|
To me the cylinder Mac Pro problem is that Apple bet big on the future of workstation computing being GPU compute to the exclusion of all else. They took out internal storage bays, they decided internal PCIe cards other than GPUs were superfluous, they even axed the second CPU socket. The sales pitch was "we fit two awesome GPUs in this crazy small/quiet form factor. Isn't this great???". But it wasn't. GPU compute is important, but it's not the only thing. Shitloads of software developers want dual CPU socket systems with a ton of CPU cores to cut down build times on their large projects. Many workstation buyers want at least a little internal storage expansion. The compromises they made to pack as much GPU as possible into such a small machine made it undesirable for everyone else. Even people who were doing serious GPU compute probably found it limiting. IIRC Apple had to limit the power consumption per GPU below what would be possible in a bigger box, and obviously there was no way to add a third or fourth GPU. The nail in the coffin was developer uptake. Apple's grand plan depended on OpenCL going big and becoming the industry standard GPU compute environment. OpenCL was Apple's baby, so of course that's the way they went, but it lost to Nvidia's proprietary CUDA. With the benefit of hindsight this battle was probably over before the trashcan started shipping, and possibly even before it was announced. (I remember arguing otherwise, but I was wrong.) It's pretty obvious the trashcan was in development hell for many years. Maybe it could have had a chance if it was on the market sooner, while there was still a window to make a much stronger OpenCL push. Regardless of what could have been, Apple wasted a shitload of money and time re-imagining the Mac Pro, and it was all for nothing. At some point in the last few years they should have cut their losses and brought back a cheese grater design with updated internals, but I suspect there are some executive egos involved. It's difficult to admit that "well, this thing we said was the future of pro computers was a failure, have a new boxy box". It's become the embarrassing thing in the corner which nobody wants to acknowledge. Presumably sells well enough to make it worthwhile to keep manufacturing the existing design, but not enough to justify a refresh.
|
#
?
Mar 19, 2017 23:06
|
|
|
|
| # ? May 31, 2024 21:05 |
|
|
People have done some pretty impressive upgrade hacking of the cheese graters, but it's a pretty sad indication of Apple overall that they went nuts trying to pursue "Pro" users, then dropped them like a hot potato once they made all their money from the iPhone.
|
|
#
?
Mar 19, 2017 23:11
|
|
|
BobHoward posted:Presumably sells well enough to make it worthwhile to keep manufacturing the existing design, but not enough to justify a refresh. That's a pretty big presumption. Apple's the kind of company that it would be easy to hide a few big failures in for a very long time. iPhone money is more than enough to bury your loss making iTrashcan line under.
|
|
#
?
Mar 19, 2017 23:14
|
|
|
Interesting side note: Nintendo Switch owners are now buying Macbook USB-C chargers because nobody trusts USB-C wall warts from other companies.
|
|
#
?
Mar 19, 2017 23:16
|
|
|
Considering that the Mac Pro is Apple's only "Made In America" product and since Trump went after them for just that, I wouldn't be surprised to see a low-effort Pro upgrade or revamp to keep the Orange-in-Chief happy.
|
|
#
?
Mar 19, 2017 23:16
|
|
|
The pro is dead
|
|
#
?
Mar 19, 2017 23:27
|
|
|
I wonder if a Mac Pro with two gtx cards could've tipped the scales but my imagination comes up short every time.
|
|
#
?
Mar 20, 2017 00:07
|
|
|
What I don't understand is why some of y'all drag your feet on just embracing and trying to make iOS better. For better or worse that's the way things are heading. If your response has anything to do with your job you need to realize you're in a super niche market that Apple isn't going for anymore. Most consumers are good with iOS and it's great and they should focus on it instead of also worrying about what random web guy thinks he needs.
|
|
#
?
Mar 20, 2017 00:20
|
|
|
~Coxy posted:You must have the JDM model.  If it's a joke it's gone right over my head.rally posted:It is anyway. My 2016 tb has it 
|
|
#
?
Mar 20, 2017 00:26
|
|
|
TACD posted:I... don't know what this means Ah, I somehow thought we were talking about three finger drag.
|
|
#
?
Mar 20, 2017 00:39
|
|
|
GoldfishStew posted:What I don't understand is why some of y'all drag your feet on just embracing and trying to make iOS better. For better or worse that's the way things are heading. Great. Give me iOS with something approaching finder and an easy way to set up folders within folders for documents and I'll jump ship.
|
|
#
?
Mar 20, 2017 00:46
|
|
|
Cyrano4747 posted:Great. Give me iOS with something approaching finder and an easy way to set up folders within folders for documents and I'll jump ship. documents by readdle can basically do this
|
|
#
?
Mar 20, 2017 00:49
|
|
|
Cyrano4747 posted:Great. Give me iOS with something approaching finder and an easy way to set up folders within folders for documents and I'll jump ship. I'm sure it's coming. Somewhere deep inside Cupertino there's an iPad with a mouse attached. It's just a matter of how seamless they can make it work. You can already see the faint glimpses of it while attaching a keyboard to an iPad and being able to use MacOS keyboard shortcuts.
|
|
#
?
Mar 20, 2017 00:56
|
|
|
GoldfishStew posted:What I don't understand is why some of y'all drag your feet on just embracing and trying to make iOS better. For better or worse that's the way things are heading. "Everyone who uses a desktop for work" isn't a niche market what decade is this post from. e: it's just smaller than "owns a cellphone that I will rebuy for one feature every year" and "need a $2k computer for browsing facebook in my college class" markets.
|
|
#
?
Mar 20, 2017 01:02
|
|
|
LOL if you think there will be IOS with mouse. No. That's not it. That's not how this is going. And finder/search on my iPad Pro is great I command space and find whatever I want.
|
|
#
?
Mar 20, 2017 01:13
|
|
|
Quantum of Phallus posted:The pro is dead Edit: BobHoward posted:Regardless of what could have been, Apple wasted a shitload of money and time re-imagining the Mac Pro, and it was all for nothing. At some point in the last few years they should have cut their losses and brought back a cheese grater design with updated internals, but I suspect there are some executive egos involved. It's difficult to admit that "well, this thing we said was the future of pro computers was a failure, have a new boxy box". It's become the embarrassing thing in the corner which nobody wants to acknowledge. Presumably sells well enough to make it worthwhile to keep manufacturing the existing design, but not enough to justify a refresh. Canned Sunshine fucked around with this message at 01:41 on Mar 20, 2017 |
|
#
?
Mar 20, 2017 01:28
|
|
|
Cyrano4747 posted:Great. Give me iOS with something approaching finder and an easy way to set up folders within folders for documents and I'll jump ship. I've mentioned a couple times that if the iPad Pro had finder, terminal and Xcode it would be enough for me to give it a shot. The thing that makes my MacBook Pro a more comfortable experience with general browsing (and work too!) is having the touchpad. I can't imagine gorilla arming all day. What's the solution for this?
|
|
#
?
Mar 20, 2017 03:01
|
|
|
I'd guess I'd be fine with iOS with full access to general computing ala Terminal/Finder/ability to run whatever software with the App Store being just a suggested gatekeeper along the lines of the Mac App Store, but I don't see this ever happening.
|
|
#
?
Mar 20, 2017 03:25
|
|
|
I'm really not trying to be rude but on my pro I bring down search and it searches all my files including Dropbox, etc. what else do you want?
GoldfishStew fucked around with this message at 04:04 on Mar 20, 2017 |
|
#
?
Mar 20, 2017 04:00
|
|
|
I want MacOS
|
|
#
?
Mar 20, 2017 04:33
|
|
|
GoldfishStew posted:I'm really not trying to be rude but on my pro I bring down search and it searches all my files including Dropbox, etc. what else do you want? I want to be able to organize my data into a folder hierarchy of my choosing and run software of my choosing on my general purpose computer. The same as any computer since 1977?
|
|
#
?
Mar 20, 2017 04:36
|
|
|
you can do all that on Android .... I think hierarchical file systems are outmoded personally. Sure it's nice for us because we understand them and are used to them, but why do they need to exist in the way that they do? Pretty much every large technology company - Microsoft (WinFS), Apple (internal research precursor to APFS), developers of ZFS and even Reiser all experimented with indexed, or relationally-database backed file systems. A lot of people think hierarchical file systems are the 'bare metal' as in they represent how data ACTUALLY exists on disk, but they are wrong - in truth, data stored on disk has no real hierarchy and folders and file system permissions and all that fun stuff is things we added just to have it make sense. There's no real reason storage is presented the way it is other than it's just the way we're used to. Really there's no reason some bits in some folder are executable by some group and others aren't, it's something we invented to make sense of the mess we inherited. Assuming most of us here are in our 20s and 30s, we grew up with folders and files, but our kids will grow up with database backed file systems where data is grouped by context more than by arbitrary hierarchy. Pivo fucked around with this message at 04:50 on Mar 20, 2017 |
|
#
?
Mar 20, 2017 04:47
|
|
|
Saying an iPad Pro can competently replace a traditional personal computer is loving dumb.
|
|
#
?
Mar 20, 2017 04:51
|
|
|
If you don't provide a hierarchical file system, all that happens is app vendors/OS vendors start letting users group only specific of data in hierarchies. Photos in albums, etc. One-off poo poo instead of general purpose. That's not even to mention collaboration across large sets of documents within an organization. I think that none of us need file systems when we're dicking around on the internet on an iPad. Many of the users using PCs and Macs to make a living right now (not necessarily using niche software for the tech-savvy) would be less productive with iPads. I like my iPad. It's nice for reading and watching videos. I would never want to use it for work, even if my work was just editing documents and spreadsheets and emailing people.
|
|
#
?
Mar 20, 2017 05:00
|
|
|
Yes I couldn't get by with just an iPad as well, knowing the hierarchy of my files does mean a lot for me. Having a git repo in a certain folder means something more than just having a git repo. BUT. There is nothing saying that Git repo can't be classified as metadata for Git. Then the metadata itself can say it is responsible for a certain set of files that can be opened by your text editor, by your code editor, by a web browser, etc. Does it matter that applications using those files know that they're in a directory you can't even get the listing for? Not really, all they care about is the files. It's all about a mindset. Folders and permissions make sense because it was an easy hack that lasted through time. Not only is there is a literal physical metaphor but it is easy to implement and reason about. You will lose no functionality by having the entire filesystem be database-backed. In fact you gain a lot of functionality. We've tried to hack that on, like with indexing in modern OSes or with the file system jails iOS and Android apps sit in (and Windows Phone apps too I think?), and guarding access to 'not your files' at the OS level.... The only reason you think files in folders (which can be hidden depending purely on their name? and depending on what interface you use to list them?) is normal is because you're used to it. There WILL be a new normal.
|
|
#
?
Mar 20, 2017 05:07
|
|
|
While we're on the topic let's talk about file extensions! And file system limits on folder names, file names... And that in some cases, 'touch' on a stale mount point can hang your terminal until WHATEVER decides to time out, decides to time out. File systems are broken and everyone's been trying to fix it for years and years.
|
|
#
?
Mar 20, 2017 05:10
|
|
|
iPad sales keep going down. I don't think it's going to ever replace the desktop/laptop.
|
|
#
?
Mar 20, 2017 05:38
|
|
|
Pivo posted:you can do all that on Android .... I can do all of that on proper desktop Linux too, this is the Mac thread, not the iOS thread, right? I was entertaining a discussion of what it would take to get me to accept iOS on my Mac desktops/laptops, should the day ever come where they try to converge the two OS. Did you mean that if/when that day comes and these features are missing, I should start running Android on my laptop? I do have an iPhone and while I would *prefer* an open computing environment there, I don't really give a poo poo since I have no expectation that the thing behave like a proper computer. It makes a cute approximation a lot of the time, but you can't even write a 'Hello World' program on one. I'm open to new ideas, I just don't want to lose functionality. I don't really know anything about these non-hierarchical filesystems, so lets consider a random scenario where I'd rely on a traditional file system to get the job done: -Let's say I wanted to grab the audio files from the latest FPS game so I can splice them up and make a hilarious meme for GBS. Mind you I don't know if the game developer stored them as mp3 files, as some other audio format. I definitely don't know their names. On OSX, I could browse the .app container and locate audio files of various formats by their file extension. If I don't have the right software to open them, I could Google some and grab it if needed. How would you approach this on a non-heirarchical OS?
|
|
#
?
Mar 20, 2017 05:57
|
|
|
See, black guys, they use post-PC computing devices like this. But see, white guys, they use post-PC computing devices like this!
|
|
#
?
Mar 20, 2017 06:06
|
|
|
have they fixed itunes? by fixed i mean burned to the ground. it's been 10 years since the iphone came out and i still can't just drag and drop music from the finder/explorer window. instead i gotta deal with this through itunes. Strong Sauce fucked around with this message at 07:43 on Mar 20, 2017 |
|
#
?
Mar 20, 2017 06:06
|
|
|
Pakistani Brad Pitt posted:How would you approach this on a non-heirarchical OS? What you consider the file extension now would be stored as metadata. So you'd just query the file system for audio files belonging to that application. Not much different from searching the .app folder for audio files manually. Neither representation is an accurate representation of how files are stored on disk, both interpretations have the same functionality (containing folder, file name/extension, and permissions are really just metadata associated with an object), except the database backed one has far richer possibilities in querying it. You're not limited to just searching for file names in folders and manually searching through their contents. Instead imagine a scenario where you have a bunch of spreadsheets. Some are on a USB drive, some are on your laptop's SSD, some are on a network share. Some are tagged 'Finance', some are tagged 'Personal'. How would you craft a search across your entire file system for personal finance spreadsheets? Well without a database I guess you'd rely on a file system index, you'd hope it's up to date, and you have to know low-level (relatively speaking) details like the file extension you used. Is it a CSV or an XLS? So you'd manually craft a search across all those locations looking for a specific file extension, which is really just part of the name... Orrr you could just search for files tagged 'finance' and 'personal' that can be opened by Excel. Now you don't care WHERE they are, WHAT they are, only that they match your criteria! Trust me this is where file systems are going. Pivo fucked around with this message at 06:23 on Mar 20, 2017 |
|
#
?
Mar 20, 2017 06:19
|
|
|
Pivo posted:What you consider the file extension now would be stored as metadata. So you'd just query the file system for audio files belonging to that application. Not much different from searching the .app folder for audio files manually. But MacOS right now doesn't know that an Ogg Vorbis file is an audio file. So how would this query find it for me if I can't browse for it?
|
|
#
?
Mar 20, 2017 06:38
|
|
|
Pakistani Brad Pitt posted:like a proper computer. This definition is evolving and you're having trouble. We become our own grandparents, eventually, maybe.
|
|
#
?
Mar 20, 2017 06:46
|
|
|
Pakistani Brad Pitt posted:But MacOS right now doesn't know that an Ogg Vorbis file is an audio file. So how would this query find it for me if I can't browse for it? Instead of a file with a name in some folder pointing to a physical location, you would have a database row with metadata pointing to a physical location. In the end it's the same. Currently we use file extensions. Often file formats have a header or preamble, a few bytes at the top of the file that tell you it's supposed to be a file of that format. The goal is to be smart enough to keep that metadata associated with the file, for applications to know they can read that file, and to decouple the file/folder structure from accessing/writing files. Currently we store a lot of information in the file's location and file name, and naively users assume this has anything to do with where it's stored on disk. It doesn't, so why have it? What changes for you if the file system uses an RDBMS to find files that claim they're OGG? Nothing, really. But by saying this stuff is just metadata we can remove the importance of this historical hack and realistically say, this is just a piece of metadata associated with the file, and its location doesn't matter and never has. Indexing the file system is how we currently implement this on top of the old hierarchical file system. iOS goes even further by completely hiding the file system from users. The goal is to make this abstraction the definition, instead of creating hacks that pretend it is - and make people think they need to view the heirarchal file structure to feel like they're in control. In reality it doesn't matter (or at least shouldn't matter) and shouldn't even exist at the low level. It's unfortunate attempts at an RDBMS backed file system have failed, even though everyone has tried it, but IMHO it was mainly due to friction of having to deal with the old file system. iOS hiding it completely goes a long way to accomplishing that. Pivo fucked around with this message at 06:52 on Mar 20, 2017 |
|
#
?
Mar 20, 2017 06:48
|
|
|
Let me give a concrete example. Sometimes when you are editing XML files, you want to do it in an IDE, sometimes you want to do it in a plain text editor. The XML file belongs to some project in some directory. Git cares about that file because it needs to track it for changes. Your IDE cares about that file because it's part of a larger project. Your text editor cares about that file because you've intended to edit it. Where does that file BELONG? Who owns it? Why does it matter where it exists in some arbitrary hierarchy? It should be part of the Git repo at the same time it is part of your IDE's project at the same time it is editable in your text editor. In this case saying it is under ~/my_project/config is really no different from NAMING it, except the name is really only one piece of metadata. Why define that object (file) by its name, when its name ultimately matters so little? The context of what that file belongs to - a git repo, a larger project, and a text file you edit manually - matters so much more than ~/my_project/config. And no one's asking you to give up that name. It's just a lot easier for everyone to find it when the name isn't the only thing you know.
|
|
#
?
Mar 20, 2017 06:59
|
|
|
Ugh, and look at how Time Machine is implemented. It has to fake the whole directory structure with links just to make it look like a snapshot is of the whole system, and tools fail to properly understand how to count links... Even look at Windows, OS X or Linux directory structure. You can edit drat near anything under /Users/yourusername (unless otherwise specified) but you can't edit poo poo under /Library without elevation, so even saying 'but but the hierarchy implies rights!' doesn't even hold true, two objects at the same depth can have wildly different permissions. A file name and location is literally just that. A name. There is so much more information that can be associated with it. And this is getting more and more important where a file isn't even necessarily on your machine. My god we even can mount SFTP locations as if they're part of the local filesystem just to hack that to work. Even Android mounts SD cards strangely and people write software that links the SD card into where we expect main storage to be just to trick the OS into knowing the reality of how much storage we have available! All of these problems just because we care so much about which name points to which group of blocks on disk! Decouple that!
|
|
#
?
Mar 20, 2017 07:18
|
|
|
I guess what I'm getting as is how do you look/query for something that isn't nicely associated with some project / app / whatever on your computer? I'm looking for the edge case. Let's say I'm studying programming, and make a program while learning file I/O that spins out of control. It generates 1 million 100KB files of random data on my hard drive, consuming 100GB of disk space. On a hierarchical system, I know what folder they were output to, so I can locate it and delete them. On this system where they now just exist in some flat blob of files, how would you locate them to delete them?
|
|
#
?
Mar 20, 2017 07:41
|
|
|
Pakistani Brad Pitt posted:I guess what I'm getting as is how do you look/query for something that isn't nicely associated with some project / app / whatever on your computer? I'm looking for the edge case. All the files will have metadata associating them with the process (tree) that created them. The operating system won’t allow you to write files without this metadata any more than current operating systems allow you to write files that belong in no directory.
|
|
#
?
Mar 20, 2017 07:55
|
|
|
By asking the file system - which files were created by a certain user/account recently? Which files were created by a certain process recently? And of course, hopefully you still have some sort of names on these things. Getting rid of the directory structure doesn't mean they're just random blobs, they still have names. In the end, putting the files in a specific directory is no different than giving them a predictable name. You just have more data to play with.
|
|
#
?
Mar 20, 2017 07:59
|
|
|
|
| # ? May 31, 2024 21:05 |
|
|
Platystemon posted:All the files will have metadata associating them with the process (tree) that created them. OK, I can buy all of this. Metadata for 'what process created this file', metadata for file creation time, metadata for 'homework' and also for 'Aiden' (so you can differentiate his homework-tagged files from his brothers Jayden and Caden.) So what does the UI look like that allows to you to make these kind of complex and arbitrarily strung together queries so you can say: 'show me all of Aiden's homework files from the last two weeks that were also touched by myShittyFirstProgram.exe'? Siri? dear god..
|
|
#
?
Mar 20, 2017 08:01
|
|