Huge Memory Bandwidth, but not for every Block

One highly intriguing aspect of the M1 Max, maybe less so for the M1 Pro, is the massive memory bandwidth that is available for the SoC.

Apple was keen to market their 400GB/s figure during the launch, but this number is so wild and out there that there’s just a lot of questions left open as to how the chip is able to take advantage of this kind of bandwidth, so it’s one of the first things to investigate.

Starting off with our memory latency tests, the new M1 Max changes system memory behaviour quite significantly compared to what we’ve seen on the M1. On the core and L2 side of things, there haven’t been any changes and we consequently don’t see much alterations in terms of the results – it’s still a 3.2GHz peak core with 128KB of L1D at 3 cycles load-load latencies, and a 12MB L2 cache.

Where things are quite different is when we enter the system cache, instead of 8MB, on the M1 Max it’s now 48MB large, and also a lot more noticeable in the latency graph. While being much larger, it’s also evidently slower than the M1 SLC – the exact figures here depend on access pattern, but even the linear chain access shows that data has to travel a longer distance than the M1 and corresponding A-chips.

DRAM latency, even though on paper is faster for the M1 Max in terms of frequency on bandwidth, goes up this generation. At a 128MB comparable test depth, the new chip is roughly 15ns slower. The larger SLCs, more complex chip fabric, as well as possible worse timings on the part of the new LPDDR5 memory all could add to the regression we’re seeing here. In practical terms, because the SLC is so much bigger this generation, workloads latencies should still be lower for the M1 Max due to the higher cache hit rates, so performance shouldn’t regress.

A lot of people in the HPC audience were extremely intrigued to see a chip with such massive bandwidth – not because they care about GPU or other offload engines of the SoC, but because the possibility of the CPUs being able to have access to such immense bandwidth, something that otherwise is only possible to achieve on larger server-class CPUs that cost a multitude of what the new MacBook Pros are sold at. It was also one of the first things I tested out – to see exactly just how much bandwidth the CPU cores have access to.

Unfortunately, the news here isn’t the best case-scenario that we hoped for, as the M1 Max isn’t able to fully saturate the SoC bandwidth from just the CPU side;

From a single core perspective, meaning from a single software thread, things are quite impressive for the chip, as it’s able to stress the memory fabric to up to 102GB/s. This is extremely impressive and outperforms any other design in the industry by multiple factors, we had already noted that the M1 chip was able to fully saturate its memory bandwidth with a single core and that the bottleneck had been on the DRAM itself. On the M1 Max, it seems that we’re hitting the limit of what a core can do – or more precisely, a limit to what the CPU cluster can do.

The little hump between 12MB and 64MB should be the SLC of 48MB in size, the reduction in BW at the 12MB figure signals that the core is somehow limited in bandwidth when evicting cache lines back to the upper memory system. Our test here consists of reading, modifying, and writing back cache lines, with a 1:1 R/W ratio.

Going from 1 core/threads to 2, what the system is actually doing is spreading the workload across the two performance clusters of the SoC, so both threads are on their own cluster and have full access to the 12MB of L2. The “hump” after 12MB reduces in size, ending earlier now at +24MB, which makes sense as the 48MB SLC is now shared amongst two cores. Bandwidth here increases to 186GB/s.

Adding a third thread there’s a bit of an imbalance across the clusters, DRAM bandwidth goes to 204GB/s, but a fourth thread lands us at 224GB/s and this appears to be the limit on the SoC fabric that the CPUs are able to achieve, as adding additional cores and threads beyond this point does not increase the bandwidth to DRAM at all. It’s only when the E-cores, which are in their own cluster, are added in, when the bandwidth is able to jump up again, to a maximum of 243GB/s.

While 243GB/s is massive, and overshadows any other design in the industry, it’s still quite far from the 409GB/s the chip is capable of. More importantly for the M1 Max, it’s only slightly higher than the 204GB/s limit of the M1 Pro, so from a CPU-only workload perspective, it doesn’t appear to make sense to get the Max if one is focused just on CPU bandwidth.

That begs the question, why does the M1 Max have such massive bandwidth? The GPU naturally comes to mind, however in my testing, I’ve had extreme trouble to find workloads that would stress the GPU sufficiently to take advantage of the available bandwidth. Granted, this is also an issue of lacking workloads, but for actual 3D rendering and benchmarks, I haven’t seen the GPU use more than 90GB/s (measured via system performance counters). While I’m sure there’s some productivity workload out there where the GPU is able to stretch its legs, we haven’t been able to identify them yet.

That leaves everything else which is on the SoC, media engine, NPU, and just workloads that would simply stress all parts of the chip at the same time. The new media engine on the M1 Pro and Max are now able to decode and encode ProRes RAW formats, the above clip is a 5K 12bit sample with a bitrate of 1.59Gbps, and the M1 Max is not only able to play it back in real-time, it’s able to do it at multiple times the speed, with seamless immediate seeking. Doing the same thing on my 5900X machine results in single-digit frames. The SoC DRAM bandwidth while seeking around was at around 40-50GB/s – I imagine that workloads that stress CPU, GPU, media engines all at the same time would be able to take advantage of the full system memory bandwidth, and allow the M1 Max to stretch its legs and differentiate itself more from the M1 Pro and other systems.

