At this rate, it will take Samsung several more years to simply catch up with the iPhone 13. They have not been able to break out of their performance curve. This is alarming because competition is good for the industry as a whole (including Apple) but other ARM makers are falling behind badly. Hopefully NVIDIA can work some magic with their ARM processors.
“Competition is good” is a myth. When Apple invented the iPhone they didn’t think of making a better Blackberry or a skinnier flip phone. Copying does NOT create innovation. Innovation creates innovation.
Samsung DOES NOT push Apple to make a faster iPhone 14. Apple and it’s needs push Apple to make a faster iPhone 14. Apple Watch and AirPods are a testament to that.
Like that Ford quote ”If I asked what people wanted they would have said faster horses.”
True but Samsung is currently failing to compete with Apple hence my concern. If Apple and Samsung were similar in performance, competition would create innovation. As it stands Apple has a monopoly on high performance ARM CPUs. As to why this is happening, if I had to guess I would say that the Israel team came up with some new technology and Apple owns the patents.
This Geekbench chart only proves what a crappy benchmarking tool Geekbench truly is. Especially when many ignorant users actually think those scores work across different platforms, and different hardware. GPU's on Android devices use both OpenGL, and Vulkan. Where as all iOS apps now use Apples Metal APIs. Not comparable at all. It use to be comparable when Apple and their developers could still create OpenGL apps, but Apple hasn't supported OpenGL for many, many years now. Then those graphic scores mean nothing across other platforms.
Also GeekBench ML scores are complete nonsense, especially when you can go to the Geekbench ML browser pages and look up the scores for lots of devices. I love how the iPhone 7 (released 2016) ML score is 354, and the Pixel 6 (released 2021) ML score is 309. The Pixel 6 Tensor SoC has dedicated AI hardware, yet the iPhone 7 gets a higher ML score, even though it has NO AI hardware on board with its A10 SoC. So how can that be? The S22 has a Snapdragon 8 gen 1 SoC and its embedded AI hardware (Hexagon) is rated at 52 TOPS, and for some other ML tasks it can go up to 104 TOPS. Yet the iPhone 13 A15 SoC embedded AI hardware with its Neural Engine is rated 15.8 TOPS. Yet Geekbench is saying that the iPhone 13 ML score is more than twice as fast as the S22 ML score? How can that be? If those scores were real across other platforms, then Qualcomm, Samsung, Google, and any other SoC OEM could be brought to court for false advertising of their specifications. However I know for a fact that my Android AI application which uses Object Detection is much faster on a number of Android devices, compared to any iOS device that I have tested. So clearly Geekbench ML scores mean nothing.
Comments
Also GeekBench ML scores are complete nonsense, especially when you can go to the Geekbench ML browser pages and look up the scores for lots of devices. I love how the iPhone 7 (released 2016) ML score is 354, and the Pixel 6 (released 2021) ML score is 309. The Pixel 6 Tensor SoC has dedicated AI hardware, yet the iPhone 7 gets a higher ML score, even though it has NO AI hardware on board with its A10 SoC. So how can that be? The S22 has a Snapdragon 8 gen 1 SoC and its embedded AI hardware (Hexagon) is rated at 52 TOPS, and for some other ML tasks it can go up to 104 TOPS. Yet the iPhone 13 A15 SoC embedded AI hardware with its Neural Engine is rated 15.8 TOPS. Yet Geekbench is saying that the iPhone 13 ML score is more than twice as fast as the S22 ML score? How can that be? If those scores were real across other platforms, then Qualcomm, Samsung, Google, and any other SoC OEM could be brought to court for false advertising of their specifications. However I know for a fact that my Android AI application which uses Object Detection is much faster on a number of Android devices, compared to any iOS device that I have tested. So clearly Geekbench ML scores mean nothing.