Safari is definitely threaded for multiple CPUs like nearly all other apps. I just managed to push Safari to 145% CPU playing four flash animations and two large QuickTime movies in separate windows. The ECMAScript engine, however, appears to not be threaded.
Why are the PC's beating the pants of our Macs. the difference cannot be explained by pure MHZ?
A profound question. There's no doubt IE on Windows is much more optimized by using some of MS undocumented kernal internals. IMHO, IE was always faster than NS on Windows. Has anyone else tested this script with Netscape/Mozilla on the same Windows machine they tested IE? The closest I've seen is mark_wilkins "7673 in Mozilla 1.2 on a 2.8 GHz Xeon (dual) running Red Hat 7.2.". This compares very well with the numbers I got on my Dual 1.25 with Safari.
This is a singularly stupid benchmark, in that the main bottleneck is the implementation of the code for printing all those numbers, which is very optimization-sensitive. It's unclear how any of these results correspond to real-world performance of anything, although obviously faster machines tend to run faster.
The IE/Windows superiority in these tests are probably as much a matter of a single clever optimization that someone added along the way as anything else. However, I can't believe that it's a great indicator for how other tasks will do.
(P.S. Woohoo! my P4 is still the winner so far! Beat THAT!)
A pessimistic number, because I (again) I didn't bother quitting my work apps:
10406 in Moz 1.5a on a P4 1.7GHz running Windows XP.
mark_wilkins: I agree that it's a kind of silly benchmark. The most interesting thing it's pointed out to me sofar is that OmniWeb is the only browser I've run it on that displayed the code and commentary on the page before running the script. Safari and Mozilla just give me a white page and a busy cursor.
This is a singularly stupid benchmark, in that the main bottleneck is the implementation of the code for printing all those numbers, which is very optimization-sensitive. It's unclear how any of these results correspond to real-world performance of anything, although obviously faster machines tend to run faster.
The IE/Windows superiority in these tests are probably as much a matter of a single clever optimization that someone added along the way as anything else. However, I can't believe that it's a great indicator for how other tasks will do.
-- Mark
We have to be careful not to dismiss every benchmark that doesn't produce the results we want as being unrepresentative. What we have here is an example of our machines loosing out to pc's on some very simple code.
Who's "we?" I've posted the lowest number so far and I still think it's a stupid benchmark.
-- Mark
Mark et al,
Although this is a purely silly benchmark (should we even call it a benchmark at all?), one thing is clear and that is IE is better optimized for windows than any other browser on the PC or Mac. And this does apply to more than just text from this script. There's certainly more that can be done to optimize Safari but it won't be an easy job when we are dealling with open source. Anyone correct me if I wrong here.
There's certainly more that can be done to optimize Safari but it won't be an easy job when we are dealling with open source.
Software optimization is software optimization. Nothing prevents it from flying along. Certainly, nothing at all prevents Apple from taking OS code and optimizing it specifically for their platform.
firebird was faster but it was hell to scroll to the bottom, the scroll bar actually disappeared, and the scroll column broke heh While i could drag the bar right through the whole thing smoothly in safari.
edit:
second attempt: firebird didn't break and it got 26368
Comments
Originally posted by PB
Why? Safari uses only one processor
Not true.
Safari is definitely threaded for multiple CPUs like nearly all other apps. I just managed to push Safari to 145% CPU playing four flash animations and two large QuickTime movies in separate windows. The ECMAScript engine, however, appears to not be threaded.
Originally posted by Addison
Why are the PC's beating the pants of our Macs. the difference cannot be explained by pure MHZ?
A profound question. There's no doubt IE on Windows is much more optimized by using some of MS undocumented kernal internals. IMHO, IE was always faster than NS on Windows. Has anyone else tested this script with Netscape/Mozilla on the same Windows machine they tested IE? The closest I've seen is mark_wilkins "7673 in Mozilla 1.2 on a 2.8 GHz Xeon (dual) running Red Hat 7.2.". This compares very well with the numbers I got on my Dual 1.25 with Safari.
