Jump to content

Jay Parlar

  • Content count

  • Joined

  • Last visited

Posts posted by Jay Parlar

  1. I have a 2 gig G5 Powermac, 1.66 dual core Mini and a 2 gig 17" iMac. All running across a gigabit (locally cabled) ethernet network and all with either additional iSights or built in.


    All works perfectly and I can carry out device to device video iChats. However if I try to initialise a 3 way conference the quality degrades significantly. I have Quicktime streaming enabled and the network definately runs at gig speeds so I'm at a bit of a loss as to why I'm losing quality.


    I've tried both over bonjour and aol but exactly the same symptoms.


    Anyone any ideas?







    I've done a 3-way conference on a local wireless network (802.11G) between 2 Macbooks and a 24" iMac, and it worked flawlessly. I'd suggest opening up Activity Monitor on all three machines during the conference call, and see if any of them are pegging either the CPU or the network really hard. Gigabit is way more than enough, so it sounds like something weird is going on.

  2. It's always the same thing: "The POP server "pop-server.rochester.rr.com" rejected the password for user. Please re-enter your password, or cancel." Every time I enter my password, click "Remember this password in my keychain," and hope the message goes away permanently. But it doesn't. It comes back 3 minutes later. I know I'm typing in the right password. I know I am. I know categorically that I am typing the right password. And yet still the message persists, with no way to stop it except canceling it (in which case I receive no new mail) or quitting the mail application (in which case I receive no new mail). Any suggestions?


    I see this sometimes, and thus far I've blamed it on problems with the ISP. I think that this is the behaviour you see if there's an error connecting to the POP server.


    Of course, Mail.app *should* really give a better error. When it happens, I usually just hit "Cancel", and try again later.

  3. Ok, here are my latest results:


    Doing a Command-Y to hide the meters chops a few points off the CPU requirements.


    Locking tracks makes a HUGE difference. I created a song with about 20 tracks, and it would only play about two seconds of it before giving up. If I lock all the tracks, CPU usage drops to almost nothing (Activity Monitor shows Garageband at 30%).


    Here's the really interesting thing: When I do it without locking the tracks, and Garageband stops playing it, it *is* in fact spreading the load over both processors (according to the "CPU Usage" graph), but it refuses to use more than about 50% of each processor. As soon as the process gets to about 80% CPU, it cuts out. So *maybe* there's something in Garageband that makes the song stop if the song hits a certain percentage. Of course, when it cuts out, there's still a ton of CPU available (because a process at 80% CPU means using each CPU at 40% each).


    So, yeah, that's about it. Locking the track works great, and I was technically wrong about it not spreading the load. It does spread the load, but refuses to put too much load on it.

  4. Close widgets to free up more RAM, try again.


    Like I said earlier, there were over 400 megs free, it's a processor thing.


    Did you try locking the tracks after they were created its supposed to cut down on the amount of procceser power needed


    I'll give that a try. Of course, even if I free up more of the processor, it'll still bug me that his Garageband is still ignoring the other one :)

  5. One would expect applications to be agnostic to multiple cores, and that the operating system handles this in stead. That is, if the application threads can be divided among the cores, because that is a prerequisite.


    That's exactly what I thought too. However, I'm beginning to believe that's not the case. If you look at Activity Monitor, pretty much every app is running with multiple threads, I believe that Cocoa pretty much demands it. However, it doesn't really look like *any* app I run ever spawns across both processors.


    Maybe Apple forced Garageband to stay on one processor, to give people reason to upgrade to Logic?

  6. I really have no idea about garageband using only one processor versus both at the same time. However, I believe that error message you mentioned actually has to do with how much ram you are using. I tend to go nuts with tracks and even with 2gb of ram I'll get that error when I get to around 15 tracks of real instruments (although it will still play the song after a few "hiccups.")


    I checked the RAM usage as well, he had tons free (at least 400 megs). The one processor core was completely pegged though, which is why I'd love for it to use both.

  7. Anyone know why Garageband only ever uses one of the processors in the Intel machines?


    My musician friend just got a Mac Mini to play around with Garageband, and after adding a ton of tracks and effects, he gets the dreaded error message from Garageband saying too much was going on.


    So we try the song again, but this time I run ActivityMonitor at the same time. What a surprise I got when I saw that only one of the processors was being pegged for Garageband, despite 18 threads running.


    Am I missing something here? I've never programmed against Cocoa before, but I would assume that if you had multiple threads, the OS would spread them out over the multiple processors. It seems that Garageband is putting ALL of the audio processing into just one thread. Either that, or I have no understanding of how threads run on OS X (which is just as likely).


    Jay P.

  8. Thanks. The wireless router is a AT&T Homeportal which gives me 802.11G. One Mac is ethernet wired. The other 2 macs are wireless described as Airport in Sys prefs.


    The main reason I asked is Nintendo indicates you got to have Windows XP and IE, but don't indicate what it is used for.


    If you buy Nintendo's USB WiFi adapter, you do need Windows. If you already have a wireless router though, you should be fine. My network has two Macs, one Linux machine, a Wii and a DS on it, no Windows in sight.


    Jay P.

  9. My friend recently brought over 15 minutes of MiniDV footage (not a lot, I know). We edited it in iMovie, and burned it with iDVD.


    The iDVD encode process took about 30 minutes, but it seemed (IIRC) that most of that time was spent doing the animated menus, of which we had a few! The audio processing especially seems to take awhile to do. We originally had a 6 minute song looping on one of the menus, we chopped that down to a 30 second loop because the encode was taking forever.


    This is on a 2.16 GHz Core 2 Duo iMac with 2 gigs of RAM.


    Jay P.

  10. Oh, and one more piece of info.


    In the second before it crashes, I can see all three of the Keychains are there (login, System and X509Anchors), but none of the items in the login keychain are showing up in the main part of the window.


    This leads me to believe that it's crashing while reading my login keychain. But it's strange that it would crash when running x86 code, and not PPC code, unless they screwed up some endianness issues. I'd imagine though that would have been caught a long time ago, were it the case.


    Jay P.

  11. Very weird problem here. I've got a new 20" iMac (well, I've had it for a little bit), and for the first time today I decided to run "Keychain Access". I use that program a lot on my old Powerbook (running Panther) to store my passwords, so I wanted to see how the program looks under Tiger.


    Well, everytime I start it, I see the application window appear, and then it crashes one second later. Without fail. This is scary, obviously, because Keychain Access is pretty important.


    I did however find a workaround. If I open up the Info dialog for the Keychain Access app, and tell it to run the program using Rosetta, then everything works, fine and dandy like sour candy!


    I've run Repair Permissions on the disk. I've run the Keychain First Aid (using Rosetta). Still, the Keychain Access app crashes if I try to run it normally.


    Has anyone seen anything like this before? Any ideas?


    Oh, and the crash report is very unhelpful. It of course has gobs of stuff in it, but the key lines (in this programmer's opinion) are:


    Exception: EXC_BAD_ACCESS (0x0001)

    Codes: KERN_INVALID_ADDRESS (0x0001) at 0xb1e06f60


    That's some bad mojo right there!



    Jay P.