From 5d2b48146ef8bab4964920d2f59d5304a3bbc0ce Mon Sep 17 00:00:00 2001 From: Patrick Sanan Date: Thu, 18 Jun 2020 08:06:52 +0200 Subject: [PATCH] Latency numbers: remove conversion to throughput Remove conversion of latency numbers to throughput. This is misleading for two reasons: 1. Latency and throughput are not computable from each other. For example: if I know it takes 1 second to turn on the hose, get 1 liter of water, and turn it off, I cannot conclude that the maximum rate I can get water out of the hose is 1 liter per second (though I can conclude it's more than that). 2. The throughput estimates seem suspect, in and of themselves. These are great numbers to know, of course, but need to be backed by a separate source. (This'll be extra important if and when this guide includes information about, say, GPUs, which are optimized for throughput). --- README.md | 4 ---- 1 file changed, 4 deletions(-) diff --git a/README.md b/README.md index f9563097..34ea5901 100644 --- a/README.md +++ b/README.md @@ -1623,10 +1623,6 @@ Notes Handy metrics based on numbers above: -* Read sequentially from disk at 30 MB/s -* Read sequentially from 1 Gbps Ethernet at 100 MB/s -* Read sequentially from SSD at 1 GB/s -* Read sequentially from main memory at 4 GB/s * 6-7 world-wide round trips per second * 2,000 round trips per second within a data center