re: Markets may be turbulent tomorrow due to contraction in the AI sector

quote:

---

But please stop acting like you understand stuff like this.

---

I understand it fine. Show me proof that Deepseek used A100 chips that are banned in China.

quote:

---

But you don't take the time.

---

I spent hours researching and watching video of experts explaining how Deepseek accomplished more efficient AI for less money and resources. But you are telling me those are all lies and it didn't happen OK Boomer

quote:

---

A pretty good thread that walks through some of the techniques

---

Nice. Appreciate the link

quote:

---

I spent hours researching and watching video of experts explaining how Deepseek accomplished more efficient AI

---

Oh dude.
You spent hours researching and watching video of experts explaining??
You're not helping yourself.
You just aren't.

quote:

---

But you are telling me those are all lies

---

No.
I'm telling you those are not all truths.

quote:

---

OK Boomer

---

You've directed your presumed "Boomer" invective at me several times now.
It's odd, because I couldn't care less whether you label me Boomer, or Silent, or Greatest, or X'er, or Millennial. I am who I am.

and ...

... you are who you are.

On the topic of the hardware and compute - or "is the $6M figure real or did they really train on billions worth of hardware?!?"

Article here

Here's the good part below. There's skepticism, but the good news is it's verifiable if someone can take the code and dataset, and replicate their training runs.

quote:

---

In the DeepSeek-V3 paper, DeepSeek says that it spent 2.66 million GPU-hours on H800 accelerators to do the pretraining, 119,000 GPU-hours on context extension, and a mere 5,000 GPU-hours for supervised fine-tuning and reinforcement learning on the base V3 model, for a total of 2.79 million GPU-hours. At the cost of $2 per GPU hour – we have no idea if that is actually the prevailing price in China – then it cost a mere $5.58 million.

The cluster that DeepSeek says that it used to train the V3 model had a mere 256 server nodes with eight of the H800 GPU accelerators each, for a total of 2,048 GPUs. We presume that they are the H800 SXM5 version of the H800 cards, which have their FP64 floating point performance capped at 1 teraflops and are otherwise the same as the 80 GB version of the H100 card that most of the companies in the world can buy. (The PCI-Express version of the H800 card has some of its CUDA cores deactivated and has its memory bandwidth cut by 39 percent to 2 TB/sec from the 3.35 TB/sec on the base H100 card announced way back in 2022.) The eight GPUs inside the node are interlinked with NVSwitch es to created a shared memory domain across those GPU memories, and the nodes have multiple InfiniBand cards (probably one per GPU) to create high bandwidth links out to other nodes in the cluster. We strongly suspect DeepSeek only had access to 100 Gb/sec InfiniBand adapters and switches, but it could be running at 200 Gb/sec; the company does not say.

We think this is a fairly modest cluster by any modern AI standard, especially given the size of the clusters that OpenAI/Microsoft, Anthropic, and Google have built to train their equivalent GPT-4 and o1, Claude 3.5, and Gemini 1.5 models. We are very skeptical that the V3 model was trained from scratch on such a small cluster. It is simply hard to accept until someone else repeats the task. Luckily, science is repeatable: There are companies with trillions of curated tokens and tens of thousands of GPUs to see if what DeepSeek is claiming is true. On 2,048 H100 GPUs, it would take under two months to train DeepSeek-V3 if what the Chinese AI upstart says is true. That’s pocket change for the hyperscalers and cloud builders to prove out.

---

AI is a scam

quote:

---

AI is a scam

---

Negative!

quote:

---

Nvidia losses a record $600 billion in 1 day

NC Boomer: It's all Lies!!

---

quote:

---

NVDA stock price plummeted 17% to close at $118.58. It was the stock’s worst day on the market since March 16, 2020.

---

FYI, I sold a bunch of NVDA on Friday at $146. I bought $50K at $117.60 today. Small position, but the downward move made no sense.

Where did you put your money?????

my NVDIA holdings took an arse whooping. Good thing I like bologna on white bread sandwhiches.