Comments on: Crunching Photons And Electrons Down Into Datacenter Switch ASICs https://www.nextplatform.com/2020/03/17/crunching-photons-and-electrons-down-into-datacenter-switch-asics/ In-depth coverage of high-end computing at large enterprises, supercomputing centers, hyperscale data centers, and public clouds. Sun, 19 Jun 2022 16:57:27 +0000 hourly 1 https://wordpress.org/?v=6.7.1 By: Timothy Prickett Morgan https://www.nextplatform.com/2020/03/17/crunching-photons-and-electrons-down-into-datacenter-switch-asics/#comment-192895 Sun, 19 Jun 2022 16:57:27 +0000 http://www.nextplatform.com/?p=125416#comment-192895 In reply to Fenton Heirtzler.

Well, I am not making any claims. But Intel is. The total number of transceivers has to exceed the number of servers, and it depends on the topology and tiering in the network. The number of switches will be fewer than the number of servers, but maybe only by half as Intel has suggested.

]]> By: Fenton Heirtzler https://www.nextplatform.com/2020/03/17/crunching-photons-and-electrons-down-into-datacenter-switch-asics/#comment-192852 Sun, 19 Jun 2022 00:17:35 +0000 http://www.nextplatform.com/?p=125416#comment-192852 Dear Mr. Morgan,
You are making surprising claims in this article. For example, you claim that a “hyperscale” data center will have:
≥ 1EE6 optical transceivers
≥ 1EE5servers
≥ 5EE4 optical switches
That would mean:
(# transceivers)/(# servers) = 1EE6/1EE5 = 10 ??
(# switches)/(# servers) = 5EE4)/1EE5 = 0.5??
So few transceivers and switches?

]]> By: DAN D NORDMANN https://www.nextplatform.com/2020/03/17/crunching-photons-and-electrons-down-into-datacenter-switch-asics/#comment-140796 Thu, 19 Mar 2020 01:14:30 +0000 http://www.nextplatform.com/?p=125416#comment-140796 Great article! Any thoughts on the Nvidia buyout of Mellanox with respect to these 100G + densities?

]]>