Hacker News Reader

Comment by traceroute66 | Hacker News Reader

"Up to 4x faster than Apple M1 Max" they say ...

Aah, that famous marketing "up to" ! So I guess that's in certain cherry-picked scenarios then ?

Also what they don't tell you is that the Apple will happily do ML for hours on end ON BATTERY with zero performance impact. Whilst an x86 laptop will need to be permanently tethered to your nearest power socket in order to obtain top performance. Otherwise you will suffer a performance drop and kill off your battery life.

I know the usual Apple bashers gonna hate, but fact of life its that for most people Apple have done a sterling job on their Apple Silicon laptops. No doubt more to come with the M2 in due course.

View on HN

Well, I am a big proponent for Apple products, but for ML-related work Apple Silicon is not the best good value for money currently. This will likely change in the future, when we have larger matrix accelerators and ML-specialized data types in the GPU/AMX, but Apple is not there yet.

>ON BATTERY

Why would I even care about that for ML?

>I know the usual Apple bashers

Apple fanboys found one benchmarks which looks fine (Performance per watt) and decided it applies to everything and the only thing that ever matters - even when it's totally irrelevant.

It used to be they cared about Mhz, until it became 'the Mhz myth' when Apple was losing.

Sadly Apple Silicon is still very much difficult to properly configure for some areas of ML whilst leveraging the GPU (object detection for example), so having that nvidia really makes the set up smoother.

Wouldn't really want to do heavy computational work on any laptop. Thermal throttling will gut your performance. Laptops just don't have the cooling necessary to have it any other way.

But that's exactly where newer Apple laptops shine. They don't have the throttling issues associated with x86 simply because they use 2-4 times less power for the same level of performance. E.g. my M1 Max CPU maxes out at 5-6 watts per core, while delivering performance comparable to that of an Intel P-core running at 4.5-4.8 ghz. In multicore, that's 40 watts for the level of performance where Intel or AMD would somewhere between 80 and 150 watts, depending on the configuration. And 40 watts is something that a laptop can easily dissipate.

The efficiency of Apple silicon is a matter of fact now, however isn't nVidia with its cuda still king in this segment? Please correct me if I'm wrong but doing ml/dl on CPU instead of GPU seems to be the least efficient way to go about it?

> nVidia with its cuda still king in this segment

I suspect this will gradually change, perhaps especially now a lot of effort has been made to bring tooling such as PyTorch over to Apple silicon.

> on CPU instead of GPU

But Apple isn't doing it on CPU.

You are thinking in terms of x86 discrete components.

Apple Silicon is a fully integrated architecture including unified memory. That's what makes it so efficient.

Yes, Nvidia still obliterates M1/M2 in Deep Learning. M1 is close to GTX1650 in real-world DL workloads though in theory based on TFlops it should be around GTX1070.

Note that I was talking about the CPU specifically. GPU on Apple is also more efficient (approx 0.25TFLOPs/watt for M1 series), but Apple GPUs lack support for ML-optimized FP representation (primary reason why Nvidia is so good in this domain). Apple does have a matrix coprocessor which offers excellent performance/watt for inference, but these units are relatively small and only offer limited aggregated performance.

I think it’s just a question of time until Apple offers hardware support for BFLOAT and other formats on the GPU and AMX (they already have BFLOAT16 in the CPU), at which point their ML performance will improve dramatically.

> Apple will happily do ML for hours on end ON BATTERY

At 1/10th of the speed...

> At 1/10th of the speed...

Yawn.

As has already been clearly explained numerous times in this discussion, e.g. @ribit[1] below, there is no throttling on Apple Silicon when running on battery.

[1] https://news.ycombinator.com/item?id=34159456

It's not about throttling, it's about mobile 3080Ti vs M1 GPU performance. M1 hits at around GTX1650 for Deep Learning workloads in GPU benchmarks.

> there is no throttling on Apple Silicon when running on battery

... yet. The computational demand of typical work will increase to eat up the extra capacity, as it always does - and the only way for a laptop to handle the same workload plugged or unplugged is to cap its plugged-in performance.

Reply on news.ycombinator.com