Not news, but: Fediverse (active): https://t.co/R9bvhzseVX / @dougall@mastodon.social Bluesky:

My view. https://t.co/wojAwRsg23

very sad that cohost is going. maybe its unique genre of posting, the css crime, will go with it. well, i'm glad that there's this worthy goodbye to it

After 27 years of providing in-depth coverage of the amazing world of PC and mobile hardware, AnandTech is saying farewell. We want to thank everyone from the AnandTech community for their support and passion for what we’ve done over the years https://t.co/3EGh4FJguE

CRC checksum is not a cryptographic hash function. It is easy to add a few bytes of garbage at the end of a file so that the CRC of the entire file becomes a desired value. But did you need to do that in the linker? Yes. https://t.co/Wu7gU2j5Hb

Analysis of the M4 by @techinsightsinc „key design choice in the Apple M4 is the use of TSMC's high-performance standard cell library for CPU1“ …“UHD libraries for GPU and CPU2“ aka TSMC FinFlex CPU1: P-Cores CPU2: E-Cores https://t.co/DMuS9Tkg3h

Starting to think this @Arm thing might have legs

Intel is now essentially following Apple's design philosophy, with an integrated memory architecture, a large front-end, a large L1 cache, removal of SMT, 4+4 cores

"Vulkan 1.3 on the M1 in 1 month", by Alyssa Rosenzweig https://t.co/ZyS9EaMBfK

With Lisa's keynote over, I will casually drop this to confirm that y-cruncher v0.8.5 will get #Zen5 optimizations with it's big and fat #AVX512. Release date still ETA - pending final silicon. https://t.co/dohnqbg7Ex ST and BBP will be where the light shines the brightest.

@dyaroshev @geofflangdale @HaroldAptroot @FUZxxl FWIW, I've also given it a decent write-up at https://t.co/lQoW9gnNMr. Only notable change is putting maskz on the shuffle rather than on the permutexvar, as https://t.co/V5fqZLYP4G suggests maskz costs a few cycles, so might as well put it on the cheaper of the two instructions.

Fantastical deep information about optimizing for Apple Silicon can now be found here:

Example of why power draw is a poor proxy for how well a program is utilizing your hardware (i7-7700K CPU in this case). In one case, execution port use is heavy and IPC is high, so the CPU's pipeline is running at high utilization. However power draw is low. Second case...

New blog post: "Entropy decoding in Oodle Data: x86-64 6-stream Huffman decoders"

Here is @Qualcomm #Oryon single threaded performance: 👉 VS Apple M2: 14% faster at 30% less power 👉 VS Intel i9-13980HK: 1% faster at 70% less power

oh my god

* 8-bit floating-point, separately configurable to use E5M2 or E4M3 formats for different operands Scalar highlights: * "Checked pointer" add/sub/multiply-add/multiply-sub – for faster/safer top-byte-ignore? * New "enhanced" pointer authentication ops.

New Arm extensions just dropped: https://t.co/9tQFfJNPOH SIMD highlights: * LUTI2 and LUTI4 – powerful pshufb/tbl-like instructions, decoding bit-packed 2-bit or 4-bit indices to 8-bit or 16-bit values * Floating-point absolute min/max