Unified memory has pros and cons. However there is no need to solder the RAM to the board for it to be unified. My key point though is that Apple is in fact using off the shelf memory and storage. The chips are a standard design with standard performance.
The only specific configurations that Apple should need to worry about are the same ones they were concerned with when they didn’t force customers to buy upgraded memory and storage from them.
You claim that, but would you care to point out how they’re supposed to get 400 GB/s memory b/w across a socketed system? I know of no PC (let alone consumer grade) that gets anywhere near that type of memory b/w with their reliance on (SO-)DIMM slots.
I see this not so much as ill will (although I have no doubt Apple loves forcing people to pay their ludicrous markup), but technical necessity to achieve certain performance levels.
I have a similar Mac-owning history to yours, but I’ve had a couple of Macs just bite the dust over these past 40 years.
My Mac SE no longer boots. I kept it as one of the last models with the original developer team signatures embossed inside the case. I also kept the special long-handled hex wrench and “Mac cracker” clamp that lets me in to see that if I ever feel like it.
My daughter’s 2000-vintage purple iMac suffered an electrical short and failed. It was possibly an electrical surge caused by a lightning strike—I found chips with little holes and white marks on them. It was essentially unrepairable.
And, an iMac 24-inch (one of the early Intel Core 2 Duo models from 2006) had what I think was a leaking capacitor failure that was so catastrophic that when I brought it to the Apple Store, they handed it back to me a week later with the hard drive pulled at my request and said “No charge.”
I balance that against the many other Macs I’ve had or used over the years (everything from an Quadra and a Mac Portable to my current setup) which are all reliable and functioning, some of them a full decade after initial purchase. I have a Wall Street Powerbook in my closet that has broken display hinges, but would still boot up if I chose.
These are all data points on the reliability curve. Who’s to say if a SSD can last 5 years? No matter how carefully they are manufactured, subtle defects will always find their way out into the world. The more tightly hardware is integrated, the more likely a single component will require at very least a radical transplant, and possibly a complete equipment replacement.
I agree with your main point, but I just hate to put the expectation out there because it just fuels influencers like Linus.
I guess I was lucky with my Macs. Between my lab and home I used to have at least 5 Macs running 24/7 and I had only one failure: a G5 tower (perforated case) in my lab whose power supply failed (which I replaced with some trouble). Sorry, I forgot about that when I claimed to have had no fatal failures. I had one Intel 13" MBP (circa 2013) repeatedly freeze on an approximately monthly basis - my wife took over this machine about 5 years ago and it has run flawlessly since then. I am still not sure whether it was a hardware or OS failure (or combination). Macs are impressively well made and reliable. Like the NeXTs which I used in my lab after they were discontinued - we got them from surplus for <$100 and the OS and hardware were bulletproof.
No flyback transformer failure in a 128k Macintosh? Interesting. Also, somewhat unlikely. And I’d love to hear engineering details on a home-brew Fat Mac (512k). The upgrade cost me at least a kilobuck.
IIRC, it wasn’t much more than carefully unsoldering 16 64 kb RAM chips (in DIP packages) and replacing them with 256 kb parts. Where I was working at the time, we had recently been through a RAM chip shortage. It almost killed me to casually toss 128 kB of memory in the trash, but it was worth it to have that HUGE amount of RAM in my Mac.
Having access to the physics department electronics shop was very useful. We “fattened” about a dozen 128k Macs using the shop’s professional solder sucker to unsolder the 16 ram chips (the ends of the pins were bent making them difficult to remove). I used MacDraw to design a small daughter board for the multiplexor (we were following the design in Dr. Dobbs) and we had about 20 small circuit boards made. All the conversions worked well, though there was an occasional trace that was accidentally cut but was easily repaired. The funny thing is that shortly after I sold my machine the flyback failure occurred - we fixed it but the buyer was a bit annoyed, understandably (it worked perfectly when I sold it).
I have made a number of “FrankenMacs” using cheap motherboards (a MacPlus motherboard could be bought for $100) and then installing add-ons like a 68020 or 68030 accelerator from Novy Corp. I mounted the board in the air vertically for cooling and used cheap surplus monitors (12") which accepted the high sweep rates of the Mac. The power supplies were cheap analog units and the HDs were purchased and placed in a box. The things were a bit ugly but worked very well.
Those were the days!
All I can say is wow. And no flyback transformer in sight. Double wow. Especially the part about “hard drives” which did not exist in 1984. As did not “MacPlus”.See ya.
These were done in the late ‘80s. The monitors of course had flyback transformers which were more reliable than the ones in the original Mac. Of course the ‘020 and ‘030 didn’t appear till then also. It was nice making a Mac which was as powerful but much cheaper than the Mac II. (Or the SE30). The HD was a Rodime (Scottish) which failed but I fortunately had a backup (40 floppies).
I’m a fan of Gamer’s Nexus… and for good reason…seems Linus is in the news and not in a positive light.