3 ms·

I didn't see any units for their "the GPU has 2.6 TFLOP/s of throughput". 2.6 TFLOPS/s at what precision? FP16, FP32, something else? Nvidia labels their Tegra

I didn't see any units for their "the GPU has 2.6 TFLOP/s of throughput". 2.6 TFLOPS/s at what precision? FP16, FP32, something else? Nvidia labels their Tegra AGX Xavier part as 5.5 TFLOP/s (FP16) and AMD's APUs say 1.8 TFLOP/s (FP32). They also say "up to" 3.5x CPU performance (total performance) of previous generation macs, then make references to a Mac with a 1.2 GHz quad core Core i7 (MacBook Air) and a quad core 3.6 GHz Core-i3 (Mac Mini). I would like to see some independent benchmarks personally...

eugeniub 6y ago
If you go to the MacBook Pro 13" page (https://www.apple.com/macbook-pro-13/ https://www.apple.com/macbook-pro-13/) and click on "See how M1 redefines speed" there are some more numbers for specific use cases (Xcode project build, Final Cut Pro ProRes transcode, etc). The footnotes for these numbers say the tests were done with the maxed-out previous-generation 2-port MacBook Pro: > Testing conducted by Apple in October 2020 using preproduction 13-inch MacBook Pro systems with Apple M1 chip, as well as production 1.7GHz quad-core Intel Core i7-based 13-inch MacBook Pro systems, all configured with 16GB RAM and 2TB SSD. I think it's a fair comparison, since the new M1 MacBook Pro 13" is a 2-port model that replaces the existing 2-port models, while Apple still sells Intel 4-port models. Similar situation on the MacBook Air page. The tests are done with the maxed-out previous-generation MacBook Air: > Testing conducted by Apple in October 2020 using preproduction MacBook Air systems with Apple M1 chip and 8-core GPU, as well as production 1.2GHz quad-core Intel Core i7-based MacBook Air systems, all configured with 16GB RAM and 2TB SSD.