Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
ryao
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
20 ms
·
271.
▲
by
ryao
1y ago
I had no idea that literary fiction was a genre rather than a generic term for fictional literature until I read this: https://en.wikipedia.org/wiki/Literary_fiction There is a very simple explanation for the decline.
272.
▲
by
ryao
1y ago
I once tried having my web browser claim to be the Google spider. It worked well and I completely forgot about it until one website soft banned me a year or two later for impersonating the Google spider. I contacted them to complain. They l
273.
▲
by
ryao
1y ago
The point of the article is to say that this is a bad thing, but to someone who was bullied as a child, the idea of being “sedentary, scrolling and alone” sounds like the ideal childhood. I remember adults saying socialization was a good th
274.
▲
by
ryao
1y ago
Your comment does not make much sense to me. The 355x and 300x are two different chips, not binned versions of one another. The RTX PRO 6000 has fp4/fp6 support too, so there is no need to use datacenter exclusive hardware to get that.
275.
▲
by
ryao
1y ago
The higher power requirements of datacenter GPUs are mostly from running hardware well past a reasonable point on the efficiency curve to eke slightly higher generational increases in performance. Nvidia for example has three versions of th
276.
▲
by
ryao
1y ago
A number of people want to purchase their own hardware, not rent cloud hardware. I recently purchased a RTX PRO 6000 for the same reason, despite having the option of renting a B200 VM for $1.49 an hour from DeepInfra until the end of June.
277.
▲
by
ryao
1y ago
It has gone through many boom and busy cycles. If you go far back enough, it was very well funded. In particular, I recall reading about the US government investing 1 to 2 billion dollars during the Cold War into AI research to translate Ru
278.
▲
by
ryao
1y ago
Most of those willing to work with AMD are very small players (with some notable exceptions). They are likely hopeful that the small players will grow.
279.
▲
by
ryao
1y ago
The pronunciation of Y in Latin is the pronunciation of I with rounded lips. That is probably the reason for the name.
280.
▲
by
ryao
1y ago
The Latin alphabet was based on the Etruscan alphabet, which was based on the Greek alphabet, which was based on the Phoenician alphabet. In early Latin, C, G and K were all pronounced the same way. They later dropped G, only to bring it ba
281.
▲
by
ryao
1y ago
If people followed the alphabetic principle for English, the written language would become unintelligible due to all of the regional variations in pronunciation causing there to be many variations of the same words.
282.
▲
by
ryao
1y ago
The table is not empty. You need to pick the windows versions to show since the table would be so huge if it showed all of them that they are hidden by default. Click “show” in the table head.
283.
▲
by
ryao
1y ago
I wonder how the DRM used in some video games that calls Windows system calls directly handles these changes. We know that some of it uses windows system calls directly because both Linux and Wine had to be patched to support it: https:&#x
284.
▲
by
ryao
1y ago
It makes sense if you are a regular and today, you want something other than your usual.
285.
▲
by
ryao
1y ago
Coincidentally, I am in NY. That said, “I want…” when ordering seems fine to me too.
286.
▲
by
ryao
1y ago
> At the dinner table a Norwegian is likely to say something like “Jeg vil ha potetene” (literally “I will have the potatoes”, which sounds presumptuous and haughty in English) where a brit might say “Could I please have some potatoes?”.
287.
▲
by
ryao
1y ago
Then that would be $100 per year and none of your numbers for annual plans are accurate. Your website claims $99 per year and claims that is equivalent to $8.50 per month. None of these 3 numbers are equivalent.
288.
▲
by
ryao
1y ago
The discount is applied to $10/month and is said to make it effectively $8.5 per month, but is actually $8.25 per month, since they claim to charge $99 per year.
289.
▲
by
ryao
1y ago
The website says “$8.50/month” and “$99 billed annually” for Pro Annual. Those numbers are inconsistent. $99 annually is $8.25/month. Usually, one would expect a wealth management service to be able to divide 99 by 12 correctly. F
290.
▲
by
ryao
1y ago
You can predict the token generation performance of a GPU or CPU by dividing the memory bandwidth by the size of the active parameters. By definition, that is a memory bandwidth bottleneck. I have no idea why you think it is not. Anyone who
291.
▲
by
ryao
1y ago
I use wayland with sysvinit and comments like this make me cringe. He was not even talking about wayland and you managed to inject this anti-X11 tirade into the conversion anyway. X11 works great for many people and if it were not for XWayl
292.
▲
by
ryao
1y ago
It should be obvious to him that it is not happening given that several contributors all responded no. As for closing the bug, no one has volunteered to be the one to do it.
293.
▲
by
ryao
1y ago
Which issue? I can reopen it and mark it as not stale.
294.
▲
by
ryao
1y ago
This seems appropriate: https://www.youtube.com/watch?v=LZXwdRBxZ0U It took a little longer than predicted, but “Googlezon” is finally happening, with or without Google and Amazon.
295.
▲
by
ryao
1y ago
Once the not-stale label is set, the bot ignores the issue and will not close it again. At least, that is how it works for the OpenZFS bug tracker.
296.
▲
by
ryao
1y ago
In 2022 and 2023, I made a push to find and fix bugs in OpenZFS using static analysis tools. I found many obscure bugs. One was a theoretical issue that should have affected big endian systems and been an annoyance for users, but there were
297.
▲
by
ryao
1y ago
I have seen OpenZFS adopt one, but whenever I have seen a bug that has merit closed by the stale bot, it is reopened by a contributor and a not-stale flag is added to prevent it from being automatically closed again. Note that I am a contri
298.
▲
by
ryao
1y ago
The memory wall (also known as the Von Neumann bottleneck) is still true. Token generation on Nvidia GPUs is memory bound, unless you do very large batch sizes to become compute bound. That said, more exotic architectures from cerebras and
299.
▲
by
ryao
1y ago
There is no reason people cannot write userland drivers in Python.
300.
▲
by
ryao
1y ago
N is a variable in what I posted, not a constant.
More ›