Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
OS X 10.8.4 Build 12E55 Seeded to Developers
- Thread starter MacRumors
- Start date
- Sort by reaction score
Not only that, but they are limiting their testing to the HD3000, which has never had the best OSX support.
----------
As far as I'm aware all of the vendor drivers work as plugins, which hook into the graphics core of the os, which Apple writes.
So Apple controls the front end, and the vendors write the back end drivers that actually talk to the card.
This is why nvidia can offer updated drivers on their site outside of os os updates.
You had me until "Solid".
Seriously, who do you think "they" is that are testing OSX? It's developers out in the real world who repeatedly report things only to find that in the 5th and even 6th iteration they are ignored.
So, you don't want solid updates? Btw. I already mentioned that I wished updates to be more ambitious. But even though they are quite narrow in their scope, they seem to have solved a lot of issues on the incremental updates, especially in terms of Safari's performance and stability.
Most of the issues seems to be gone on OS X 10.8.4.
Apple buy the chips from these companies, so they'd have all the technical documentation needed to develop their own drivers. They would never need to reverse-engineer open-source drivers if they were writing their own. There shouldn't be any issue with patents or copyrights - a license would presumably be included as part of the sale of the devices (what's the point in a bunch of GPUs you can't use?).
I used to write graphics drivers for ARM. The reason they don't open-source their drivers is because it could potentially expose some type of IP infringement. I should add that they weren't aware of any actual infringements, but if you publish your source code you give people something to dig through. It's better to be safe than sorry.
Modern graphics drivers are freakishly complex (especially in a world of GPGPU). I've always been told that Apple wrote their own drivers using the technical documentation they get from the manufacturers. If you're wondering, I didn't get that info from ARM - we never wrote drivers for Apple, but then Apple wasn't a Mali GPU licensee so we wouldn't have. That said, everybody there makes jokes about how much Apple customise from the IP they license; they tweak and change basically everything.
Last edited:
To be fair OSs also get more complicated with each new release. Perfection from the outset is a near-impossibility.
Whole-heartedly agreed! Sometimes one can't help but wonder though, if APPLE is spreading themselves too thinly these days. It's a well-documented fact they move their human resources talent pool around, as needed, rather than have software people, dedicated to one particular line. The former has imho many distinct advantages, not only for the company, but especially for us, the users. One reason why most APPLE software is so seamlessly integrated and runs so well. OK, before I get inundated, "relatively well".
Haven't experienced problems with these 'slow shutdown issues' as I rarely ever shutdown my MBA which runs ML;
I simply close the lid. The desktop is rock-solid running SL. No shutdown issues there.
Now if we could only get APPLE to add Multiple Time Machine Back-ups to SL, that would be fantastic.
I know, I know, not gonna happen.
Last edited:
I smell troll but I'll bite.
Yes there were as much beta testing in Jobs era as well.
Are we going to regret 'biting', but I'll add my two cents to your reply to koban4max that a lot of beta testing by any company is not only good, but commendable. Who wants to buy buggy, laggy software that freezes and fails to live up to expected and advertised standards? Todays OSs are becoming exponentially more complicated, so thorough testing is imperative if a company wants to keep their customers happy.
With today's software, perfection from the get-go is a near-impossibility. There always still seem to be issues that slip through the cracks, and need to be updated later. Many glitches in fact.
Without this extended beta testing, these issues/glitches would be a lot worse.
As always (in my case):
com.apple.coreservices.appleevents
com.apple.securityd
Lucky man. I shutdown my MBA because in sleep mode it discharges the battery much more than when it's shut down. And because it's the 11" one it really feels (yes, I've got the hibernatemode set to 0 so that it never uses SSD for saving the state, only RAM and thus more power is used). So now I use the known workaround with very short ExitTimeOut values (2s). Definitely not to be considered a clean solution, but at least brings some improvement.Haven't experienced problems with these 'slow shutdown issues' as I rarely or never shutdown my MBA which runs ML; I simply close the lid. The desktop is rock-solid with SL.
