.... They compare their 12-core X Elite using 80W with M2/M3 with 8 cores using around 23W. A ”fair” comparison would be to use 12-core M3 Pro using only 24W. So again no, Apple doesn’t need M4 to ”rise to the challenge”.
View attachment 2367598
(
Source: Just Josh on YouTube)
Do you even read the graph your are posting? Line
one above is a 23W X Elite and the closet M3 variant to that is a the "23W M3". So yes, they are comparing those two. There is actually a horizontal line
separating the the MBP 14"'s from the 80W mode X Elite. So they are trying to compare across the barrier line they put in the graph? Errr Probably not.
Apple built TWO different dies; the plain M3 and the M3 Pro. Qualcomm built one that sits in-between those two. They are not trying to exactly what Apple did. Nor should they ( on the first iteration). Apple built the A__X variants for
YEARS before essentially slapping a Mn name on them ( and moving on to addition die sizes).
There is two major modes they can set the die/package to. One that is plays in roughly the same zone as the plain Mn SoC from Apple. And competes very handily with the actual real competitors of AMD/Intel with a 23W constraint. The other Windows/Linux laptop competitive threat is "repurposed desktop SoC" in a laptop. Those chassis typical have thermal capacity that handily accepts 80W. Getting the SoC placed into those chassis won't be hard. And they absolutely do not have to compete for placement in Mac chassis at all.
The 80W X Elite is up-clocked into a zone where really start to push past where the fab process's sweet spot on efficiency. Apple tends to just stop before the efficient curve goes bad ( Perf/Watt being the primary target metric) and simply makes a bigger die to walk up the multicore performance ladder. Qualcomm is trying to 'kill two birds with one stone'. Since their primary competitors push their chips into the diminishing returns zone (especially Intel), it isn't going to 'hurt' Qualcomm much to do the same exact thing to gather some incremental additional system placements. There is at least one , if not two, orders of magnitude more possible laptop chassis to be placed into than what exists in the Mac line up ( as if that was the entire possible area of competition. ). X Elite has to be selected and bought by system vendors for the line up to survive to additional generations. Qualcomm needs some positive cash flow here before they go to multiple dies in the SoC line up. ( buying Nuvia just gives them a $1B hole to at least partially crawl out of first).
Is Qualcomm going to dominate the 90+ W laptop market? Probably not since don't appear to do dGPUs , but decent chance they can get a few placements there as an 'alternative' SoC configuration. Nvidia isn't going to get banned from Windows/Linux laptops any time soon. So it isn't the Mac chassis restrictions that Qualcomm really has to compete with over the long term. In second , or third, iteration will probably see Qualcomm 'adapt' to that issue (probably only on 'bigger' die variants. Qualcomm is probably going toward phones/tablets smaller too; so dGPU ability is very likely not going to be ubiquitous in their line up. ). However, Qualcomm probalby is not going to feel same pressure that Apple has by banning dGPU, so unlikely going to see "Max like" die area commitment for GPU subsystem resources from Qualcomm. Scaling down to phone SoC like allocations is likely going to have higher priority than going 'max area'.
There is nothing suggestive in those charts that indicate that Qualcomm cannot keep efficiency is made more appropriate sized dies to walk deeper into the Pro/Max CPU compute zone. There is no big show stopper present.
Qualcomm can continue to make the entry X Elite is bit bigger than a plain M3 because Android tablet ( iPad Pro ) isn't going to pull the same volume. The bigger die could end up in-between the Pro and Max targets also. It makes little sense to try to land exactly on top of the Apple SoCs when the Qualcomm SoCs are not going into the exact same set of chassis constraints.
