Hardkernel announces new ODRIOD H2

[VictordeHolland]

Hardkernel announces a new project with an X86 Intel CPU:

https://forum.odroid.com/viewtopic.php?f=29&t=32536

[LaurV]

What's "odriod"?

[henryzz]

I am not sure this would be great for Prime95. It looks like there is no AVX and it is Intel Atom based. It wouldn't surprise me if it is incapable of maxing out the memory bandwidth.

[M344587487]

110x110x43mm for 4C/4T 4MiB cache x86 SSE4.2 no AVX. From a compute density perspective for PRP I wonder how this compares to the best ARM solution. Without AVX, limited to a boost of 2.5GHz and having 4MiB of cache will this be able to saturate dual channel?

[science_man_88]

Originally Posted by LaurV

What's "odriod"?

https://en.m.wikipedia.org/wiki/ODROID

[Happy5214]

Originally Posted by LaurV

What's "odriod"?

Originally Posted by science_man_88

https://en.m.wikipedia.org/wiki/ODROID

I lack a sense of humor. Even I know that wasn't a serious question. ;)

[ewmayer]

I've been checking the odroid fora weekly for updates on the N2 (the refactored version of the big/little dual-core N1 ... they had engineering samples of the N1 and I even got one of the beta testers to do an Mlucas build/test on that back in March, but that project ran into issues with the memory-chip supplier, hence the need for the re-do in midstream), originally slated for a September rollout but takig longer than expected, and saw this - my first thought was "how does it compare in bang-for-buck terms with my little Intel NUC, which cost ~$400 but has an AVX2-capable dual-core Broadwell CPU?" Even if the H2 turned out to be slightly better in FLOPS-per-$ (which seems dubious) the number of H2s needing to be stacked to match the NUC wrecks the micro-footprint aspect of the Odroid.

But if any of our local enthusiasts decide to buy one, it would be useful to compare the following:

1. Prime95 on H2 vs on Intel NUC;

2. Prime95 on H2 vs Mlucas on C2 and N1 ... looking back at what I wrote at time of the N1 beta numbers (specifically post #158 on that page), using all 6 cores (one 2-threaded job running on the 'big' 2-core a72 cpu, another 4-theaded one on the 'little' 4-core a53 cpu) I got 2.2-2.3 the total throughput of an Odroid C2, which means ~3.5x the total throughput of a Pi3. I expect N2 to be similar, perhaps even somewhat better.

[kladner]

Originally Posted by Happy5214

I lack a sense of humor. Even I know that wasn't a serious question. ;)

Maybe it's not a serious answer, either.