Apple's M3 Max Outshines RTX 4090 in AI Speech Transcription Test

The M3 Max is Apple's most powerful 3nm SoC developed for the Mac, with multi-core performance that outperforms the M2 Ultra and a 40-core GPU that can compete with a laptop RTX 4080. Despite these advancements, there is still a gap between laptop GPUs and NVIDIA's 'top of the line' RTX 4090, but in the newest voice transcribing test, the M3 Max outperforms the world's fastest consumer graphics processor, given the correct test circumstances.

Apple's M3 Max Outperforms an RTX 4090 in OpenAI's Whisper Speech Transcribing Test

Apple has introduced MLX, an open-source framework created exclusively for the company's M-series of chipsets, according to WCCFTech. This framework may be used by developers to create, test, and enhance their applications, as Oliver Wehrens did with several Macs and an RTX 4090 GPU and gave benchmark results below. Whisper, OpenAI's voice recognition model, is used in the MLX framework, and Wehrens undertook a transcription test to check how quickly various processors performed.

The initial test results using a 10-minute audio file suggest that the M2 Ultra and M3 Max are unexpectedly quicker than the desktop RTX 4090, although this is also due to the use of an unoptimized program to carry out the benchmark, as seen in the graph. Wehrens also released data using the optimized tool, and the results are much as expected: no competition for the RTX 4090, which blows away the M2 Ultra, M3 Max, and the remaining SoCs.

In the refresh test, the RTX 4090 completed the audio transcription test in under 8 seconds, while the M2 Ultra took 95 seconds and the M3 Max took 100 seconds. In terms of raw performance, the new findings demonstrate that Apple's M-series family has a long way to go. Power efficiency is one area where the custom silicon range shines, as it uses far less power than NVIDIA's flagship consumer-grade graphics card.

At the end of the test, Wehrens states that the results are not meant to be scientific measurements, but they do give some insight into how capable Apple's MLX framework is. Because it is designed specifically for Apple silicon, a MacBook Pro equipped with the M1 Pro is as fast as an RTX 4090, assuming no NVIDIA-optimized tools are used to perform the tests.

Apple’s Fastest and Most Efficient CPU By Far

To recall, in October, Apple announced the M3, M3 Pro, and M3 Max CPUs, which feature game-changing innovations that provide drastically enhanced performance and unlock new Mac possibilities. These are the first personal computer chips made using the industry-leading 3-nanometer manufacturing technology, which allows for more transistors to be crammed into a smaller space while boosting speed and efficiency. Together, the M3, M3 Pro, and M3 Max demonstrate how far Apple silicon for the Mac has progressed since the introduction of the M1 series of CPUs.

The M3 chip series has a next-generation GPU, which marks the most significant advancement in graphics architecture for Apple hardware to date. The GPU is faster and more efficient, and it includes a new technique known as dynamic caching as well as new rendering technologies like hardware-accelerated ray tracing and mesh shading for the first time on Mac. Rendering rates have increased up to 2.5x over the M1 series of CPUs.

The CPU performance and efficiency cores are 30% and 50% faster, respectively, than those in M1, while the neural engine is 60% faster than the neural engine in the M1 series of CPUs. A new media engine now supports AV1 decode capability, allowing streaming services to provide more efficient and high-quality video experiences. The M3 processor family maintains Apple's incredible rate of silicon innovation, bringing major advancements and new capabilities to the new MacBook Pro and iMac.

“Apple silicon has completely redefined the Mac experience. Every aspect of its architecture is designed for performance and power efficiency. With 3-nanometer technology, a next-generation GPU architecture, a higher-performance CPU, faster Neural Engine, and support for even more unified memory, the M3, M3 Pro, and M3 Max are the most advanced chips ever built for a personal computer,” Johny Srouji, Apple’s senior vice president of Hardware Technologies, stated.

The performance and efficiency cores of the next-generation CPU in the M3, M3 Pro, and M3 Max have been architecturally improved. The performance cores are up to 30% quicker than those in the M1 family, allowing for speedier operations like generating and testing millions of lines of code in Xcode, enabling artists to use hundreds of audio tracks, plug-ins, and virtual instruments in Logic Pro.

The efficiency cores are up to 50% quicker than the efficiency cores in M1, allowing the system to maximize battery life while performing routine activities faster than before. These cores work together to build a CPU that gives the same multithreaded performance as M1 while utilizing half the power and up to 35% higher performance at peak power.

Photo: Hostaphoto/Unsplash