A number of media outlets in our neck of the woods have been raving about the arrival of Apple’s new Silicon M4 chip in the iPad Pro Oled. Some even went so far as to use Apple’s well-worn superlatives before the latter had even used them. However, others believe that Apple is marking time and that the M4 is merely a refresh of the M3, allowing Apple to take advantage of a more mature process at TSMC to increase profitability. Thus, it seems that the gain in performance, notably in IPC, is only 3% compared to the M3… Moreover, during Apple’s presentation, beyond the usual Apple superlatives, we read that the chip would use “second generation” 3 nm, that the M4 would integrate 28 million transistors, that the power of its “new” NPU would be 38 TOPS… An impressive performance of 50% compared to the M2. No comparison of the new chip with the M3 appeared during the presentation.
Geekbench comparisons of the M4 with the M3 Max
The contrast is striking between Apple’s presentation of its first M1 chip, which was a real revolution. If we have to make an objective comparison, the M4 has four performance cores and six efficient cores. That’s two more efficient cores than M3. This increase in the number of CPU cores accounts for a large part of the performance gains over the M3. Single-core processor performance hasn’t increased much with Apple’s M4 chip.
Apple outstripped by AMD, Intel and…Qualcomm?
If we have to talk about the GPU part, the M4’s seems to be largely similar to that of the M3 chip. Then there’s the AI trend, and here again, Apple isn’t exactly celebrating. By announcing an NPU delivering 38 TOPS, Apple has declared that its M4 chip is an outrageously powerful chip for AI. But the comparison with what’s on the market isn’t particularly flattering. Qualcomm’s new Snapdragon X Elite chip boasts a 45 TOPS NPU. As for AMD and Intel, both are expected to unveil new chips with faster NPUs at Computex, well ahead of this M4’s scores…Will we see a trend reversal in 2024?
