Mail startup: 11 seconds
Xcode (3.1.4): 10 seconds
iPhone Simulator startup: 20 seconds
Snow Leopard:
Startup chime to true, usable desktop: 49 seconds
Shutdown: 5 seconds
Safari startup: 2 seconds
Microsoft Word 2008 startup: 17 seconds
Mail startup: 7 seconds
Xcode (3.2): 16 seconds
iPhone Simulator startup: 20 seconds
Upgrade time: 1 hour, 20 minutes
Hard drive recapture: 12GB (see update)
![TUAW [Cafepress]](http://www.blogsmithmedia.com/www.tuaw.com/media/tuaw-cafepress-promo.png)
Reader Comments (Page 1 of 4)
Chroma said 9:37AM on 9-04-2009
"using my iPhone's stopwatch function to time various functions before and after upgrade"
Uhh, thats were I stopped reading.
Reply
Ray Duran said 9:42AM on 9-04-2009
Because obviously the iPhone's timer taints the experiment? Or shows a *gasp* bias of the author? Please.
Devon said 9:49AM on 9-04-2009
Why? The iPhone has easily enough resolution for this test. Actually it has timer resolution better than 50ms. Yes I have written a program on the iPhone that works as a timer and it's accurate.
Matt said 9:52AM on 9-04-2009
Chroma, give me a break! I personally don't have a stopwatch laying around, so if I ever think of using a stop watch, I use my iPod or iPhone.
Great article showing the practical improvements in upgrading to Snow Leopard!
Chroma said 10:00AM on 9-04-2009
No, using an iPhone to mesasure software speeds is akin to using a 12" ruler to measure a wall in an building. Right idea, but bad excution.
For the record, I've have an iPhone since launch month.
ProfessorDex said 10:15AM on 9-04-2009
@ Chroma
"I undertook an intentionally low-tech approach to find out"
Perhaps you missed this part? This was by no means a scientific experiment, it was real world testing using everyday tools and situations.
"Right idea, but bad excution."
Right idea = People making posts like this and allowing others to comment.
Bad Execution = Allowing you to make useless comments.
julian said 12:57PM on 9-04-2009
I am actually with chroma.
everyone is entitled to their opinion.
and well, although the iphone has a stopwatch it doesnt mean it is one.
Posting stuff on a blog that is read by thousands might ask for a more "scientific" approach.
then again i liked the article.
Szrog said 1:00PM on 9-04-2009
Chroma has a point, even if he delivered it with a bit of snark.
The point of the article is to measure real world use as opposed to the benchmarks that have been measured ad nauseum. In order to find value in the article, we have to accept the assumption that the benchmarks don't represent our best efforts to accurately measure real world use.
The choice of _what_ to measure makes a lot of sense. The author measured things that people experience every day.
The problem Chroma brought up was the flaw in _how_ it was measured. This is not the "scientific" vs. "real world" argument, ProfessorDex. It is an "accurate" vs. "inaccurate" argument.
Chroma said 1:07PM on 9-04-2009
Snark is all i gots!
Lauren Hirsch said 1:08PM on 9-04-2009
The problem is that the differences I was looking to see were ones that were well outside the margin of error of a standard stopwatch. I was looking for practical, tangible differences--the kind easily measured by a stopwatch. If I need more accuracy than, say, a whole second, then I'm achieving useless precision because I'm quite confident nobody is PRACTICALLY impacted if a program takes 9.7 or 9.9 seconds to boot up. That certainly isn't going to create any kind of usability difference. So, given that was my aim, a standard stopwatch was certainly an appropriate tool. We've posted other articles on the very precise benchmarks--this post was meant to come at it from the entirely opposite direction: nothing more precise than could be felt by the end user.
iGO said 1:31PM on 9-04-2009
Lauren,
You did good.
I clearly understand the gist of your "blog"
If I did not want to read a "blog", I would not come here,
to T-he U-nnoficial A-pple W-eblog.
I'd go to something like Arstechnica or such other, and seek out a White Paper on "Measurements and Standards", and how it was applied to Snow Leopard.
People are just !@#$%-Up !
Carry on
ProfessorDex said 1:49PM on 9-04-2009
Lauren, I agreeyou did a great job!! Ignore the whiners and complainers who have nothing better to do than moan about your testing methodology (and/or the tools you used). Until they do their own testing and present it as you did, their opinions are moot!
As for Chroma, if "Snarks is all you gots", then perhaps you should work on fixing that.
Szrog said 2:29PM on 9-04-2009
@Lauren
The problem with using the stopwatch isn't the reduced precision of the stopwatch, but the unnecessary inclusion of human error into the measurements. I say unnecessary because you can always round any unwanted precision away. To use Chroma's wall-measuring analogy: You might not need the accuracy of a tape measure, but why use a ruler when the tape measure is right there?
But this discussion has grown banal enough already. I'm willing to cede the point that a stopwatch was close enough for your purposes. You've written a fine article about your experiences which I found interesting. Thank you very much.
@ProfessorDex
As long as you brought up the subject of what people should "work on fixing" perhaps you can try to address the semantic content of posts instead of name-calling and pretending the opinions of those who disagree with you don't matter.
ProfessorDex said 3:16PM on 9-04-2009
Clearly you've completely misinterpreted my posts.
I had no intent of "pretending the opinions of those who disagree with you don't matter". I'll try be a little more clear for you to understand.
The opinions of those who disagree with me don't matter.
Thank you. That is all.
sorrel said 12:12AM on 9-05-2009
do you actually think that tenths or hundredths of a second actually matters?...and did you not read his admittedly low tech approach comment...he is going for user experience here not laboratory quality that is only of value to some low percentage of the using population...
rd said 11:53AM on 9-23-2009
You spelt "where" wrong, that's where I stopped reading.
TmiY said 9:44AM on 9-04-2009
you guys have anything with Snow Leopard skin?
Reply
mentalsticks said 9:50AM on 9-04-2009
@mykshop: sounds great! I've always wanted Jordan Gucci shoes! How can I reach this "website" you are talking about?
Reply
Cycomachead said 9:50AM on 9-04-2009
You didn't get 15GB of file space. And everyone one of your files takes up more room on your HD. Also notice how this size of the actual drive grew???
STOP REPORTING DRIVE SIZES LIKE THIS! PLEASE. Apple is not lying when you free up disk space-depending of if you remove things during install it could be 6GB (and part of that is it clearing the caches and etc.) Unless you do the math or have utilities to measure then 15GB more space is inaccurate. How big is your HD 10.6? The math is so different that you really have to list every detail. 10.5 1GB = 2^30 bytes. 10.6 1GB=10^9 bytes.
Reply
Andy Smith said 10:09AM on 9-04-2009
I'd like to agree with the technicality of the above comment, the article doesn't state what proportion of the hard drive space saved is just a function of the changes mathematics.
However, I don't agree with the above commentators rudeness and arrogance.