Now if we could only get APPLE to add Multiple Time Machine Back-ups to SL, that would be fantastic.
I know, I know, not gonna happen.
Last edited:
what about OPEN GL drivers ?
Intel's Linux OpenGL Driver Faster Than Apple's OS X Driver
http://apple.slashdot.org/story/13/...-opengl-driver-faster-than-apples-os-x-driver
Petition requesting Apple to support OpenGL 4.3 and ZFS in OS X 10.9
https://www.change.org/petitions/apple-os-x-10-9-support-opengl-4-3-and-zfs
Intel's Linux OpenGL Driver Faster Than Apple's OS X Driver
http://apple.slashdot.org/story/13/...-opengl-driver-faster-than-apples-os-x-driver
Petition requesting Apple to support OpenGL 4.3 and ZFS in OS X 10.9
https://www.change.org/petitions/apple-os-x-10-9-support-opengl-4-3-and-zfs
I think the ZFS requirement (in that petition) is based on the fact that it was already being tested at one point in the past. However, for example I don't want ZFS myself and don't like it either though I signed that petition. I believe some bigger filesystem change is really required. HFS+ is ancient at this point, it isn't even robust enough (you can for example easily corrupt data without any possibility to detect this corruption at the fs level) and even the performance is poor (for example ext4 easily outperforms it in practically every type of workload). At this moment it doesn't have any real advantage over other common filesystems.
So what. Things started to change fast and it was removed. Due to licensing and what not.
Just take it as "bring some big FS change" instead of "ZFS and only ZFS" petition. The person who created that petition didn't do it the best possible way, but it's at least some effort. Though I don't believe it will mean much to Apple and that they will change anything in the near future.
Another reason they dumped ZFS might be OS X crashing after you unmounted a USB ZFS formatted drive and then disconnect the Cable, this was a bug and might still be the problem.
Last edited:
They are not going to help him if he goes in with a beta OS on it. He would need to reinstall 10.8.2.
A lot of the time, the stuff developers and testers report are fixed in 1-2 years time. Not in 2 weeks. There's a long list of bugs for OS X, like any other OS. Unless the bug is life threatening it will be put at the end of the list. But of course new point updates introduce new bugs which can be squashed fast enough if it's easy to do. But nobody is being ignored.
----------
I don't think Apple will ever implement ZFS. If they ever change HFS, they'll probably switch to an in-house file system again.
A license is something that you can only give out if every copyright owner agrees and that's the exact problem in most cases: there are many people who have contributed code and it is very hard and sometimes impossible to track all of them. Giving out a license can be a very difficult thing because of things like this. Having documentation doesn't change anything. We've seen various real world examples where copyrights and patents create issues. Mostly due to the USPTO way of doing it (thorough checks are only done when somebody complains after the patent is granted instead of before granting the patent in the first place combined with the ability to patent nearly everything, even the most common and silly things). The H.264 codec is a good one as well as Googles WebM/VP8. There simply are too many legal issues in today's world.
The most likely and most realistic option would be that Apple is getting the chips from those manufacturers and they are helping them to write the drivers for OS X. In Linux there are proprietary binary blobs for both AMD/Ati and Nvidia. The kernel only has some stuff that allows it to access those blobs. That's probably how it works for OS X as well: there is only teamwork between Apple and AMD/Nvidia for accessing those blobs from OS X. That would also be the most easiest way of doing things for AMD/Nvidia technically.
They only offer driver updates for the optional cards they release for the Mac Pro (e.g. Quadro 4000). Those drivers are not for the GPU's in say the MBP or iMac.
To me, it makes sense that they are taking longer with this release. It's the first one since Forstall was fired. That only happened very late in 2012. That means that the clock was reset by roughly 6 months. (I am sure that Apple is already developing the next .x or x.0 revision long before the "current, new" revision is released.) The much ballyhooed "de-Forstall-ization" process has to take a while.
