Sujet : Re: I Deleted Nemo :-)
De : recscuba_google (at) *nospam* huntzinger.com (-hh)
Groupes : comp.os.linux.advocacyDate : 15. Nov 2024, 02:19:29
Autres entêtes
Organisation : A noiseless patient Spider
Message-ID : <vh67j1$2q1pl$1@dont-email.me>
References : 1 2 3 4 5 6 7 8 9 10
User-Agent : Mozilla Thunderbird
On 11/14/24 7:00 PM, Lawrence D'Oliveiro wrote:
On Thu, 14 Nov 2024 10:21:42 -0500, -hh wrote:
Today, Apple's M- architecture employs a unified memory configuration.
The tight integration significantly improves performance, but the
trade-off is that it isn't "old school" modular anymore. I guess that
if you really wanted a RAM upgrade, you could swap out the whole CPU
package.
Which is why it’s an evolutionary dead-end. A friend suggested to me,
around the time of the M1 chips, that the rest of the PC industry was
going to follow Apple’s lead, but Intel has already admitted it was a
mistake doing so, and is going back to modular memory.
It really depends on system level trades, and lifecycle contexts.
For example, the need for supporting incremental upgrades after deployment depends on the expected lifespan for which there will be software updates that could outgrow the original hardware, so when there's not much change in software across its useful lifespan, then there's not much need to support hardware changes (eg, upgrade RAM).
At that point, other value metrics such as lower manufacturing costs, higher reliability, etc, can prioritize solutions with lower parts counts and fewer discrete points of failure. Adding a modular connector for RAM is an increase in parts count and decreases reliability.
-hh