HNHacker News
TopNewBestAskShowJobs

KaoruAoiShiho

3,254 karma · joined October 1, 2011

Looking for investors & partners: kaoruaoishiho@gmail.com
submissionscomments
KaoruAoiShiho··on Google is a monopoly – the fix isn't obvious
Can you explain what large enterprises and educational institutions would be paying Chrome for?
KaoruAoiShiho··on Show HN: I made a Note-Taking app for people who keep texting themselves
I just use google keep and apple notes, works fine for me.
KaoruAoiShiho··on Show HN: I am building an open-source Confluence and Notion alternative
Really wish a designer would go and help you out, this looks really good except for the rather dated look and feel.
KaoruAoiShiho··on AMD MI300x GPUs with GEMM tuning improves throughput and latency by up to 7.2x
Yeah, that makes sense. Here's another benchmark that shows trt being about 2-3x faster, 2x on fp16 and 3x on int4. https://blog.premai.io/prem-benchmarks/ Though to be fair it's 512 tokens and not 4k tokens. Would love to see more public testing done for all the different variations.
KaoruAoiShiho··on AMD MI300x GPUs with GEMM tuning improves throughput and latency by up to 7.2x
That's cool but a 25-30% improvement wouldn't make up for the 300% advantage LMDeploy and TRT has over vllm on the 70b benchmark. Where are your results for vllm vs lmdeploy and trt?
KaoruAoiShiho··on AMD MI300x GPUs with GEMM tuning improves throughput and latency by up to 7.2x
Is it really that close in production? The benchmarks I've seen show it's almost 3x slower compared to the SOTA. https://bentoml.com/blog/benchmarking-llm-inference-backends
KaoruAoiShiho··on HH70, the first high-temperature superconducting Tokamak achieves first plasma
Don't forget their best game tears of themis
KaoruAoiShiho··on AMD's MI300X Outperforms Nvidia's H100 for LLM Inference
Pretty bad benchmarks to the point of being deliberately misleading. They benchmarked vllm which is less than half the speed of the inference leader lmdeploy: https://bentoml.com/blog/benchmarking-llm-inference-backends

They also used Flywheel for AMD while not bothering to turn on Flywheel for Nvidia, which is crazy since Flywheel improves Nvidia performance by 70%. https://mk1.ai/blog/flywheel-launch

In this context the 33% performance lead by AMD looks terrible, and straight up looks slower.

KaoruAoiShiho··on Tinygrad 0.9.0
He's not wrong.
KaoruAoiShiho··on Show HN: Openpanel – An open-source alternative to Mixpanel
Do you have a way to automatically prune old data? I ran into the problem with posthog where the old data just overflowed my db and made it inoperable and they had no way to prune it.
KaoruAoiShiho··on The OpenAI board was right
It sounds nothing like johansson, like nothing at all.

https://www.reddit.com/r/singularity/comments/1cx1np4/voice_...

This whole controversy makes no sense and is obvious that the media is looking for something to attack openai with.

KaoruAoiShiho··on New exponent functions that make SiLU and SoftMax 2x faster, at full accuracy
This does help the gguf usecase of partial offloading to GPU? It'll help the CPU part be faster too?
KaoruAoiShiho··on PaliGemma: Open-Source Multimodal Model by Google
Isn't this outdated already. It's 2 models slapped together rather than natively multimodal.
KaoruAoiShiho··on GPUs Go Brrr
Eh if you're running in production you'll want something lower level and faster than pytorch.
KaoruAoiShiho··on Did GitHub Copilot increase my productivity?
When you review pull requests you're unable to get the mental picture. But it's different with copilot because it fits in your mental picture, you can compartmentalize what you're working on, and the copilot output can come in the form of small pieces that are easily understood and slot in the prepared places where you need it.

I agree 100% that reviewing human code is a lot of work but not with copilot IMO because of the immediacy of it all.

KaoruAoiShiho··on Did GitHub Copilot increase my productivity?
It's shorter, you're working directly with the code base you're already thinking about, and it's plugged into it rather than having to start a new mental task when you're looking at a pull request.
KaoruAoiShiho··on Using a LLM to compress text
I dunno, I'm not a lawyer, it might matter.
KaoruAoiShiho··on Using a LLM to compress text
Can this be used to save on token cost in LLM inferencing?
KaoruAoiShiho··on Using a LLM to compress text
Was it that NYT articles are routinely pirated on reddit comments and the like?
KaoruAoiShiho··on How to Make More Money Renting a GPU Than Nvidia Makes Selling It
Just adding some information, the article claims Tesla has 15k H100s, but they actually have 40k H100s and 85k by the end of the year. https://www.reddit.com/r/NVDA_Stock/comments/1cbwvnr/tesla_4...

As usual need to have a bit of doubt on the third party estimates.

KaoruAoiShiho··on The man who killed Google Search?
While entertaining it doesn't actually say anything about what the villain guy actually did, am I crazy? There's 2 serious charges he levied to google.

1. Ads look more like results.

2. Google results got more useless spam.

While 1 is kinda icky it's not that big of a deal, especially since I use an adblocker... and for 2 why does the author think this is the fault of google? Does shittier results increase in more people using google? I feel like it's the opposite, this doesn't seem right to me. Can it not just be that spammers and SEO freaks got more sophisticated and the problem got more challenging?

KaoruAoiShiho··on Show HN: I taught ChatGPT to find cool domain names – with availability check
People are self-interested. Electron only offers the maker of the tool value but not the end user. The end user finds AI extremely valuable and the costs are on the maker.
KaoruAoiShiho··on Nvidia CEO Jensen Huang announces new AI chips: ‘We need bigger GPUs’
But maybe you should. Once the software stack is ready for it there'll be more people since the performance gains are so massive.
KaoruAoiShiho··on Nvidia is now more valuable than Amazon and Google
Do you have any knowledge of what it runs on? I was pretty sure it was NVIDIA (accessed through cloud providers as a customer).
KaoruAoiShiho··on OpenAI drops ban on military tools to partner with The Pentagon
This is the sort of thing that becomes possible after dropping the ethical members of their board.
KaoruAoiShiho··on Game-icons.net: Free icons for your games
Now that we have genAI...
KaoruAoiShiho··on Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
Did you use claude?

https://www.anthropic.com/index/claude-2-1-prompting

KaoruAoiShiho··on Ask HN: How do I train a custom LLM/ChatGPT on my own documents in Dec 2023?
Not true.
KaoruAoiShiho··on Nvidia CEO: We bet the farm on AI and no one knew it
No that doesn't make sense because NVIDIA physically built this field with their own hands. So there's no way it could have "not happened" or "not been GPU friendly" lol.

Here is Jensen talking about building the industry 9 months before the founding of OpenAI: https://www.youtube.com/watch?v=_iBLoNG0qHk

KaoruAoiShiho··on Nvidia CEO: We bet the farm on AI and no one knew it
That makes no sense, there's no way AI "could never have happened".
← PreviousPage 7 of 34Next →