For single-threaded, 20% is great. For multi-threaded, it's exactly equivalent to their core-count increase, meaning virtually no IPC or clockspeed improvement. I'm guessing the CPU cores are the same as the baseline M2, so we should have some idea about the answer... I just haven't looked it up, yet.The M2 Ultra chip will have 24 CPU cores ... for 20% faster performance than M1 Ultra ... which offered 20 CPU
I highly doubt that. The M1 series' memory bus was already scaled up to crazy widths.I'm guessing they doubled the channels to get the 2x memory bandwidth over their M1 siblings, but that's only a guess.
|Variant||Memory data bus width (bits)|
I think I mess read somewhere that they had double memory bandwidth with the M2 line, but a quick check shows the base M2 is the only one that got a memory bandwidth bump (100 GB/s vs 66.7 GB/s). Would make sense if they are using the same LPDDR5-6400 and same bus width.I highly doubt that. The M1 series' memory bus was already scaled up to crazy widths.
Variant Memory data bus width (bits) M1 128 M1 Pro 256 M1 Max 512 M1 Ultra 1024
According to this the M2, M2 Pro, and M2 Max all have the same width as their M1 counterparts.
Supercharging MacBook Pro and Mac mini, M2 Pro and M2 Max feature a more powerful CPU and GPU, up to 96GB of unified memory, and power efficiency.www.apple.com
The base M1 used LPDDR4. All of the other M1s used LPDDR5. So, that should explain why the base M2 got a speed increase (i.e. from switching to LPDDR5).a quick check shows the base M2 is the only one that got a memory bandwidth bump (100 GB/s vs 66.7 GB/s). Would make sense if they are using the same LPDDR5-6400 and same bus width.
Source?No H.265, No AV1 encoding
I agree that I/O is a sore spot. However, other than a few SSDs, there's not really any PCIe 5.0 device out there. Plus, there's such limited selection of cards you can put in a Mac.2 x16 PCIe 4 slots; 4 x8 PCIe 4 slots