basiccalendar74

Weights will be fetched 4-bit from memory, but compute happens in FP8 (W4A8) or BF16 (W4A16). This is still better than fetching 8-bit weights from memory.

Both W4A8 and W4A16 schemes are supported by Hopper GPUs and commonly used to serve mxfp4 Kimi models on Hopper.

can you show the math for your last statement?

it is only "right" statistically as in conforming to the same distribution. but there is no guarantee of exact same output.

A large cap stock increasing by 6x is not "unlikely to be achievable".

Apple went from 1T to 4T valuation in just 6 years. why did Apple not award 500 billion to Tim Cook for this achievement?

by 'allocated', I mean allocated during the purchase decision. not that money is sent to respective company.

Passive investing is not an issue, but the default bias towards large cap equities like SP500, Nasdaq100. Passive investing through total market ETFs (like VTI) maintains the status quo.

For example, if they are only two companies, say with 1T and 4T market cap. If one invests 5M into a total market ETF, 1M is allocated to company A and 4M to company B. But since company B is 4x bigger than company A, the upward price pressure is the same for both companies.

Yes, the renewals are included

curious how much of the growth is solely due to renewals since the cap on first time H1-Bs has stayed the same at 85k per year.

the green card queue is severely backlogged for India (and to some extent China, Mexico). this causes people who would usually get a green card after 3-5 years be on a continuous H1-B renewal cycle every 3 years.

This is normal for most state border in India. They mostly inspect commercial trucks

yes, it was in the 90s last August.

Deepseek v3 needs 16 H100s or 8 H200 GPUs for inference.

it is 23% of total energy. if you consider only acceleration energy, it will be much larger percentage, probably 80% (as in electric cars).

you can add an '/s' at the end next time to avoid ambiguity. :)

this is a well-studied tradeoff with immigration. high-skilled immigration helps US economy and innovation but low-skilled immigration depresses wages for citizens. you are making this distinction for a narrow case i.e. tech.

yes

you invest in companies that would be biggest market cap in future.

RSUs are treated as cash income at vest and taxed at same rates.

Stock buybacks benefit general shareholders (i.e. beyond employees) since they push up stock value without causing a taxable event. The alternative is dividends which are immediately taxed.

Nvidia was a $750B company in 2021 (pre-ChatGPT). Cause of gaming and all of pre-LLM machine learning using nvidia gpus.

you are comparing a sedan and suv though. byd seal is a better comparison to model 3

you can sell index funds and buy stuff. you will mostly be ahead of inflation.

not a data breach

one is a public company whose stock can be pumped up with "bullshit", other is not.

how much did it change compared to 2019?

hate how all YoY commentary completely ignores recession years like 2022 or boom years like 2021.

Forward PE (assuming same earnings as current quarter for full year) is around 40. Trailing PE can be misleading in high growth phases.

credit card processors charge around 3%

each of us has apps we like and Tesla only supports a few apps. Ex. I use YouTube music but tesla only supports Spotify and Apple Music.

main use case seems to be simulations in 2D, 3D or nD spaces. spaces -> spatial.

you seem to be comparing energy units with power units. 1 GW of installed solar capacity should be give about 1 TW hour of energy per year (365 days * 3 hrs per day as a conservative estimate).

except META was $90 when all tech was in recession. TSLA is crashing even when all tech and broader market has been rising.