## Meta 的 Llama 3.1

Meta 發佈了 Llama 3.1：「Introducing Llama 3.1: Our most capable models to date」，這本來就只是個發佈而已，但讓我注意到的是 AWSGCP 都同時宣佈在雲端上支援 Llama 3.1 了：

It will be doing so by tapping into the same wave of puzzle-mania that helped simple games like Wordle find viral success and millions of players. Three early efforts are games called “Queens”, “Inference” and “Crossclimb.”

“We’re playing with adding puzzle-based games within the LinkedIn experience to unlock a bit of fun, deepen relationships, and hopefully spark the opportunity for conversations,” the spokesperson said in a message to TechCrunch. “Stay tuned for more!”

## 估算 YouTube 影片總量的方式

Here’s how this works: YouTube URLs look like this: https://www.youtube.com/ watch?v=vXPJVwwEmiM

That bit after “watch?v=” is an 11 digit string. The first ten digits can be a-z,A-Z,0-9 and _-. The last digit is special, and can only be one of 16 values. Turns out there are 2^64 possible YouTube addresses, an enormous number: 18.4 quintillion. There are lots of YouTube videos, but not that many. Let’s guess for a moment that there are 1 billion YouTube videos – if you picked URLs at random, you’d only get a valid address roughly once every 18.4 billion tries.

We refer to this method as “drunk dialing”, as it’s basically as sophisticated as taking swigs from a bottle of bourbon and mashing digits on a telephone, hoping to find a human being to speak to. Jason found a couple of cheats that makes the method roughly 32,000 times as efficient, meaning our “phone call” connects lots more often. Kevin Zheng wrote a whole bunch of scripts to do the dialing, and over the course of several months, we collected more than 10,000 truly random YouTube videos.

By comparing our results to other ways of generating lists of YouTube videos, we can declare them “plausibly random” if they generate similar results. Fortunately, one method does – it was discovered by Jia Zhou et. al. in 2011, and it’s far more efficient than our naïve method. (You generate a five character string where one character is a dash – YouTube will autocomplete those URLs and spit out a matching video if one exists.) Kevin now polls YouTube using the “dash method” and uses the results to maintain our dashboard at Tubestats.

In our case, our drunk dials tried roughly 32k numbers at the same time, and we got a “hit” every 50,000 times or so. Our current estimate for the size of YouTube is 13.325 billion videos – we are now updating this number every few weeks at tubestats.org.

## GCP 的 IPv4 也要漲價了

External IP 的部分，漲 25%：

Static and ephemeral IP addresses in use on standard VM instances will go from \$.004 to \$.005.

Static and ephemeral IP addresses in use on preemptible VM instances will go from \$.002 to \$.0025.

Cloud NAT 吃的 IPv4 address 的部分，從本來沒有收變成要收費：

Static and ephemeral IP addresses mapped to Cloud NAT Gateway will go from No Charge to \$0.005.

## GCP 提供每個區域的 Standard Tier Networking 每個月 200GB 的免費頻寬

GCP 提供每個區域 Standard Tier Networking 每個月有 200GB 的免費頻寬：「Announcing 200 GB free Standard Tier internet data transfer per month」。

## AWS Aurora Xanadu？

franckpachot

Google has Spanner. AWS is working on something similar (project Aurora Xanadu). And both have YugabyteDB in their marketplace. Those are Distributed SQL (Global ACID), not Citus. For DataWarehouse which doesn't need ACID, there are other services.

## GCP 的 Disks 與 AWS 的 EBS 的比較...

AWS 最常用的 `gp3` 是 \$0.096/GB，無論空間大小，效能上都提供 3000 IOPS 與 125MB/sec throughput，另外可以加價購買 IOPS 與 throughput。不過也因為這個性質，拿來當開機碟很好用。

GCP 如果是 Balanced Persistent Disk (Balanced PD) 是 \$0.1/GB，效能上會提供 6 Read IOPS/GB + 6 Write IOPS/GB + 0.28MB/sec/GB throughput；以 10GB 的 disk 來說會是 60 Read IOPS + 60 Write IOPS + 2.8MB/sec throughput。

AWS 這邊代號是 `standard`，價錢是 \$0.08/GB，另外 IOPS 每 1M 個 IOPS 也要收 \$0.08，如果是拿來開機的話還好，但如果是有應用在上面操 IOPS 的話就不太便宜了。

GCP 這邊是 Standard Persistent Disk (Standard PD)，價錢是 \$0.04/GB，效能上提供 0.75/GB Read IOPS + 1.5/GB Write IOPS + 0.12MB/sec/GB throughput；以 10GB 的 disk 來說會是 7.5 Read IOPS + 15 Write IOPS + 1.2MB/sec throughput。

At both the Tweet level and the account level, we will remove any free promotion of prohibited 3rd-party social media platforms, such as linking out (i.e. using URLs) to any of the below platforms on Twitter, or providing your handle without a URL:

• Prohibited platforms:
• Facebook, Instagram, Mastodon, Truth Social, Tribel, Post and Nostr
• 3rd-party social media link aggregators such as linktr.ee, lnk.bio

Hacker News 的討論上面，Paul Graham 有回應 (帳號是 pg)，他又提出了一些猜測與見解，包含了他覺得這個新政策會被收回：「Paul Graham is leaving Twitter for now (twitter.com/paulg)」。

I'm not leaving Twitter. It seems more likely than not that Elon will reverse the ban on links to other social media sites. I just don't want to hang out there in the meantime. Plus given the way things are going, it seemed like a good time to learn about alternatives.

I still think Elon is a smart guy. His work on cars and rockets speaks for itself. Nor do I think he's the villain a lot of people try to make him out to be. He's eccentric, definitely, but that should be news to no one. Plus I don't think he realizes that the techniques that work for cars and rockets don't work in social media. Those two facts are sufficient to explain most of his behavior.

He could still salvage the situation. He's the sort of person it would be a big mistake to write off. And I hope he does. I would be delighted to go back to using Twitter regularly.

## GCP 推出 AlloyDB，一套相容 PostgreSQL 協定的資料庫服務

AlloyDB was also two times faster for transactional workloads than Amazon’s comparable service.

AlloyDB 設計上是考慮了 HTAP (Hybrid transactional/analytical processing) 的使用，所以同時可以提供 OLAP 與 OLTP 的應用：

[...] This makes AlloyDB a great fit for business intelligence, reporting, and hybrid transactional and analytical workloads (HTAP).

(這邊剛好回顧一下 "Amazon’s comparable service" 這段，不確定他是跟 Intel-based 比還是跟 ARM-based 比，畢竟 ARM 除了比較便宜外，還有效能的提昇)

## 搜尋影片的串流平台

Hacker News Daily 上看到「Show HN: API to query catalogs of 20 streaming services across 60 countries (movieofthenight.com)」這個，但這個服務反而不是重點，有許多人發現裡面錯誤率頗高，而且也沒有台灣的資料，反倒是裡面有人提到 JustWatch 這個服務看起來比較好用...