10.2.8 introduced a bug in which fix_prebinding or redo_prebinding (I forget which now) would run before any application launch - including background processes. This is why looking up a webpage, or if you misspelled a word, or any number of other basic tasks would give a 30 second pause.
Strange. I never saw that on any of our 10.2.8 machines, and I definitely would have noticed it. The only vaguely similar problem I've seen is that Mail will occasionally block the entire OS while it's "writing changes to disk", but it did that before and after 10.2.8. Haven't seen it in Panther yet, so maybe it's fixed.
I have seen other speed reports and they have said that panther speeds up G4s a lot but the G5s, not much. Your tests, though limited, seem to confirm that. I thought it was curious.
Maybe they put G4s in the iBooks because panther is optimized for the G4 and not so much yet for the G5s.
Strange. I never saw that on any of our 10.2.8 machines, and I definitely would have noticed it. The only vaguely similar problem I've seen is that Mail will occasionally block the entire OS while it's "writing changes to disk", but it did that before and after 10.2.8. Haven't seen it in Panther yet, so maybe it's fixed.
Fair enough 3.1416 but it does exist for some users.
It's hard to track down when it was introduced, I merely noticed it after 10.2.8.
Anyway it seems in Jaguar there was a change in how/when various pre-binding utilities ran as opposed to older version. I can't speak to the details since the problem was so poorly documented and I just said F-it and just worked through the beachballs.
I don't presume everyone had the problem, don't presume no one had it.
If you saw a beachball or waited 30 seconds for something to happen and then suddenly several queued actions all suddenly are allowed to run, then you probably had this pre-binding problem and didn't know it.
If you saw a beachball or waited 30 seconds for something to happen and then suddenly several queued actions all suddenly are allowed to run, then you probably had this pre-binding problem and didn't know it.
It's gone for now, I'm happy. Yay Panther.
You may also have had NetInfo problems. I know for a fact that can cause system-wide pauses as the NI system times out on various requests... and it acts just like that.
Changed one little thing on a G3 server a few months ago in the NI setup and suddenly the speed was *much* nicer.
I know for a fact that top -u would show fix_prebinding at a huge CPU% for about 30 seconds.
That DID always seem like a timeout issue (annoyingly exactly 30 seconds!), so maybe NetInfo was involved too, thanks.
I do think it is a particular issue that will exist on non-upgraded systems for a long time from now, so it's a shame it was never really fixed within the Jaguar updates. I doubt it will be now.
Feel sorry for Jag users that have that problem and who will assume that Panther will also "suck" in that same manner. Past experiences can and will effect future purchases and decisions, so the fact that I don't have the problem in Panther isn't reassuring, knowing so many people will either never upgrade, or wait a few years to do so.
I'm sure Apple will hasten the Jaguar-to-Panther migration by introducing new versions of QT, iApps, etc., that no longer work in 10.2 -- as with 10.1 and 10.0. If this is really a modern structured OS, shouldn't software and firmware work in all releases (10.0,10.1,10.2,10.3) from this century? When was the last time you looked at any software or firmware that said simply Mac OSX compatible -- 10.0 or greater?
Yeah, that's right. They break stuff on purpose so you have to upgrade. It has nothing to do with the fact that the frameworks changing or new frameworks being introduced AT ALL.
You want them to write apps that work on systems lacking new features? Wow, that'd be a neat trick... :P
Look at it this way, they write new systems that allow your old apps to continue working. *THAT* is compatability, and about as good as you can hope for.
The common thing in all those tests is that they have a lot of number crunching activity and almost no GUI activity. As you can see, in all cases the difference between Jaguar and Panther is minimal, except in the third test where Jaguar was faster than Panther by almost 6%.
You cannot expect something else since the two OSs have been produced with compilers not that different, and share about the same optimisations at this level. When and if Apple moves to a better compiler (e.g. the IBM one), then you can expect improvements in processor performance.
I could care less how fast Panther crunches numbers as opposed to Jaguar. My beef with Jaguar on my 1.8 G5 at home, or my 2x1.4 G4 at work, wasn't that it didn't process fast enough. It was GUI sluggishness.
The interaction between the GUI and the user IS the computer from the user perspective. It is the most important aspect of the computer usage experience. By fixing this, drastically improving it, Panther has drastically improved Mac OS X. Regardless of Tidris' numbers, no matter how much he thinks we should all just accept his standard as the gospel metric.
I could care less how fast Panther crunches numbers as opposed to Jaguar. My beef with Jaguar on my 1.8 G5 at home, or my 2x1.4 G4 at work, wasn't that it didn't process fast enough. It was GUI sluggishness.
The interaction between the GUI and the user IS the computer from the user perspective. It is the most important aspect of the computer usage experience. By fixing this, drastically improving it, Panther has drastically improved Mac OS X. Regardless of Tidris' numbers, no matter how much he thinks we should all just accept his standard as the gospel metric.
