HNHacker News
TopNewBestAskShowJobs

shaklee3

3,013 karma · joined February 21, 2015

submissionscomments
shaklee3··on The GPU is not always faster
To save some time for people, the title is absolutely incorrect. The GPU is significantly faster for this test, but the author is measuring PCIe bandwidth from a PCIe generation of 10 years ago. If instead they had used pcie5 the bandwidth would be quadrupled.
shaklee3··on Launch HN: Regatta Storage (YC F24) – Turn S3 into a local-like, POSIX cloud FS
The public sector is typically air-gapped, so not really.
shaklee3··on Wait Until 8th
This is exactly what the Fitbit Ace LTE does. So far it's been great since my daughter has no desire to ask anything else on it besides increase physical activity to get prizes
shaklee3··on Writing in Pictures: Richard Scarry and the art of children's literature
Like the giving tree. That one hits deep
shaklee3··on The AI Investment Boom
Most data center GPUs do not have game rendering hardware in them.
shaklee3··on Bluesky grew by 10% in just two days this week, passing 12M accounts in total
I can't tell if you are being facetious, but that's not how spikes work. Just like Mastodon and threads, it's a temporary spike and will go back to its normal growth in less than a week.

I had several people I follow on Twitter make a grandiose speech about going to Mastodon after musk bought Twitter, but almost all of them came back because nobody uses it. The ones that didn't come back have less than 5% of the followers they used to have.

shaklee3··on Starlink direct-to-cell enabled for hurricane helene emergency messaging
I'm not sure what you are getting at. Each carrier would have to make a deal with SpaceX. Doesn't matter if they share the same tower. If SpaceX is interfering with their frequency band, they are going to file a complaint with the FCC
shaklee3··on Starlink direct-to-cell enabled for hurricane helene emergency messaging
Right, but carriers without deals (att) won't
shaklee3··on Starlink direct-to-cell enabled for hurricane helene emergency messaging
They are, but that doesn't mean people have to use them that way. They will get use all over and not just in rural areas
shaklee3··on Starlink direct-to-cell enabled for hurricane helene emergency messaging
That's not how satellites work. The beam is far larger than a cell tower, so they're absolutely going to be transmitting where there are towers.
shaklee3··on Were RNNs all we needed?
See my comment above. It's definitely doable and very fast.
shaklee3··on Were RNNs all we needed?
Yes. See this paper: http://cs.txstate.edu/~mb92/papers/asplos18.pdf

And things have improved a lot since then.

shaklee3··on Visual Studio Code is designed to fracture (2022)
The bottom of the screen says exactly what you're connected to, so I'm not sure what they're referring to.
shaklee3··on Ask HN: What are you working on (September 2024)?
Thanks for the interest. I will post it publicly soon, but right now it's on my private github. Basically it's a small python app running on a raspberry pi with 2 USB QR scanners. It uses sqlite with a database of QR codes assigned to the students. They wear lanyards with those codes on them. When they start running it logs their start, and every lap they can again. It tracks personal records, total laps, laps per day, and time per lap.

We've received a lot of positive feedback from the kids compared to the old system that was using proprietary software. The problem with that software is it used cell phones for scanning codes, which is very error-prone and takes a long time to register on some phones.

shaklee3··on Ask HN: What are you working on (September 2024)?
My kids' elementary school has a running club that I manage. In past years they pay a subscription to a company that's quite expensive to track laps. I decided to implement some software that's better and runs on an rpi.

It's working very well, and the kids like it more because it has audio feedback on laps and time.

Besides the cheap USB qr scanners there's no expenses either.

shaklee3··on Real-time Linux is officially part of the kernel
You may need to modify the jiffy frequency
shaklee3··on David Chang on the long, hard, stupid way
I don't get your point. There are many other companies using the name chili crunch. He was not the first. But he tried to trademark it to stop all the others.
shaklee3··on David Chang on the long, hard, stupid way
He also tried to sue all the other small businesses using the chili crunch name. I stopped buying his after that.
shaklee3··on QUIC is not quick enough over fast internet
This is GPUDirect. GPUDirect is the technology that enables any third-party device to talk to a GPU (like a NIC).

> but probably pretty awful over the Internet. In the DC though it will be beaten by real GPUDirect on RDMA naturally

It's being used in many places successfully over the internet. RDMA is fine, but completely breaks the abstraction of services. In many places you do not want to know who is sending or what address to send/receive.

shaklee3··on QUIC is not quick enough over fast internet
Again, you can do what I said. You may be using different terminology, but you can do GPUdirect in dpdk without rdma
shaklee3··on QUIC is not quick enough over fast internet
You can do GPUdirect over the Internet without RDMA though.
shaklee3··on Show HN: Wealthfolio: Private, open-source investment tracker
The amount of work you'd have to do to correct a stolen identity is far less than the amount of work to log in to 10 accounts periodically,dump the data, format it, import it, and fix any issues
shaklee3··on Show HN: Wealthfolio: Private, open-source investment tracker
This is a huge waste of time and not worth it. The amount of hours you spend fixing errors manually vs the small risk your info gets out I don't think is worth it.
shaklee3··on GPU utilization can be a misleading metric
I'd be curious to see how you can do it. Try launching an fft of any size and batches and see if you can hit 100%
shaklee3··on GPU utilization can be a misleading metric
I don't mean a single fft. I mean the fft algorithms are inherently not going to use the GPU at 100% utilization by any metric.
shaklee3··on GPU utilization can be a misleading metric
This is not true. Lots of algorithms simply can't use 100% of the GPU even though they're written as optimal as possible. FFT is one.
shaklee3··on Making your own hot sauce
Tabasco, tapatio, and Cholula are very common in Mexican restaurants in CA
shaklee3··on Run CUDA, unmodified, on AMD GPUs
A 20k GPU will be passively cooled and you'll need a real server for that. Even the old MI210 another poster sent is passive.
shaklee3··on So you want to rent an NVIDIA H100 cluster? 2024 Consumer Guide
RoCE does not encapsulate infiniband
shaklee3··on So you want to rent an NVIDIA H100 cluster? 2024 Consumer Guide
If you use ucx it does all that automatically without you choosing.
← PreviousPage 2 of 34Next →