Live data from Hacker News

AmpereOne: Cores Are the New MHz

jeffgeerling.com

81–90 of 132 posts

Re: AmpereOne: Cores Are the New MHz

#81
post #75

Earlier quoted context omitted.

3 phase power is near universal inncentral and northern europe. And it's useful for powering saunas, electric warm water or bigger tools

why? The benefits of 3-phase power are really only used by factories with lots of large electric motors. 3 phase has no benefit for resistance heating.

You can move more power per conductor with three conductors than with two.

As a sidenote, my apartment has a 3x63A main fuse. Three-phase is everywhere here. Really convenient for ev charging too.

Re: AmpereOne: Cores Are the New MHz

#82
post #36
post #35

Earlier quoted context omitted.

Sounds like x86-64 is starting to lose some of it’s market share very soon.

Half of all CPUs in AWS are ARM already. Update: "Over the last two years, more than 50 percent of all the CPU capacity landed in our datacenters was on AWS Graviton." So not half of all but getting there. https://www.nextplatform.com/2024/12/03/aws-reaps-the-benefi...

Not to mention no one pays AMD or Intel the Suggested Retail Price for those CPU. You could expect any larger order from DELL or M$ / Google to be 50% off those prices. Of course Amphere would offer some discount as well but when you put together and the performance the difference isn't as big as most claimed.

Re: AmpereOne: Cores Are the New MHz

#83
I worried that AmpereOne isn't competitive with ARM core design. Its 512 CPU is only equivalent to x86-64 / Zen 6 256 Core with 2 threads. Which means AmpereOne may still not be competitive against AMD Zen6c unless they have their Core IPC massively improved.

But other than that we are only 2 years away from getting this into our hand a single server / node with 512 vCPU. Or may be 1024 vCPU if we do Dual Socket. Along with PCI-e 6.0 SSD. We can rent one or two of these server and forget about scaling issues for 95%+ of most use cases.

Not to mention nearly all dynamic languages has gotten a lot faster with JIT.

Re: AmpereOne: Cores Are the New MHz

#84
post #74

> I'm also happy to see systems like this that run without exotic water cooling and 240V power. Always funny to see 240V power described as exotic (or at least exotic-adjacent). It's the standard across practically the entire world except North and Central America!

They actually distribute 240V in USAmerica, one neutral wire and two opposite-phase 120V wires. Outlets in homes are 120V but wiring 240V is not difficult. https://en.wikipedia.org/wiki/Split-phase_electric_power This is different from Europe where we usually distribute 3 phases 130V.

3 phases 230v you mean?

Re: AmpereOne: Cores Are the New MHz

#85

> I'm also happy to see systems like this that run without exotic water cooling and 240V power. Always funny to see 240V power described as exotic (or at least exotic-adjacent). It's the standard across practically the entire world except North and Central America!

To be clear the water cooling part is what I meant to be 'exotic'. 240V is pretty prevalent in datacenters or network closets when specced, but is rarely seen outside of large appliance installs in US homes and small offices.

Re: AmpereOne: Cores Are the New MHz

#86

Earlier quoted context omitted.

why? The benefits of 3-phase power are really only used by factories with lots of large electric motors. 3 phase has no benefit for resistance heating.

You can move more power per conductor with three conductors than with two. As a sidenote, my apartment has a 3x63A main fuse. Three-phase is everywhere here. Really convenient for ev charging too.

And welding .. nice when you have to use it.

Re: AmpereOne: Cores Are the New MHz

#87
post #31

Earlier quoted context omitted.

When was the last time you saw a server with 10GB Ram no matter the number of cores\threads?

I've run plenty of microservices with 256 or 512 GB of RAM, and they were handling large loads. So long as each request is short lived, and using a runtime with low per request overhead (e.g. Node), memory is not really a problem for many types of workloads.

I know it's a typo, but you have accidentally coined the term megaservice.

Re: AmpereOne: Cores Are the New MHz

#88

> I'm also happy to see systems like this that run without exotic water cooling and 240V power. Always funny to see 240V power described as exotic (or at least exotic-adjacent). It's the standard across practically the entire world except North and Central America!

Every home in the US with an electric stove or water heater is using 240v electricity. It isn't really very exotic. 3 phase power to homes IS very exotic though and I'm not sure why it would be useful in homes.

Charging electric vehicles.

Re: AmpereOne: Cores Are the New MHz

#89

Earlier quoted context omitted.

Last time I looked at Ampere it was way less power hub but then the competition. EPYC is over 250W at idle! Off the top of my head, the Ampere 128 core was 250W at max load. Big difference. Only issue I had was cache size, and you could only buy them from NewEgg. Oh… and the dual socket gigabyte board had the sockets too close together that I’d you put a big ass heat sink and fan, the outlet of one fan will go straig…

I think you might be comparing your memories of Ampere's CPU power draw to either a full Epyc server's wall power draw or stated TDP. The 128 core EPYC 9754 in the chart linked above has a min of 10 W and a max draw of 397 W even though it outperforms the new 192 core variant AmpereOne A192-32X which had a min of 102 W and a max of 401 W. The min on the A192 does seem anonymously high though. That said, even the M128…

Hmmm. Weird because I was definitely out odd EPYC because of idle draw. Ok thanks, that’s back on the table now!!

Re: AmpereOne: Cores Are the New MHz

#90
post #28
post #20

I found this part particularly interesting: > Also, with 512 gigs of RAM and a massive CPU, it can run a 405 billion parameter Large Language Model. It's not fast, but it did run, giving me just under a token per second. If you're serious about running LLMs and you can afford it, you'll of course want GPUs. But this might be a relatively affordable way to run really huge models like Llama 405B on your own hardware. T…

It's really slow. Like, unusably slow. For those interested in self-hosting, this is a really good resource: https://github.com/XiongjieDai/GPU-Benchmarks-on-LLM-Inferen...

I would never chatgpt my code because I don't want to send it to Microsoft. Slow is better than nothing.
Post reply on HN