I know that many people don't do computationally intensive tasks on their machines. That is why Apple sells a ton of iMacs and iBooks. However, there are many other people who do. That is why Apple also sells a ton of dual G5s. If your work involved having to wait 3 hours for a dual G5 to finish converting a single movie to MPEG2 format, perhaps you would understand my point of view.
Obviously this thread is directed to those who do computationally intensive tasks. iMac users feel free to move on.
I know that many people don't do computationally intensive tasks on their machines. That is why Apple sells a ton of iMacs and iBooks. However, there are many other people who do. That is why Apple also sells a ton of dual G5s. If your work involved having to wait 3 hours for a dual G5 to finish converting a single movie to MPEG2 format, perhaps you would understand my point of view.
Obviously this thread is directed to those who do computationally intensive tasks. iMac users feel free to move on.
That was a bit disingenious of you... several of us *do* require computational performance in our work, but didn't have (I think) unrealistic expectations of what Panther was going to provide is all. I recall rumors of the newest gcc showing a 20% increase in *certain* operations, and that was extrapolated out by uninformed parties to a 20% increase across the board... that was never the case. You were misinformed is all.
That was a bit disingenious of you... several of us *do* require computational performance in our work, but didn't have (I think) unrealistic expectations of what Panther was going to provide is all. I recall rumors of the newest gcc showing a 20% increase in *certain* operations, and that was extrapolated out by uninformed parties to a 20% increase across the board... that was never the case. You were misinformed is all.
Actually I was not misinformed because I had been telling Mac users I know that Panther was not going to increase number crunching speed in apps. However, nobody wanted to believe me because of all the missinformation being parroted in all the Mac forums. Now that Panther is finally out I am able to show them I was right using real life number crunching apps. Despite that, some remain skeptical of my claim.
Actually I was not misinformed because I had been telling Mac users I know that Panther was not going to increase number crunching speed in apps. However, nobody wanted to believe me because of all the missinformation being parroted in all the Mac forums. Now that Panther is finally out I am able to show them I was right using real life number crunching apps. Despite that, some remain skeptical of my claim.
So you knew all along, and yet you are 'disappointed'??
How very odd.
Quote:
I am a retired software developer, by the way.
That's nice. I'm in the final stages of my dissertation in software engineering.
What exactly does any of that have to do with the speed of the OS?
Time for a lock, methinks.
BTW, many of us do appreciate the time spent in creating those tests and reporting the benchmarks. Thanks for the data.
Comments
Originally posted by johnq
10.2.8 introduced a bug in which fix_prebinding or redo_prebinding (I forget which now) would run before any application launch - including background processes. This is why looking up a webpage, or if you misspelled a word, or any number of other basic tasks would give a 30 second pause.
Strange. I never saw that on any of our 10.2.8 machines, and I definitely would have noticed it. The only vaguely similar problem I've seen is that Mail will occasionally block the entire OS while it's "writing changes to disk", but it did that before and after 10.2.8. Haven't seen it in Panther yet, so maybe it's fixed.
Maybe they put G4s in the iBooks because panther is optimized for the G4 and not so much yet for the G5s.
Originally posted by 3.1416
Strange. I never saw that on any of our 10.2.8 machines, and I definitely would have noticed it. The only vaguely similar problem I've seen is that Mail will occasionally block the entire OS while it's "writing changes to disk", but it did that before and after 10.2.8. Haven't seen it in Panther yet, so maybe it's fixed.
Fair enough 3.1416 but it does exist for some users.
It's hard to track down when it was introduced, I merely noticed it after 10.2.8.
http://www.osnews.com/comment.php?news_id=3210
http://forums.macrumors.com/archive/topic/39265-1.html
Anyway it seems in Jaguar there was a change in how/when various pre-binding utilities ran as opposed to older version. I can't speak to the details since the problem was so poorly documented and I just said F-it and just worked through the beachballs.
I don't presume everyone had the problem, don't presume no one had it.
If you saw a beachball or waited 30 seconds for something to happen and then suddenly several queued actions all suddenly are allowed to run, then you probably had this pre-binding problem and didn't know it.
It's gone for now, I'm happy. Yay Panther.
Originally posted by johnq
If you saw a beachball or waited 30 seconds for something to happen and then suddenly several queued actions all suddenly are allowed to run, then you probably had this pre-binding problem and didn't know it.
It's gone for now, I'm happy. Yay Panther.
You may also have had NetInfo problems. I know for a fact that can cause system-wide pauses as the NI system times out on various requests... and it acts just like that.
Changed one little thing on a G3 server a few months ago in the NI setup and suddenly the speed was *much* nicer.
I know for a fact that top -u would show fix_prebinding at a huge CPU% for about 30 seconds.
That DID always seem like a timeout issue (annoyingly exactly 30 seconds!), so maybe NetInfo was involved too, thanks.
I do think it is a particular issue that will exist on non-upgraded systems for a long time from now, so it's a shame it was never really fixed within the Jaguar updates. I doubt it will be now.
