Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Simple CPU Design

simplecpudesign.com

Post reply on HN
Print view View original

21–30 of 51 posts

Re: Simple CPU Design

#21

Post by ranger207 »

nxobject wrote: post #17

As much as I really benefited from being able to internalize system architectures like these many times over… I do wish now, as someone who ended up in software, that there were similar hand-holdy third guides to implementing the “core” of out-of-order superscalar execution engines, too. They’re crucial to understanding how modern processors _kinda actually work to a zeroth order approximation_, even though it’s impo…

At Georgia Tech I had one class (CS 2110) that dealt with implementing a simple in-order non-pipelined processor, one class that dealt with implementing a pipelined processor (CS 2210), then two classes (CS 4290 and CS 3220 IIRC) that dealt with implementing an out-of-order processor (4290 was more theory and also covered caches; 3220 was entirely implementing it on an FPGA). So, that sort of thing does exist, but IDK if most universities will let you take single classes like that

Re: Simple CPU Design

#23

Post by robinsonb5 »

artemonster wrote: post #2

how awesome that this exists. I was learning how CPU works and designing my own CPU w emulator like 20 years ago as a teenager by googling into obscure forums, blog posts and homemade cpu webring. I made an experiment not long ago "would I be able to find in google by myself all learning materials to do that again". The outcome of that experiment deeply unsettled me. Google just gives you shit and total garbage. Half…

Part of me thinks we need a new protocol, and a new lightweight web built around markdown with absolutely no (client side) active content allowed.

What I'm not sure about is how to combat bad actors / spammers / low-effort pages and AI slop. I'm leaning towards some kind of git-like storage with history as a mandatory part of the protocol, and some kind of cryptographic web-of-trust endorsement system.

Re: Simple CPU Design

#25

Post by linguistbreaker »

robinsonb5 wrote: post #23
artemonster wrote: post #2

how awesome that this exists. I was learning how CPU works and designing my own CPU w emulator like 20 years ago as a teenager by googling into obscure forums, blog posts and homemade cpu webring. I made an experiment not long ago "would I be able to find in google by myself all learning materials to do that again". The outcome of that experiment deeply unsettled me. Google just gives you shit and total garbage. Half…

Part of me thinks we need a new protocol, and a new lightweight web built around markdown with absolutely no (client side) active content allowed. What I'm not sure about is how to combat bad actors / spammers / low-effort pages and AI slop. I'm leaning towards some kind of git-like storage with history as a mandatory part of the protocol, and some kind of cryptographic web-of-trust endorsement system.

I'm starting to use Claude.ai more and more instead of googling. For the moment this seems to cut through the noise of the modern web.

Re: Simple CPU Design

#26

Post by spencerflem »

robinsonb5 wrote: post #23
artemonster wrote: post #2

how awesome that this exists. I was learning how CPU works and designing my own CPU w emulator like 20 years ago as a teenager by googling into obscure forums, blog posts and homemade cpu webring. I made an experiment not long ago "would I be able to find in google by myself all learning materials to do that again". The outcome of that experiment deeply unsettled me. Google just gives you shit and total garbage. Half…

Part of me thinks we need a new protocol, and a new lightweight web built around markdown with absolutely no (client side) active content allowed. What I'm not sure about is how to combat bad actors / spammers / low-effort pages and AI slop. I'm leaning towards some kind of git-like storage with history as a mandatory part of the protocol, and some kind of cryptographic web-of-trust endorsement system.

Sounds kinda like Gemini on top of IPFS/Dat/Hypercore. Imo some cool things but I'm not sure the problem is a technical one.

Content addressing has some real benefits in allowing something like the internet archive to be transparent (ie: it doesn't matter who hosts it). But that's mostly solving linkrot.

Searching through everything is still as hard as ever, and if the incentives are the same will be just as gamed. And people would have to make good content in the first place which is hard to justify without a good audience at the same time

Re: Simple CPU Design

#27

Post by spencerflem »

linguistbreaker wrote: post #25
robinsonb5 wrote: post #23

Earlier quoted context omitted.

Part of me thinks we need a new protocol, and a new lightweight web built around markdown with absolutely no (client side) active content allowed. What I'm not sure about is how to combat bad actors / spammers / low-effort pages and AI slop. I'm leaning towards some kind of git-like storage with history as a mandatory part of the protocol, and some kind of cryptographic web-of-trust endorsement system.

I'm starting to use Claude.ai more and more instead of googling. For the moment this seems to cut through the noise of the modern web.

I believe that it does, I'm worried long term that it will discourage people from making and curating webpages themselves though.

Re: Simple CPU Design

#28

Post by spencerflem »

artemonster wrote: post #19
spencerflem wrote: post #8

Earlier quoted context omitted.

I agree. Its hard though, SEO people are malicious, persistent and with modern tech, have incredible tools. And with hand curation, its hard to feel like its 'worth it' when instead of being able to build a community, your results are scraped and shown out of context. If you have any thoughts on how to get that sort of culture back I'm open to it

tbh I have dreamt about what could be possible if we were making some sort of "closed doors" internet branch. You can access with a single account bound to you, invite only, something like PGP with a web of trust. Good "legacy" internet websites can be chained and indexed through some sort of thematical webring, with a good search and comment functionality added on top, as a global HN. Any external content is opt-in…

In the same vein, I feel like the 'fair source' movement makes sense - pay a fixed percent of profit and get access to a massive collection of licenced software.

Just like with yours though, allocating it fairly is centralized and very hard to make everyone happy. And nobody wants to pay for something that used to be free.

Re: Simple CPU Design

#29

Post by LeftHandPath »

artemonster wrote: post #2

how awesome that this exists. I was learning how CPU works and designing my own CPU w emulator like 20 years ago as a teenager by googling into obscure forums, blog posts and homemade cpu webring. I made an experiment not long ago "would I be able to find in google by myself all learning materials to do that again". The outcome of that experiment deeply unsettled me. Google just gives you shit and total garbage. Half…

I am glad you followed up on this, to see if you could do it again! That matches my experience.

I remember feeling like the big tech corps had turned "consumer" into a pejorative and started relentlessly abusing their customers circa 2016 or so... Especially microsoft, post Windows 8. Consumer devices don't need to work. That's for pro devices. Consumer devices just need to sell ads, soak up user time, and let businesses market their goods for consumption!

The majority of search results from late 2019 or so and onwards have only degraded. Even on other platforms, like YouTube -- you get 4-5 real results, and the rest are "suggested for you", even if you've logged out. Google and Youtube both feel like "consumer" search engines, where advertising and eyeball time trump usefulness and user authority (i.e. the user being able to ask for what they want, and get it).

Re: Simple CPU Design

#30

Post by linguistbreaker »

spencerflem wrote: post #27
linguistbreaker wrote: post #25

Earlier quoted context omitted.

I'm starting to use Claude.ai more and more instead of googling. For the moment this seems to cut through the noise of the modern web.

I believe that it does, I'm worried long term that it will discourage people from making and curating webpages themselves though.

Definitely a possibility - hopefully AI will similarly empower creation of better content instead of AI slop noise.

In fact I wonder if Claude.ai could come up with similar CPU teaching tools and a syllabus based on some of the great resources linked in this discussion.

Post reply on HN