Andreessen-Horowitz craps on “AI” startups from a great height

Re:Andreessen-Horowitz craps on “AI” startups from a great h... by eyegor on Monday February 24, 2020 @08:06PM (#22410052)

10gbe will depend on the workload. In general, I'd assume it's fine because it takes a parallel raid setup to saturate. Upgrading to 100gbe is pretty unreasonable cost wise unless you buy network gear from a back alley van dealer.

Although once you reach 4 2080ti, you ought to consider switching to a titanium grade psu and rewiring if you're in a 100-120v country. If you're feeling cheap, just steal the phases from two different circuits. Last I looked, most psu operate around 5% lower efficiency on 115 vs 230.

[ Parent | on HN ]

Re:Andreessen-Horowitz craps on “AI” startups from a great h... by liuliu on Monday February 24, 2020 @10:09PM (#22410612)

I've encountered some latency issues with allreduce on transformer models due to vocabulary sizes when communication have to cross PCIe lanes. Increasing batch size helps a lot, but low-latency & high-throughput is universally more helpful to lift these minor concerns (I don't really want to care about my batch size to improve allreduce performance). Hence worried not only throughput, but also latency on consumer-grade 10ghe equipments.

[ Parent | on HN ]

Re:Andreessen-Horowitz craps on “AI” startups from a great h... by xyzzyz on Tuesday February 25, 2020 @02:46PM (#22416649)

If you're feeling cheap, just steal the phases from two different circuits.

I chukled, but more seriously, if you can't rewire your house to get a normal 240V circuit, you should not be fucking around with hacks like above.

[ Parent | on HN ]