M1 Pro & M1 Max: Performance Laptop Chips Power Behaviour: No Real TDP, but Wide Range
Comments Locked


View All Comments

  • Speedfriend - Tuesday, October 26, 2021 - link

    This isn't their first attempt. They have been building laptop version of the A series chips for years now for testing. There have been leaks about this for years. Assuming that the world best SOC design team will make a significant advancement from here after 10 years of progress on A series is hoping for a bit much
  • robotManThingy - Tuesday, October 26, 2021 - link

    All of the games are x86 translated by Apple's Rosetta, which means they are meaningless when it come to determining the speed of the M1 Max or any other M1 chip.
  • TheinsanegamerN - Tuesday, October 26, 2021 - link

    Real-world software isnt worthless.
  • AshlayW - Tuesday, October 26, 2021 - link

    "The M1X is slightly slower than the RTX-3080, at least on-paper and in synthetic benchmarks."
    Not quite, it matches the 3080 in mobile-focused synthetics where Apple is focusing on pretending to have best-in-class performance, and then its true colours shows in actual video gaming. This GPU is for content creators (where it's excellent) but you don't just out-muscle decades of GPU IP optimisation for gaming in hardware and software that AMD/NVIDIA have. Furthermore, the M1MAX is significantly weaker in GPU resources than the GA104 chip in the mobile 3080, which here, is actually limited to quite low clock speeds, it is no surprise it is faster in actual games, by a lot.
  • TheinsanegamerN - Tuesday, October 26, 2021 - link

    Rarely do synthetics ever line up with real word performance, especially in games. MatcHong 3060 mobile performance is already pretty good.
  • NPPraxis - Tuesday, October 26, 2021 - link

    Where are you seeing "actual gaming performance" benchmarks that you can compare? There's very few AAA games available for Mac to begin with; most of the ones that do exist are running under Rosetta 2 or not using Metal; and Windows games using VMs or WINE + Rosetta 2 has massive overhead.

    The number of actual games running is tiny and basically the only benchmark I've seen is Shadow of the Tomb Raider. I need a higher sample size to state anything definitively.

    That said, I wouldn't be shocked if you're right, Apple has always targeted Workstation GPU buyers more than gaming GPU buyers.
  • GigaFlopped - Tuesday, October 26, 2021 - link

    The games tested were already ported over to the Metal API, it was only the CPU side that was emulated, we've seen emulated benchmarks before, the M1 and Rosetta does a pretty decent job at it and when they ran the games at 4k, that would have pretty much removed any potential bottleneck. So what you see is pretty much what you'll get in terms of real-world rasterization performance, they might squeeze an extra 5% or so out of it, but don't expect any miracles, it's an RTX 3060 Mobile competitor in terms of Rasterization, which is certainly not to be sniffed at and very good achievement. The fact that it can match the 3060 whilst consuming less power is a feat of its own, considering this is Apple first real attempt at desktop level or performance GPU.
  • lilkwarrior - Friday, November 5, 2021 - link

    These M1 chips aren't appropriate for serious AAA Gaming. They don't even have hardware-accelerated ray-tracing and other core DX12U/Vulkan tech for current-gen games coming up moving forward. Want to preview that? Play Metro Exodus: Enhanced Edition.
  • OrphanSource - Thursday, May 26, 2022 - link

    you 'premium gaming' encephalitics are the scum of the GD earth. Oh, you can only play your AAA money pit cash grabs at 108 fps instead of 145fps at FOURTEEN FORTY PEE on HIGH QUALITY SETTING? OMG, IT"S AS BAD AS THE RTX 3060? THE OBJECTIVELY MOST COST/FRAME EFFECTIVE GRAPHICS CARD OF 2021??? WOW THAT SOUNDS FUCKING AMAZING!

    Wait, no I, misunderstood, you are saying that's a bad thing? Oh you poor, old, blind, incontinent man... well, at least I THINK you are blind if you need 2k resolution at well over 100fps across the most graphics intensive games of 2020/2021 to see what's going on clearly enough to EVEN REMOTELY enjoy the $75 drug you pay for (the incontinence I assume because you 1. clearly wouldn't give a sh*t about these top end, graphics obsessed metrics and 2. have literally nothing else to do except shell out enough money to feed a family a small family for a week with the cost of each of your cutting edge games UNLESS you were homebound in some way?)

    Maybe stop being the reason why the gaming industry only cares about improving their graphics at the cost of everything else. Maybe stop being the reason why graphics cards are so wildly expensive that scientific researchers can't get the tools they need to do the more complex processing needed to fold proteins and cure cancer, or use machine learning to push ahead in scientific problems that resist our conventional means of analysis

    KYS fool
  • BillBear - Monday, October 25, 2021 - link

    The performance numbers would look even nicer if we had numbers for that GE76 Raider when it's unplugged from the wall and has to throttle the CPU and GPU way the hell down.

    How about testing both on battery only?

Log in

Don't have an account? Sign up now