LOL. If “swap” is faster than RAM, it’s not a swap file. It’s called L1/L2 cache, and the M1 has 12MB L2 (for performance cores) of it.
It could be a while. From what ive read, it could be until 2nd half of 2021.
Well the country I live in has 3 Apple Stores, which is already a significant improvement from 4 years ago when there were 0 - the Apple Store app used to tell me my nearest Apple Store was somewhere over a 4 hour flight away!
Or just bury it in cold storage until those became collectibles. Just like the powerbooks.
Of course that’s when they started to experiment running their own chips (T series chip on those models) are essentially exercise in preparation of the M1 macs. First touchID in T1. In T2 they had the chip took over part of the boot process and bypassing intel PCH in addition to taking over SSD controls/encryption.
for collection purposes.
This also means I can nab them T2 macs for cheap now probably. I’d probably nab one of the 2019 16” with 5600M once price is low enough
Not all of us do or did at point asking. This is why me going from a 2010 Mac Pro 8GB of RAM to a 2019 i9 iMac with 128GB of RAM and the most taxing thing I do is 1080p video editing. I got 128GB of RAM from recommendations when it made literally 20 seconds difference on a 15+ export. Yay I guess? I am not Hollywood so the extra 20 seconds is useless to me. And that 20 seconds might have just been an error in statistical analysis since I did not do dozens of tests. It could be my mac was in the middle of something else when I first tested that caused my 8GB version to be 20 seconds slower.
Not necessarily. NSObject retain/release is much much MUCH faster on Apple Silicon. Which means memory transactions are faster, which means clearing out old memory and reserving new memory when needed is faster. Which can lead to a faster system overall.
I can’t now, its priced too high to make sense right now. I just also canceled my M1 Air return once I realized that it could be until 2nd half of next year until the 16” M1X arrives. Well that and a Razer blade 2019 Silver edition with 2070 max-Q I just nabbed for half off. (Its even cheaper than my M1) with plans to mostly game on it (Cyberpunk among other things) and possibly hackintosh and experiment with. As hackintoshes are far more flexible in this regard. (Wanted to learn more about the underpinnings of OS X/ macOS)
Someone mentioned (I think it was Max Tech) that the prediction is 8/16GB maybe 32GB on the SOC itself, and some DIMMs for getting to the 1.5TB. That way some OS level/GPU level items needing the very fast RAM access can use the SOC RAM, and things like programs can use the extra DIMMs memory. Will be interesting to see. It would have to be the case, I don't think they can release an SOC with 1.5TB of RAM on the Mac Pro.
1 and 3 are already addressed technically. With the M1 matching/beating a GTX 1050Ti and the fans are barely running most of the time. Upgradeable RAM is all that is left. The Intel Mini had replaceable RAM but the M1 does not, and probably won't.
I believe DRAM still uses one transistor per bit (and some overhead for logic), so 1.5 TB would need ~13 trillion transistors. That's why the Mac Pro has 12 DIMM slots, and each DIMM can have tens of memory chips.
It does not mean that. Retain/release has absolutely nothing to do with malloc/free. It is merely a matter of accounting which objects want or no longer want a strong reference to another object. It happens with great frequency in Cocoa, but a retain follows an allocation that has already happened, and a release merely decrements the retain count, which may or may not result in memory being deallocated. Please fail to overstate the situation.
