3 ms·

I don't get why people are pouncing on the memory BW thing. There are very few applications which are actually bound by memory bandwidth (most of these will nev

I don't get why people are pouncing on the memory BW thing. There are very few applications which are actually bound by memory bandwidth (most of these will never be used on a laptop) and I've seen nothing so far to suggest that the CPU cores in the M1P/M1M can actually use it. The M1P/M parts are basically a midrange GPU and CPU in the same package and use a wide, high-speed memory interface (LPDDR is conceptually more similar to GDDR than standard DDR) to get GPU-like bandwidth because there's an actual GPU in there - which is actually rather impressive. For reference, in Zen 2/3 a chiplet (of eight cores) is limited to around 25 GB/s write, 50 GB/s read. Edit: The anandtech test has memory B/W numbers and gives 100 GB/s for a single core and around 220 GB/s for all cores, which is extremely high, but also not the full memory bandwidth.

zozbot234 5y ago
Memory bandwidth per core, per clock cycle is actually very limited on many systems. So you end up being bound by memory bandwidth any time you're running compute-intensive apps on multiple cores, unless thermal constraints hit sooner - which is quite unlikely on Apple silicon, and ARM silicon more generally.
sliken 5y ago
One clear benefit is decent GPU performance, without the cost, physical size, and power/cooling for a discrete GPU. It's way faster than Intel or AMD's integrated graphics. You can see this difference when you unplug a MBP vs any of the laptops with a discrete GPU. Marketing brags about memory bandwidth based on clockspeed * bus width = 400GB/sec. Getting 60% of peak on some memory bandwidth benchmark is pretty common. Try McCalpin's stream benchmark on your platform of choice to verify. I suspect you'll find similar on Intel or AMD on desktops, laptops, or servers. Not sure I buy the memory bound argument, sure peak performance will not change, but worst case performance can be caused by cache misses. So I expect that the new MBPs will be more evenly fast than the x86-64 competition, even while doing stressful things. I don't currently need a $3k laptop, but am hoping they ship a mini with a M1 Max or Pro. To quote anandtech: The fp2017 suite has more workloads that are more memory-bound, and it’s here where the M1 Max is absolutely absurd. The workloads that put the most memory pressure and stress the DRAM the most, such as 503.bwaves, 519.lbm, 549.fotonik3d and 554.roms, have all multiple factors of performance advantages compared to the best Intel and AMD have to offer. So an apple M1 Max is 19% faster on SpecFP (floating point applications) as a Ryzen 5950x which has twice as many fast cores (16 vs 8) and runs at 105 watts TDP. That's pretty amazing in my book.

zozbot234 5y ago

Memory bandwidth per core, per clock cycle is actually very limited on many systems. So you end up being bound by memory bandwidth any time you're running compute-intensive apps on multiple cores, unless thermal constraints hit sooner - which is quite unlikely on Apple silicon, and ARM silicon more generally.

sliken 5y ago

One clear benefit is decent GPU performance, without the cost, physical size, and power/cooling for a discrete GPU. It's way faster than Intel or AMD's integrated graphics. You can see this difference when you unplug a MBP vs any of the laptops with a discrete GPU. Marketing brags about memory bandwidth based on clockspeed * bus width = 400GB/sec. Getting 60% of peak on some memory bandwidth benchmark is pretty common. Try McCalpin's stream benchmark on your platform of choice to verify. I suspect you'll find similar on Intel or AMD on desktops, laptops, or servers. Not sure I buy the memory bound argument, sure peak performance will not change, but worst case performance can be caused by cache misses. So I expect that the new MBPs will be more evenly fast than the x86-64 competition, even while doing stressful things. I don't currently need a $3k laptop, but am hoping they ship a mini with a M1 Max or Pro. To quote anandtech: The fp2017 suite has more workloads that are more memory-bound, and it’s here where the M1 Max is absolutely absurd. The workloads that put the most memory pressure and stress the DRAM the most, such as 503.bwaves, 519.lbm, 549.fotonik3d and 554.roms, have all multiple factors of performance advantages compared to the best Intel and AMD have to offer. So an apple M1 Max is 19% faster on SpecFP (floating point applications) as a Ryzen 5950x which has twice as many fast cores (16 vs 8) and runs at 105 watts TDP. That's pretty amazing in my book.