Not the parent, but: it isn't unfortunate that the M1 is so good; it's unfortunate that everything else is so far behind.

Is it that far behind? Benchmarks aren't everything, but it's probably as good as other metrics. A14 is ahead of SD888 in Geekbench (which has historically favored Apple CPUs), and behind on Antutu (which is more considerate to Snapdragons).

Geekbench 5:

- Single core: A14 1609, SD888 1127

- Multi core: A14 3872, SD888 3731

Antutu:

- A14 615796, SD888 723674

Benchmarks aren’t everything as you said. I used to have the last gen MAcBookPro with Intel for 3 years, and replaced it with M1 MacBook Pro. It’s almost an apple to apple comparison as they are same spec except the new processor. The difference is night and day. It just simply feels faster to use. It last almost 1.5 workdays of my use, while the intel one would not even last one full work day. I don’t need to lug around a power brick anymore. I charge at home and bring only the Mac to work without feeling anxious about battery. The cost I pay is that building x86 Docker images are quite slow on M1, but I don’t do it that often to bother me much.

> and behind on Antutu (which is more considerate to Snapdragons).

It is more considerate to Multi Core workload.

The main point is Single Core benchmarks. Which A14 set itself apart to all of its competitors.

Note that M1 is faster than A14:

Geekbench 5:

- Single core: 1744

- Multi core: 7676

The quoted comment referred only to battery life.

If Samsung just commits to outspend Apple on the node commitment, transistor counts, and size, then the outcome of the race may not be so pre-determined yet.

Transistor for transistor, ARMs latest licenseable cores are not so much far behind.

Last time I checked Samsung tablets and phones, they lagged already in the store. I don't know whether it's all the crapware that gets pre-installed, or just inefficient programming. But I think CPU speed might not help them.

Their Exynos are infamous for being slower, running hotter and using more power than equivalent Snapdragon and even Kirin (and of course, Apple SoCs).

But I doubt that shows in light usage like scrolling or opening an app, that's more likely their (lack of) software optimization.

My experience is that Exynos hw was great... but Samsung software (including drivers) was problematic.

Combine to with legacy of android design going for flexibility, including allowance for inefficient options (especially important in comparing graphics - Several Apple models ran on edge of being able to paint one frame without stutter assuming well-optimized code - Apple spent a lot of time ensuring you didn't see that) and you get certain reputation.

The only thing that I noticed really problematic is graphics intensive software optimized for Qualcomm.

Also, the real issue is not that Samsung doesn't have the capability. A lot of Apple "secret sauce" is that they don't have contractually separated design teams that have to "shop" around for suppliers/buyers, which means that both Qualcomm and Samsung are forced to make more mediocre CPUs because it brings wider selection of buyers - whereas Apple can design device with SoC together, which let's them easily take decisions like "ok, let's put A LOT MORE L1/L2 cache on each core" because they aren't going to deal with customers not wanting to buy them.

But Samsung has its own huge mobile business. Wouldn’t it be justified to make an OP SoC for their Galaxy Tab(or whatever they sell)? I’m sure their execs are aware of the possibility.

Supposedly they are forced to keep a "chinese wall" strategy due to possible legal concerns. How true that is hard to check, but there seems to be considerable variation and definitely less "design this cpu specifically for this phone" in Exynos line up, except maybe for first Samsung Galaxy S and SGS2.

P.S. Exynos and Apple A-series have common ancestry

Outspending Apple seems like a tall order, to put it mildly.

probably not when you're the second largest

You would need to not only outspend them, but at least match their engineering. Either one alone would be a tall order. But they should at least try, consumers will benefit even from a close miss.

Not OP, but Apple doesn't sell their chips to other companies. It's their right to do so, but if Apple sold M1, that'd be great for consumers.

M1 is that good because Apple customises their software for it(or the other way around). I doubt any other company would dedicate that much effort to it.

M1 is quite good at general-purpose workloads.