The device is an entry-level device with a chip that has been used in smartphones (iPhone).
One of the crucial points is, that it is limited to 8 GB DRAM regardless of the model.

Nowadays a reasonable entry level for a 64-bit operating system is 16 GB DRAM.

My Windows 10 system with a small Cubase 15 Pro project open has just 11.8 GB DRAM in use.
Who would want to do serious audio editing with only 8 GB DRAM?

I think the device has not been designed with such tasks in mind, and you do yourself a favor by buying a system
with 16 GB DRAM and a more regular CPU for audio related tasks with Near-Realtime processing demands.

Thanks Maggie33, Ki-Google answer was the same but not explained with arguments.

I wanted to have it in the forum as knowledge-base.
This Neo-book becomes a birthday-present or christmas gift for students probably and the question of software compatibility should be discussed before.
8 GB Ram is enough for beginners and as it is fanless silent, a great recording studio for cheap.

> But unified memory is architecturally different from traditional DRAM.
> It is shared across CPU, GPU and Neural Engine at 60 GB/s bandwidth,
> making direct comparisons to Windows DRAM usage figures misleading.

If the memory is shared between three components—CPU, GPU, and Neural Engine—then there is (much likely) effectively even less DRAM available to the CPU/NPU for processes at certain times. Another point is that these components may not always be able to read from or write to DRAM simultaneously.

Also, what happens if there is not enough free DRAM? In that case the system needs to page out, swap, or compress memory, all of which introduce latency.

For this reason—possibly to avoid such latencies—other components might reserve a certain amount of DRAM so that a portion remains directly available when needed. But this all means less DRAM for CPU and processes.

Regardless of the exact mechanism, 8 GB is not a large amount of memory, and effectively even less if the DRAM is shared in this way.

And I have a reason for this "speculation".

I once had a bad experience with a Lenovo business laptop running Windows where the CPU had an integrated GPU. I experienced audio dropouts during music playback while doing only light browsing. The CPU load was very low, yet I could only stabilize playback with a 256-sample ASIO buffer size, which is unusually high given the minimal system load. That discrepancy made me investigate further.

It turned out the issue was related to the iGPU; DRAM was also shared between the CPU and the GPU, which was inside the CPU.

Fortunately, the laptop also had a discrete NVIDIA GPU, and the system allowed configuring which applications used the iGPU or the NVIDIA GPU. As soon as I forced Firefox to run on the NVIDIA GPU, the audio issues disappeared completely. I was then able to run my RME UFX at a 48-sample ASIO buffer size without any audio dropouts, even under higher load and browsing the internet.

After that experience with shared DRAM between CPU and iGPU, I have remained skeptical of such shared-memory designs.

I also remember reports of strange audio dropouts on certain Macs running the same macOS version. This was years ago, and at the time one theory was that the problem might be related to the installed DRAM capacity, because systems with more memory did not show the issue. However, it was impossible to verify, since upgrading DRAM on those machines was not possible.

In the end, Apple is still subject to the same physical constraints as everyone else; they also only boil with water.

Incidentally, there have been enough problems lately with Apple systems that were significantly better equipped. For this reason alone, a certain amount of skepticism is certainly warranted considering such meager DRAM equipment.

And you should always plan for certain DRAM reserves, because operating systems tend to become more and more bloated over the years. What works today may not necessarily work tomorrow.

In my opinion, these devices are mainly there to attract more buyers to the Apple platform and get people used to the Apple ecosystem as early as possible. Consider also the additional school and student discounts, which make the device even more affordable.

It will be interesting to see how these devices perform in practice.

Everyone reading here for a longer time knows about the M1 8 GB machine and its Memory Pressure issues. Unusable for audio, and still is. That's why I have zero hopes for the Neo and would recommend against it.

Don't know what exactly is meant..

However, IMHO I can say:
- A 8GB Tahoe VM runs without issues
- Even my old Intel MBP 2010 with 8GB (pimped to Ventura via OCLP) has no audio issues with 802. At least as long as I don’t go overboard with the number of tracks/load in Logic (I know, it uses DRAM).

To add: Im not an Apple/Win/Linux Fanboy. All Systems have their Pros and Cons (while the first two ones are MaxProfit oriented, sure). However that’s not the point here.