Hacker News new | past | comments | ask | show | jobs | submit | bmh's comments login

They decided to use the silicon space for other things. The Pi5 CPU is powerful enough to decode most h264 streams in software. The Pi5 has an h265 hardware decoder.


Shame. Cause h265 is problematic wrt licensing etc. I would have preferred that they kick out h265 and used the silicon for h264.

And maybe the pi5 has enough cpu to do the decoding. But how much would be left over for more interesting AI tasks ?


Well it's not fast enough for most AI tasks as it is, you'd need to offload that bit to a pcie accelerator.


Not true actually. If by offloading you mean to a device that uses a really small model with 8-bit quantized weights you are not actually solving anything.


I agree 100%


sure it's powerful enough but I'd rather it built into the hardware for energy efficiency. But I guess we can't have our cake.


You need active cooling if you want to extract all the performance out of it. The standard cooling kits sold by Raspberry Pi are neat and cheap though, so not a problem in my experience. They have two options: A case with a fan on top, and a small heatsink + fan combo. The fans run at variable speed, so generally pretty quiet.


Who actually uses multipath?


It's quite popular actually, even there's an market for openwrt based mptcp routers, sometime also called SDN routers.

Say have eth, wifi(2.4+5Ghz+6Ghz) and cellular network, you may want to use them fully all together, there's a way for mptcp to shine;

or you have multiple proxies, and wants to improve reliability, you can also use mptcp.


And even commercial broadband services.... In full disclosure I worked as a supplier to implement MPTCP on the CPE for BT at the launch of their service (time marker 1:48 starting point) : https://youtu.be/eMKAFWy6940


MPTCP has been in use at massive scale on iOS for years now for core Apple services and any iOS application that uses Network.framework can leverage it too.


Anyone who uses SIRI is probably a great example.


Apple


It's interesting that the standard "K" (number of elements with a shared scale) is 32. That seems to imply that the neural network will somehow learn to group weights at those 32-element boundaries. Does anybody understand how that works? I mean, what is the mechanism that naturally causes the model to group weight scales into those K-element clusters?


There is no mechanism per-say, it's more of a bit space vs quality issue. You could think of MX4 with an 8 bit exponent scale as a 12 bit number if the block size is one, "MX12" with E10M1. You can share the scale with some error per element in a block, with that error going up as you increase the size of the block. As the block size is increased, the effective size per element goes down and the hardware implementation gets smaller/cheaper.


Haha. I first read that headline as "Hans Niemann dies using vibrating sex toy"


"It took our doctors two hours just to get the smile off his face."


Me too. What a way to go.


"Modulation can be turned off on alternate scans to identify velocity using unmodulated carrier frequency shift. This allows range and velocity to be found with just one radar set."

quote from https://navtechradar.com/explore/fmcw-radar/#:~:text=Frequen....


The short answer is:

Yes, you must disable CPU frequency scaling in your BIOS if you're doing this kind of work (i.e. building cryptographic primitives that don't leak information via timing).

See https://timing.attacks.cr.yp.to/overclocking.html


TL;DR: The December 5th experiment was about proving to ourselves and our adversaries that we still understand how to build and validate nuclear weapons, despite the ban on testing.


As far as I know, computer models and simulations have been running the show for a while, that after decades of testing and modeling.


It's conceivable that within a year or two, Copilot will be good enough to make the pain of the CMake language much more bearable.


A year or two might be enough for the c++ community to sufficiently complect the build process enough to befuddle any AI


A prime specimen for "impossible to tell if sarcasm".


Windows Defender says that the binary has Trojan:Win32/Wacatac.H!ml inside it.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: