Stuff The Internet Says On Scalability For November 11th, 2016
Hey, it's HighScalability time:
- 9 teraflops: PC GPU performance for VR rendering; 1.75 million requests per second: DDoS attack from cameras; 5GB/mo: average data consumption in the US; ~59.2GB: size of Wikipedia corpus; 50%: slower LTE within the last year; 5.4 million: entries in Microsoft Concept Graph; 20 microseconds: average round-trip latencies between 250,000 machines using direct FPGA-to-FPGA messages (Microsoft); 1.09 billion: Facebook daily active mobile users; 300 minutes: soaring time for an AI controlled glider; 82ms: latency streaming game play on Azure;
- Quotable Quotes:
- AORTA: Apple’s service revenue is now consistently greater than iPad and Mac revenue streams making it the number two revenue stream behind the gargantuan iPhone bucket.
- @GeertHub: Apple R&D budget: $10 billion NASA science budget: $5 billion One explored Pluto, the other made a new keyboard.
- Steve Jobs: tie all of our products together, so we further lock customers into our ecosystem
- @moxie: I think these types of posts are also the inevitable result of people overestimating our organizational capacity based Continue reading
Rip and replace tactics almost always results in failure.
Deep learning isn't just for supercomputers any more.
These facilities will account for 53 percent of traffic by 2020.