tenthousandthings

About

Username
tenthousandthings
Joined
Visits
105
Last Active
Roles
member
Points
1,144
Badges
1
Posts
677
  • Leaked Apple Silicon roadmap hints at new Mac Pro, MacBook Air

    rob53 said:
    blastdoor said:
    rob53 said:
    blastdoor said:
    rob53 said:
    Not going to join the off-topic comments, I'm getting back on topic.

    I hope these rumors are true because they do signal a fantastic direction for the M-series SoCs. Multiple dies networked together with their high-speed bus (PCIe or home-designed) all with unified memory possibly sharing all memory and storage among all CPUs and GPUs. This is what I was hoping for. If Apple makes these multi-die SoCs with sockets for the Mac Pro that could be absolutely outrageous, giving Mac Pro users upgradeability. 

    I presume this means at least 20 CPUs and 64 GPUs in a M1 Max-Duo. If Apple can work out a socketed motherboard with empty SoC sockets, the Mac Pro could start with this Max-Duo and quickly become a Max-Quartet, Max-Sextet simply by plugging in a matching Max-Duo. There comes a point where fast is fast enough (not really) but as I've said before, there are scientist who like Macs and having a supercomputer on their desktop just for themselves would be great. I could also see this Mac Pro Max-Sextet being used by movie studios, producing animated features in 8K in real time. 
    I doubt “sockets” or “SOC upgradability” are words that will be associated with the new Mac Pro.

    To maintain very high bandwidth and low latency connections among all logic and memory, there needs to be very tight integration. For example, look at the picture here: https://www.anandtech.com/show/16921/intel-sapphire-rapids-nextgen-xeon-scalable-gets-a-tiling-upgrade. I’ll bet apple’s four die system looks more like that than, say, this: http://neconocone.cocolog-nifty.com/blog/files/genesis_brochure.pdf 

    robaba said:
    Sockets would expose the design to huge latency costs, I don’t think they will go that route.  I’m just not conversant enough with leading-edge ultra-high performance cpu design theory to guess what they might do instead.  Once they place their markers down, that will establish their patterns going forward—it’s in the next few iterations of Apple Silicon that we will begin to see what they are really capable of!  

    And of course, their competitors will not be standing still.  Intel is starting to put out innovative (though brutally hot) chips, and AMD is still on an up-cycle.  NVidia have purchased an Apple splinter group and may soon join the fracas on the RISC side, even if they are eventually blocked from purchase of ARM Holdings.

    It’s an exciting time in CPU design again!  Almost like the way the market looked prior to the Itanium debacle.
    Sockets don't need to be the traditional thousand pin sockets, they can be something new. I don't know what but Apple probably has something in the design stage that will allow multiple SoCs to be attached to the same unified memory architecture with low latency. Saw a video with two Apple designers talking about how many years ago Apple silicon started to be developed. The fact Apple is talking about multiple dies means they already have some well into development. The 400GB/s bus is something that renders external RAM and storage worthless so they have to have ideas on how to integrate multiple SoCs along with multiple dies. I would think there comes a point where multiple dies get too big but I have no doubt Apple will figure this out.

    As for anything Intel or AMD, they're way far behind Apple unless you want to use their CPUs to heat rooms. Apple's upside down design philosophy is something a CPU manufacturer can not replicate. Apple designs the device package first, figures out what software should run then designs the computer to make it work. Chip manufacturers can only design a chip and hope someone can make it work with their software in their computer package (laptop, desktop, phone). Nobody else does what Apple does. Find that video and you'll see what I mean.
    With respect to packaging and using multiple dies apple is not ahead. AMD has the most real products with multiple dies. They started with Epyc multi-chip modules. 

    AMD and apple both use TSMC and TSMC has a wide range of packaging options available. Apple has used some interesting options with Apple Watch (system in package). 

    I think it’s super unlikely that we will see anything that could be called a socket or user upgradable. It will be highly integrated to provide ultra high bandwidth and low latency.
    Like I said, AMD only makes CPU/GPU, they don't make the whole package allowing them to design like Apple is designing. You really need to find and watch that video to see what I am referring to. A huge motor doesn't mean anything in a car if the car isn't designed to use it from the ground up. Look for the RR videos. Don't want AI to ban me for referring to other sites.
    I don’t think AI bans you for pointing to other sources of information, unless your account only exists to do that. I assume you’re referring to the interview that the VPs Tom Boger and Tim Millet gave. So you could just link to AI’s own story about it, here:

    https://appleinsider.com/articles/21/11/01/apple-executives-discuss-how-the-m1-pro-m1-max-were-developed

    The Six Colors “Upgrade” (via Relay FM) podcast interview is titled “They Feed on Memory Bandwidth” — about an hour and quite thorough. They are well-rehearsed, and the interviewers aren’t getting anything out of them that Apple hasn't prepared them to say, but it’s still an illuminating picture that is very useful.
    rob53GG1argonaut
  • Apple promises macOS scanner fix in a future update

    I had this problem with a multifunction HP LaserJet 100 colorMFP M175. Resolved a few days ago using a Terminal command that did the same thing as the fix described in this article, thanks to an HP.com support thread. The answer came from a user, several HP employees were on there but only suggested things that didn’t work.

    Apple wasn’t much help either, basically just saying “Use AirPrint” — but that didn’t work for the scanner. Not sure if the bug is with regard to Apple’s implementation of AirPrint, or some kind of permissions bug. I’ll admit I had never heard of AirPrint before this.

    Good for Apple posting this support document, because it’s an easy fix. 

    *Editing to clarify that I’m not using AirPrint for the driver. I had to install the HP Printer Drivers v5.1 package to be able to set up the scanner:

    https://support.apple.com/kb/dl1888
    darkvaderwatto_cobra
  • 2022 Mac Pro said to use Intel Ice Lake Xeon W-3300 CPU

    The negative “trashcan” crowd here are engaging in a bit of selective memory with regard to what Apple has said about the problems with the 2013 cylindrical Mac Pro. 

    It’s instructive to go back and look at what was actually said in 2017, now with the added benefit of hindsight:

    https://daringfireball.net/2017/04/the_mac_pro_lives

    For example, the remarkable description of how a subset of Apple's “pro” users had been drifting away from the Mac Pro, toward increasingly-powerful iMacs. In addition, they talked about how the 2013 cylinder was an attempt to meet the needs of a further subset of Mac Pro users, for whom the regular iMac just wasn’t enough. The iMac Pro was also aimed at them, and this glimpse into Apple’s thinking at that time would argue for its rebirth into the Apple Silicon world. 

    They do admit pretty much outright that they failed the subset of Mac Pro users who want a modular system. Thus the current Mac Pro was born. But that doesn’t negate the reality or weight of the other two subsets Apple was catering to. Nothing in the 2017 mea culpa (or since) suggests Apple has given up on the users who liked the cylinder and/or the iMac Pro, at least in concept. Especially now in a world where the XDR display exists.
    Xedwilliamlondonwatto_cobra
  • Linksys AX4200 Wi-Fi 6 mesh routers gain support for Apple HomeKit

    These are us. Setup took a few minutes, as I had to use their serial numbers to manually identify which of our three nodes was in which room, but it was easy and basically nothing to it. The Linksys app just asked if I wanted to integrate with HomeKit, then sent me to the Home app.

    One thing I’m not sure about is parental controls over internet access for kids’ devices. I used to manage that from the Linksys app, and it looks like I still can, but I’m not sure if I’ve got additional options now, with regard to Screen Time and the like. If I learn anything interesting, I’ll post it here.
    Japheyright_said_fredFileMakerFellerwatto_cobra
  • Apple Car effort gains BMW electric car executive Ulrich Kranz

    dk49 said:
    Xed said:
    dk49 said:
    "It will take another 5 to 7 years to ship". I am afraid that's quite long, and if Apple finally launches the car in that time gap, it might be too late. Other car makers like Tesla would have evolved quite a lot. Even If Apple manages to release a Level 5 autonomy vehicle, it's highly likely that others like Tesla and Waymo would have done so already. And given that Tesla and other are increasingly focusing on the in-car entertainment and productivity, Apple's offering might not be as compelling as it seems right now. 

    Tim Cook always says, "we want to be the best, not the first", but if you enter a nascent market too late, the odds of failure increases. Homepod is the perfect example of that.
    Apple has a long history of entering established—non-nascent markets—with "experts" saying that it's too late, too radical, too expensive, etc., with a resultant owing of said market in terms of mindshare, direction, and profits.

    The HomePod isn't one of them at this time, but there are reasons why a Siri-controlled device that puts security first can't compete with the likes of Google and Amazon when customers have little regard about their own data privacy… and I say this as someone who has four creepy Echos in his home. Even Apple had to scale back HomeKit security to get 3rd-party vendors to adopt it.
    And what would be an example of Apple entering a non-nascent market too late but still making it big? 
    I can’t speak for the OP, but it seems to me the iPhone is the prime example of that.

    Smartphones were not new when it launched. Windows Mobile, Palm, BlackBerry, and Symbian phones all existed, and more. iPhone was derided as being far behind the competition—why? Because it didn’t allow third-party apps. The irony in that criticism, of course, is later, when Apple did introduce the iOS App Store, it crushed that competition almost literally overnight—mobile developers flocked to it because it wasn’t an expensive, pay-to-play scheme like all the others. You basically only paid Apple if you were successful. 
    tmaythtbyronlwatto_cobra