I see a future for a bananacoin collab
377 karma · joined February 20, 2014
If we can't agree on that baseline, then its quite obvious that we'll continue to have an escalation in the types of violence that we've seen in the past few years, against the political and corporate classes in the US, with very little end in sight.
For example, the fact that right-wing governments in central and eastern Europe are protecting their borders, represents a very popular perspective, apparently shared by very few in the EU governing body.
Consolidating power at a moment when many EU policies are clearly unpopular seems like it will have unintended consequences.
The question "why run CPU code on GPU when GPU cores are slower?" assumes you're running ONE program. But GPUs execute in SIMD wavefronts of 32 threads - and here's the trick: each of those 32 lanes can run a DIFFERENT process. Same instruction, different data. Calculator on lane 0, text editor on lane 1, file indexer on lane 2. No divergence, legal SIMD, full utilization. Suddenly you're not running "slow CPU code on GPU" - you're running 32 independent programs in parallel on hardware designed for exactly this pattern.
The win isn't throughput for compute-heavy code. It's eliminating CPU roundtrips for interactive stuff. Every kernel launch, every synchronization, every "GPU done, back to CPU, dispatch next thing" adds latency. A persistent kernel that polls for input, updates state, and renders - all without returning to CPU - changes the responsiveness equation entirely.
A few things to try at home if you're curious:
2. Put 32 different "process states" in a buffer and have each SIMD lane execute instructions for its own process. Watch all 32 make progress simultaneously.
3. Measure the latency from "input event" to "pixel on screen" with CPU orchestration vs GPU polling an input queue directly. The difference surprised me.
The persistent kernel thing has a nasty gotcha though - ALL 32 threads must participate in the while loop. If you do if (tid != 0) return; then while(true), it'll work for a few million iterations then hard-lock. Ask me how I know.
Most people in Seattle "tech" are middle management with no discernible skills other than organizational deckchair arrangement. It is a place to optimize for work-life balance, and not take risk - this is why the region, despite its technology density, has such a disproportionately small startup scene.
AI IS a huge threat to a place like this and I am not optimistic about the ability for people to adapt.
Edit: for those downvoting me, Google literally shut down their China operations because they were unwilling to comply with censorship requirements. Conversely, Google and other US companies seem completely willing to comply with national security letters that compel them to spy on non-US persons, which should make other countries where US companies operate equally uneasy.
On one side, target IRR and thus valuations are fixed for the investor by their LPs
On the other side, runway and thus survival rate are fixed by local market prices for founders who are risk tolerant and maximize for the same IRR as the investor.
The only solution is to physically move to a place that isn’t as stupidly priced as the Bay Area.
I think there is a reckoning coming, where investors will realize that backing startups in locations where costs continuously increase will become a suckers game. Survival rate naturally will go down due to shortened runway, while fundraising milestones will stay fixed - decreasing the expected IRR of the asset class.
I agree that Google benefits from monopolizing headcount, even if the employees waste their lives doing nothing, because it reduces competition. I would even Google poisons people's work ethic/priorities permanently when they try to leave and go to other startups (this is my personal experience working with former non-technical google employees who only play politics).
If I were a large pension fund or asset manager, I would be asking questions to the board like, "Given nothing gets done on many teams, what would happen if Google reduced its headcount by 50%?" and "Why is increasing headcount currently an incentive for managers within organizations, especially without any apparent penalty function?"