DiabloD3

5 ms·

31.
by DiabloD3 2mo ago
One of the biggest fixes I've seen is just getting rid of traditional sampling. But first, let me say something about quantization, just to get this out of the way. Like, lets say you already did the sane thing, your model[1] is alread
32.
by DiabloD3 2mo ago
This is a problem with long context models. To put it as simple and as bluntly as possible: just because they claim you can use 1M tokens in your context doesn't mean its true and you should do that. Due to extreme quantization of mode
33.
by DiabloD3 2mo ago
That is called a hallucination.
34.
by DiabloD3 3mo ago
I don't understand the joke. Intel has already stated their deal with Nvidia, this came out in December. If you think there will be shipping Druid products, I guess you can find out sometime in 2028.
35.
by DiabloD3 3mo ago
What GPU? Its usually Nvidia owners that complain about lackluster Wine/Proton performance.
36.
by DiabloD3 3mo ago
Battlemage was still while Pat Gelsinger was CEO, and he was the only real champion of that product, understanding its place in Intel's overall strategy. He doesn't work there anymore, and neither does a significant fraction of th
37.
by DiabloD3 3mo ago
There will be no future Xe GPUs. Druid, the 4th generation, has been shelved entirely, and all consumer DGPU products, possibly _all_ DGPU products, have been killed for Celestial. Its likely the only products shipping with Celestial will b
38.
by DiabloD3 3mo ago
A lot of games no longer run properly in Windows, and the only way to keep playing them is with Wine/Proton. Windows performance is also often worse than Wine/Proton's.
39.
by DiabloD3 3mo ago
GPU clusters have, largely, no actual value. If anything, you might have to pay to have them disposed of, they don't really have any meaningful used eBay market outside of the randos that want to do high end extreme local inference in
40.
by DiabloD3 3mo ago
Neat, can't wait to see a llama.cpp PR for this.
41.
by DiabloD3 3mo ago
OnePlus basically already went under a few years back. They merged with Oppo, and ever since its just been rebranded Oppo phones... they're fine, they work, but they're not the magic OnePlus was.
42.
by DiabloD3 3mo ago
Skip Samsung, they tend to not honor warranties and are ROM swap resistant. Google is okay for now (I recently got a Pixel 10, its fine, it can run Graphene). Next year Motorola flagships will also have official Graphene support.
43.
by DiabloD3 3mo ago
Ironically, this is what people claim AI can do with a snap of the fingers. Should be real simple if the HN AI echochamber is right, right?
44.
by DiabloD3 3mo ago
I love how people say things like "extension spaghetti", as if all other non-standard APIs have the same problem: hardware gets new features that people want to use from that API, API gains extension to use that hardware feature.
45.
by DiabloD3 3mo ago
Weird, most people have the exact opposite experience. Having to deal with closed source opaque poorly documented stacks sucks.
46.
by DiabloD3 3mo ago
Weird, since the most used open source inference engine is faster on Vulkan on platforms that offer multiple options, with the sole exception being Nvidia, due to poor Nvidia driver quality (which I am forced to assume is intentional, Nvidi
47.
by DiabloD3 3mo ago
Its easier to just get rid of your legacy code entirely and use Vulkan for compute, or have your compiler emit SPIR-V directly. No reason to tie yourself to Nvidia's moat.
48.
by DiabloD3 3mo ago
You'd have less problems with 27B, btw.
49.
by DiabloD3 3mo ago
I was expecting Linda Hamilton, but apparently more than one set of twins were in there.
50.
by DiabloD3 3mo ago
I still find it funny how the AI Bros bribed Trump, and he immediately turned around and cost them significant revenue. As they say, qui cum canibus concumbunt cum pulicibus surgent.
51.
by DiabloD3 3mo ago
Serpent Lake (and its siblings) cover both consumer desktop and laptop. Arc's team was disproportionately effected by the layoffs. A lot of the major engineers that worked there changed their Linked in details to list Nvidia, AMD, etc.
52.
by DiabloD3 3mo ago
HN front page circa September through December of last year. Its part of the $5B investment into Nvidia, and the new CEO, Lip-Bu Tan, happily did whatever Nvidia told him to do. Nova Lake (Series 4) is already too late in development to swi
53.
by DiabloD3 3mo ago
Yep, but the flip side is you can get two B70s for the price of a single 3090 (MSRP, obviously; 3090s used go for about the same as a new B70), and they're true 2 slot, so they can fit on x8/x8 consumer boards fine. The side effec
54.
by DiabloD3 3mo ago
That is incorrect. They both have GDDR6. The B70 has 256 bit it bus at a clock speed of 2375mhz (608 GB/s), the 3090 has a 384 bit bus at a clock speed of 2438mhz (936 GB/s). It isn't slower, it just has less channels, ie, it
55.
by DiabloD3 3mo ago
Sundaras aren't great, but are sufficient for this. Scarlett series is also good enough. At 48khz, with a tone generator (not rando Youtube videos or anything), you should be able to clearly hear up to around 16khz (as in, can tell pit
56.
by DiabloD3 3mo ago
The problem with RocksDB is Facebook uses it, and has a similar system that replicates data across the world... and still faces unrecoverable data loss because of RocksDB.
57.
by DiabloD3 3mo ago
Yeaaaaaaaaaahhhh.... you might be a tiny bit deaf. OTOH, we know nothing of your audio equipment nor how its setup.
58.
by DiabloD3 3mo ago
Thats a weird way for Sony to announce the end of the Playstation
59.
by DiabloD3 3mo ago
Interesting, they used to be the largest ZFS user. Hard to Google for it without getting AI slop on it, but apparently they built their own stack in 2019. Not sure I like their solution, "Meta-data is persisted in RocksDB databases usi
60.
by DiabloD3 3mo ago
In one pool, sure. You can have more than one pool.

More ›