The IE/Windows superiority in these tests are probably as much a matter of a single clever optimization that someone added along the way as anything else. However, I can't believe that it's a great indicator for how other tasks will do.
(P.S. Woohoo! my P4 is still the winner so far! Beat THAT!)
-- Mark
10406 in Moz 1.5a on a P4 1.7GHz running Windows XP.
mark_wilkins: I agree that it's a kind of silly benchmark. The most interesting thing it's pointed out to me sofar is that OmniWeb is the only browser I've run it on that displayed the code and commentary on the page before running the script. Safari and Mozilla just give me a white page and a busy cursor.
Score one for the Omni guys.
Originally posted by mark_wilkins
This is a singularly stupid benchmark, in that the main bottleneck is the implementation of the code for printing all those numbers, which is very optimization-sensitive. It's unclear how any of these results correspond to real-world performance of anything, although obviously faster machines tend to run faster.
The IE/Windows superiority in these tests are probably as much a matter of a single clever optimization that someone added along the way as anything else. However, I can't believe that it's a great indicator for how other tasks will do.
-- Mark
We have to be careful not to dismiss every benchmark that doesn't produce the results we want as being unrepresentative. What we have here is an example of our machines loosing out to pc's on some very simple code.
Originally posted by Amorph
OmniWeb is the only browser I've run it on that displayed the code and commentary on the page before running the script.
IE 6 does that too... you still want to "score one" for OmniWeb??
-- Mark
Originally posted by Addison
We have to be careful not to dismiss every benchmark that doesn't produce the results we want as being unrepresentative.
Who's "we?" I've posted the lowest number so far and I still think it's a stupid benchmark.
-- Mark
Originally posted by mark_wilkins
IE 6 does that too... you still want to "score one" for OmniWeb??
I don't count IE.
21758 : Netscape 7.0
10064 : Safari 1.0
8112 : Camino 0.7
OW 4.2 crapped out on me twice.
Safari 1.0: 10559
Originally posted by mark_wilkins
Who's "we?" I've posted the lowest number so far and I still think it's a stupid benchmark.
-- Mark
Mark et al,
Although this is a purely silly benchmark (should we even call it a benchmark at all?), one thing is clear and that is IE is better optimized for windows than any other browser on the PC or Mac. And this does apply to more than just text from this script. There's certainly more that can be done to optimize Safari but it won't be an easy job when we are dealling with open source. Anyone correct me if I wrong here.
Originally posted by DVD_Junkie
There's certainly more that can be done to optimize Safari but it won't be an easy job when we are dealling with open source.
Software optimization is software optimization. Nothing prevents it from flying along. Certainly, nothing at all prevents Apple from taking OS code and optimizing it specifically for their platform.
Originally posted by DVD_Junkie
And this does apply to more than just text from this script.
Are you saying this benchmark is indicative of general slowness issues? On the basis of what do you say this?
-- Mark
pb 550
10.2.6
had some programs running in the backround,
safari 1.0(85) 31157
Firebird 0.6.1 26807
firebird was faster but it was hell to scroll to the bottom, the scroll bar actually disappeared, and the scroll column broke heh
edit:
second attempt: firebird didn't break and it got 26368
Powerbook G4 1000Mhz 1024MB RAM OS X 10.2.6
IE 5.22 - 55537
Netscape 7.1 - 10584
Safari 1.0 - 10204
Now same powerbook running Windows 2000 on VPC 6.0 (320MB PC RAM)
IE 6.0 - 14430
Netscape 7.1 - 24956
Hmmmm... says something about IE 5 Mac and IE 6 Windows
Also,
Powermac G4/500 OS 9.22
IE 5.16 - 76943
Netscape 7.02 - 12510
Compaq Pentium 3 350mhz
IE 6.0 - 17740
Netscape 7.1 - 37700
No other applications running during all tests on all computers.
Originally posted by bayside
Here are some interesting results I got:
That's pathetic!
(the IE 5.2.2 on the Mac number I mean.)
-- Mark