Russia's 16nm Elbrus CPU: 16-Core, 2.0 GHz, 4-socket 16TB capable, VLIW CPU
Russian Company Tapes Out 16-Core Elbrus CPU: 2.0 GHz, 16 TB of RAM in 4-Way System https://www.tomshardware.com/news/russian-company-tapes-out-16-core-elbrus-c... So the anarchist crowd spec's "open source/ open hardware" designs, for what it's worth. Step 1 is understanding the development models and the possibilities around various "open"ness options. "Open" anything has engineering tradeoffs, as do other models, the primary competition being "proprietary", which is mostly also "closed". The advantages of "open" include: - maintaining relevance and interest outside of the primary sponsor (e.g. govt. and military) - the possibility of development of an 'ecosystem' (software, hardware, community, startups/ companies/ industries) - to the extent that an ecosystem is born and develops, there grows a "demand economy" for the hardware at issue - this becomes a symbiotic thing where interest drives innovation drives demand which drives down costs, which in turn drives the interest and demand IF such an ecosystem is found to be a goal, certain steps foster this pathway, such as a "Rasberry-Pi"-type dev board, then a desktop workstation class computer, etc. Not unlike the steps IBM has taken with their Power architecture.
On Sat, Oct 10, 2020 at 01:37:56AM +1100, Zenaan Harkness wrote:
Russian Company Tapes Out 16-Core Elbrus CPU: 2.0 GHz, 16 TB of RAM in 4-Way System https://www.tomshardware.com/news/russian-company-tapes-out-16-core-elbrus-c...
So the anarchist crowd spec's "open source/ open hardware" designs, for what it's worth.
Step 1 is understanding the development models and the possibilities around various "open"ness options.
"Open" anything has engineering tradeoffs, as do other models, the primary competition being "proprietary", which is mostly also "closed".
The advantages of "open" include:
- maintaining relevance and interest outside of the primary sponsor (e.g. govt. and military)
- the possibility of development of an 'ecosystem' (software, hardware, community, startups/ companies/ industries)
- to the extent that an ecosystem is born and develops, there grows a "demand economy" for the hardware at issue - this becomes a symbiotic thing where interest drives innovation drives demand which drives down costs, which in turn drives the interest and demand
IF such an ecosystem is found to be a goal, certain steps foster this pathway, such as a "Rasberry-Pi"-type dev board, then a desktop workstation class computer, etc. Not unlike the steps IBM has taken with their Power architecture.
One can imagine a rather cheeky "pip Intel at their own posted intention" scenario where say (using TSMC's very cool TSV tech) an enthusiast class CPU has a mix of cores, say: - 4 Elbrus cores - a couple of old (no longer in patent) x86 (ASM level) cores - or probably better yet, a POWER core - a Xilinx-style programmable core (this ought be recreated as "original IP" due to the relatively simple/repetitive nature of such cores) - if the programmable core is not exciting enough, a custom media core which say just does de/encode of MP3 and VP9 - and while you're at it, throw in say 1GiB of L3 cache to spike the fun in an almost "illicit" way :) - oh, and to rekindle the glory days of low latency keyboarding from the halcyon days of home computing, throw in an on-chip "zero latency" "keyboard controller" (really just some serial line controller - perhaps this is what PS2 connection is?) which would totally 'rock' for certain gaming genres The point being, something a bit quirky yet inspiring of possibilities, may spike interest and even passion in the enthusiast community.
participants (1)
-
Zenaan Harkness