aurareturn

6 ms·

1.
If Astra was trained on 100k Blackwell GPUs, what happens with 1M Rubin?(twitter.com)
2 points by aurareturn 2d ago|1 comments
2.
by aurareturn 3d ago
Because I’ve lived there for a long time and understand the locals more than whatever the western media tells.
3.
by aurareturn 4d ago
The growth seems to be accelerating.
4.
by aurareturn 4d ago
What indication do you see that we will see a slow down in model capabilities or AI use?
5.
by aurareturn 5d ago
They absolutely deserved imprisonment. They broke laws and went to foreign countries for aid. Imagine if Hawaii tried advocated for succession and went to China for help. The US isn't going to do anything? Come on. This double standard
6.
by aurareturn 5d ago
Huh? I'm just wondering why you even linked me to a source that is commissioned by a British news agency done by an institution that has historically been pro-sucession. And then another one by authors at The China Quarterly which has
7.
by aurareturn 5d ago
"Peer reviewed". My goodness. No, I'm not mistaken. Regardless, how was your experience living in HK in 2019? Hopefully you've talked to many locals in cantonese?
8.
by aurareturn 5d ago
See for example the contemporaneous polling done by Reuters below - 59% of the population supported the protests at the end of 2019 and 30% had participated [1]. I looked into this source. First, it was commissioned by Reuters, a ne
9.
by aurareturn 5d ago
The University of Cambridge, UK source is the 2nd one right? What about it makes you think that it's right and not the other sources citing housing? How much do you know about the HK housing situation in 2014-2019? The survey data
10.
by aurareturn 5d ago
There is a difference between demonstrations and violent protests. The protests lasted nearly a year. Lastly, let's just say that the survey source is highly suspect. It's commissioned by a British news agency with an institute wh
11.
by aurareturn 5d ago
The source is literally in your link. Read it carefully. residents polled in a survey conducted **for Reuters** by the Hong Kong Public Opinion Research Institute It's literally commission by Reuters, a British news agency. It i
12.
by aurareturn 5d ago
They collected the data FOR Reuters. I guess you didn’t check your own source.
13.
by aurareturn 5d ago
Okay use my 2nd source then? Data source: That's the Reuters one. I summarized it with a quote? What do you want? Alternatively/also read 2nd source, which also discuss it, and again they say economic/housing was
14.
by aurareturn 5d ago
I mean, the numbers speak for themselves right? Only 37% participated in any demonstration - likely only the non-violent ones in the beginning. His source doesn't actually disprove what I said - that most locals hated the protests.
15.
by aurareturn 5d ago
A poll conducted for Reuters by an institute whose ex-CEO has historically expressed anti-Beijing and HK pro-succession views and has since moved to UK.[1] Yea, no bias there. Reuters is a British news agency wholly owned by Thomson Reu
16.
by aurareturn 5d ago
At this point there is really no Hong Kong of old. China now fully controls that area, breaking a treaty with the UK. For much of Hong Kong's history under British rule, there was no freedom. It was whatever a single British ma
17.
by aurareturn 5d ago
No one said people are property. Hong Kong people still enjoy disproportionally high quality of life compared to Mainland Chinese people. In fact, they are ranked #14 world wide in GDP PPP per capita.[0] The biggest reason for this is becau
18.
by aurareturn 5d ago
Reported by who? The famously neutral The Guardian? There were large participation in non-violent protests in the beginning. By the time it got violent, it was almost all just students and the youth protesting. It lasted for almost an entir
19.
by aurareturn 5d ago
I was at the 2019 Tiananmen Square vigil that kick started the protests. Don't mistake the local population being sympathetic to some of the protester's messages over supporting the protests. When the protestors became disruptive
20.
by aurareturn 5d ago
Yes, exactly. See this report: https://www.scmp.com/news/hong-kong/politics/article/3030209... Basically, it was pent up frustration from the youth over housing controlled by Hong Kong's property ty
21.
by aurareturn 5d ago
I was living in Hong Kong when the protests first started and stayed through the entire 2019 year. First, the vast majority of the locals hated the protests and didn't support them. This isn't reported by the western media. The me
22.
by aurareturn 5d ago
Both of those examples wouldn't have gotten traction in the pre-AI HN in my opinion. The Lenovo laptop jet is experimental and the details suggests that it uses 7.5w to cool 30w. That's not very efficient. You might as well have a
23.
by aurareturn 5d ago
Sometimes Intel chips would regress generation on generation. Broadwell in 2015 was sometimes 10% slower than the generation it replaced.
24.
by aurareturn 5d ago
They should use LPDDR6 which is a sizable improvement but the other thing they can do is increase prompt processing even further.
25.
by aurareturn 5d ago
Absolutely massive improvement in CPU performance in a single year. Looks like TSMC N2 is going to be a very good node. +22% faster single thread +34% faster multi thread +40% GPU This is suppose to be the generation that Apple wants to fas
26.
by aurareturn 5d ago
geekbench results cannot be compared across platforms. Why not?
27.
by aurareturn 5d ago
I think the clear next thing for the AI economy is that companies who can afford it will be buying private Nvidia racks and using open source models. Next, OpenAI and Anthropic will respond by developing on-premise services where their mode
28.
by aurareturn 6d ago
Also GPU test: https://browser.geekbench.com/v6/compute/compare/6874665?bas...
29.
Apple A20 Pro Geekbench Result(browser.geekbench.com)
36 points by aurareturn 6d ago|29 comments
30.
by aurareturn 6d ago
Would that be ok in your mind? It would in my mind. Hopefully companies have looked through the agreement.

More ›