Feel sorry for Jag users that have that problem and who will assume that Panther will also "suck" in that same manner. Past experiences can and will effect future purchases and decisions, so the fact that I don't have the problem in Panther isn't reassuring, knowing so many people will either never upgrade, or wait a few years to do so.
Upgrade, my people! </booming morphius voice>
Originally posted by Locomotive
I'm sure Apple will hasten the Jaguar-to-Panther migration by introducing new versions of QT, iApps, etc., that no longer work in 10.2 -- as with 10.1 and 10.0. If this is really a modern structured OS, shouldn't software and firmware work in all releases (10.0,10.1,10.2,10.3) from this century? When was the last time you looked at any software or firmware that said simply Mac OSX compatible -- 10.0 or greater?
Yeah, that's right. They break stuff on purpose so you have to upgrade. It has nothing to do with the fact that the frameworks changing or new frameworks being introduced AT ALL.
Look at it this way, they write new systems that allow your old apps to continue working. *THAT* is compatability, and about as good as you can hope for.
Originally posted by Tidris
The common thing in all those tests is that they have a lot of number crunching activity and almost no GUI activity. As you can see, in all cases the difference between Jaguar and Panther is minimal, except in the third test where Jaguar was faster than Panther by almost 6%.
You cannot expect something else since the two OSs have been produced with compilers not that different, and share about the same optimisations at this level. When and if Apple moves to a better compiler (e.g. the IBM one), then you can expect improvements in processor performance.
The interaction between the GUI and the user IS the computer from the user perspective. It is the most important aspect of the computer usage experience. By fixing this, drastically improving it, Panther has drastically improved Mac OS X. Regardless of Tidris' numbers, no matter how much he thinks we should all just accept his standard as the gospel metric.
Originally posted by Kirkland
I could care less how fast Panther crunches numbers as opposed to Jaguar. My beef with Jaguar on my 1.8 G5 at home, or my 2x1.4 G4 at work, wasn't that it didn't process fast enough. It was GUI sluggishness.
The interaction between the GUI and the user IS the computer from the user perspective. It is the most important aspect of the computer usage experience. By fixing this, drastically improving it, Panther has drastically improved Mac OS X. Regardless of Tidris' numbers, no matter how much he thinks we should all just accept his standard as the gospel metric.
I know that many people don't do computationally intensive tasks on their machines. That is why Apple sells a ton of iMacs and iBooks. However, there are many other people who do. That is why Apple also sells a ton of dual G5s. If your work involved having to wait 3 hours for a dual G5 to finish converting a single movie to MPEG2 format, perhaps you would understand my point of view.
Obviously this thread is directed to those who do computationally intensive tasks. iMac users feel free to move on.
Originally posted by Tidris
I know that many people don't do computationally intensive tasks on their machines. That is why Apple sells a ton of iMacs and iBooks. However, there are many other people who do. That is why Apple also sells a ton of dual G5s. If your work involved having to wait 3 hours for a dual G5 to finish converting a single movie to MPEG2 format, perhaps you would understand my point of view.
Obviously this thread is directed to those who do computationally intensive tasks. iMac users feel free to move on.
That was a bit disingenious of you... several of us *do* require computational performance in our work, but didn't have (I think) unrealistic expectations of what Panther was going to provide is all. I recall rumors of the newest gcc showing a 20% increase in *certain* operations, and that was extrapolated out by uninformed parties to a 20% increase across the board... that was never the case. You were misinformed is all.
Originally posted by Kickaha
That was a bit disingenious of you... several of us *do* require computational performance in our work, but didn't have (I think) unrealistic expectations of what Panther was going to provide is all. I recall rumors of the newest gcc showing a 20% increase in *certain* operations, and that was extrapolated out by uninformed parties to a 20% increase across the board... that was never the case. You were misinformed is all.
Actually I was not misinformed because I had been telling Mac users I know that Panther was not going to increase number crunching speed in apps. However, nobody wanted to believe me because of all the missinformation being parroted in all the Mac forums. Now that Panther is finally out I am able to show them I was right using real life number crunching apps. Despite that, some remain skeptical of my claim.
I am a retired software developer, by the way.
Originally posted by Tidris
Actually I was not misinformed because I had been telling Mac users I know that Panther was not going to increase number crunching speed in apps. However, nobody wanted to believe me because of all the missinformation being parroted in all the Mac forums. Now that Panther is finally out I am able to show them I was right using real life number crunching apps. Despite that, some remain skeptical of my claim.
So you knew all along, and yet you are 'disappointed'??
How very odd.
I am a retired software developer, by the way.
That's nice. I'm in the final stages of my dissertation in software engineering.
What exactly does any of that have to do with the speed of the OS?
Time for a lock, methinks.
BTW, many of us do appreciate the time spent in creating those tests and reporting the benchmarks. Thanks for the data.