Can image sensors take a leap like Apple did with their own chips?

Started 5 months ago | Discussions thread
norjens Regular Member • Posts: 403
It's not such a massive leap, and ARM isn't universally better

JimKasson wrote:

fcracer wrote:

Not sure where else to put this post, but I figure people in the MF format will know the answer.

Apple recently launched their M1 chip for the low-end MacBook Air/Pro and Mini, and the results are spectacular. It's a step-change in performance vs. Intel and even AMD.

Does anyone here know if there is a step-change like that available for sensors or are we already at the highest levels of performance for mass produced sensors?

Intel has had one hand tied behind their backs with x86 ISA. Apple went with ARM, a newer ISA. The same thing could happen with image pipeline processors, but compatibility, which has held the x86 back for years, is not an issue there.

Intel has their hands tied behind their backs with internal problems, being stuck on their 10nm node and frequent leaks of bad management and unhappy workforce.

Looking at benchmark comparisons:

The title says it "crushes" last gen, but last gen is on an older manufacturing process than TSMC's 5nm that the M1 is on, and process shrink can typically account for 15-30% performance per watt improvement. In that context, a 25% faster CPU and ~15% slower GPU isn't so impressive from one generation to the next. Other benchmarks may show different results, and we don't know the power consumption details of the benchmarks, so ymmv.

AMD last month launched a ~25% performance jump from their 2019 generation of processors, still on the same 7nm node and x86. They don't seem to be hamstrung by x86, which makes sense because neither architecture is straight better than the other. Advantages and disadvantages.

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow