Nvidia GB10's Memory Subsystem, from the CPU Side
chipsandcheese.com76 points by ingve 16 hours ago
76 points by ingve 16 hours ago
What’s stopping Mediatek from putting some of those cores in a laptop/desktop CPU package? Is it the infrastructure around it? OEM support? All of the above?
I assume that the author here is testing against one of these boxes, right? https://marketplace.nvidia.com/en-us/enterprise/personal-ai-...
Are these considered a good deal at $3-4k? What's the software support like on them? I've got 2x 3090s and I'm curious how this compares.
DGX Spark vs. Strix Halo vs. M4 Max is hotly debated. You can find plenty of HN discussions and YouTube videos about it.
> https://marketplace.nvidia.com/en-us/enterprise/personal-ai-...
In Europe these cost 5k EUR. I guess I'm not buying computer ever again and hopefully the 10 years old ones I have will never die.
I don't understand on one of the later graphs the core to core latency for strix halo goes out to 32 cores but he says only has 16 cores?
AMD's cores have SMT, allowing them to run two threads at a time and appear to the OS and its scheduler as two logical cores despite being implemented as a single physical core.
What pattern in the data shows that's what's being measured? I would expect to see basically 0 latency between adjacent "cores" then since L1 is shared per thread?
Co resident threads might not get any speed up here since coherency instructions are functionally operations on the L2 cache.
Kind of an odd choice to have clusters that are so different