DiabloD3

16 ms·

301.
by DiabloD3 1y ago
Optimal viewing distance for most media (including text found in programs and websites made in the "Vista" and "Retina OSX" era to today, but also all movies and TV shows cut for modern 16:9 displays) is between 30 (SMPT
302.
by DiabloD3 1y ago
I can almost see this complaint with poorly designed (ie, common and cheap) polarizers in LCDs and/or non-IPS/IPS-like LCDs at normal DPI. However, that doesn't really work with OLEDs or MicroLEDs at any DPI, or any HighDPI I
303.
by DiabloD3 1y ago
I don't get the use case of curved displays. I game, games do not support spherical "fisheye" rendering, thus the entire product concept is effectively dead, as no software supports correcting for these, since high end monito
304.
by DiabloD3 1y ago
I suggest figuring out what your configuration problem is. Which llama.cpp flags are you using, because I am absolutely not having the same bug you are.
305.
by DiabloD3 1y ago
Same calculation, basically. Any given ~30B model is going to use the same VRAM (assuming loading it all into VRAM, which MoEs do not need to do), is going to be the same size
306.
by DiabloD3 1y ago
I don't load all the MoE layers onto my GPU, and I have only about a 15% reduction in token generation speed while maintaining a model 2-3 times larger than VRAM alone.
307.
by DiabloD3 1y ago
Thats not a meaningful question. Models can be quantized to fit into much smaller memory requirements, and not all MoE layers (in MoE models) have to be offloaded to VRAM to maintain performance.
308.
by DiabloD3 1y ago
But you just described the existing Apple ecosystem!
309.
by DiabloD3 1y ago
Anthropic needs to continue burning cash and goodwill in hopes they extend the runway to IPO. They do not seem to care at all that what they're peddling is just elaborate smoke and mirrors.
310.
by DiabloD3 1y ago
(Everyone say it with me) Thats a weird way for Anthropic to announce they're going out of business.
311.
by DiabloD3 1y ago
Oh no, Warner Bros has figured out an infinite money machine!
312.
by DiabloD3 1y ago
There is a difference between recommended and max achievable. Zen 5 can hit that (and that's what I run), and Arrow Lake can also. The recommended from AMD on Zen 4 and 5 is 6000 (or 48x2), for Arrow Lake is 6400 (or 51.2x2); both of t
313.
by DiabloD3 1y ago
"No." DDR5-8000 is 64GB/s per channel. Desktop CPUs have two channels. PCI-E 5.0 in x16 is 64GB/s. Desktops have one x16.
314.
by DiabloD3 1y ago
Thats a weird way for Coinbase to announce they're going out of business. (This is almost my catch phrase at this point, its kinda wild)
315.
by DiabloD3 1y ago
Spectre and Meltdown send their regards.
316.
by DiabloD3 1y ago
I would also like to mention Chezmoi. I switched to it after maintaining a custom written POSIX sh script that kept getting further and further out of hand. I don't see myself going back to a self-maintained one unless I'm writing
317.
by DiabloD3 1y ago
And Intel keeps designing chips the "right" way, and they keep digging their own grave. Its almost like Intel spends more money on PR and stealth marketing than they do R&D.
318.
by DiabloD3 1y ago
I've been seeing it more and more, especially with vendors that don't speak a western language on their spec sheets. Everyone can tell what L1$ means, but what would L1 缓存 mean?
319.
by DiabloD3 1y ago
Latex doesn't work here ;)
320.
by DiabloD3 1y ago
So, its just a clone of current gen Zen5C products, and will fall behind Zen6-based products.
321.
by DiabloD3 1y ago
That's a weird way for Apple to announce Xcode is being sunset.
322.
by DiabloD3 1y ago
Wait, the Anubis people _didn't know_ 3 core machines were sold for years? AMD was famous for it!
323.
by DiabloD3 1y ago
It isn't a "bit of a joke" when the graphics teams behind normal desktop apps and browsers put more work into making sure the thing works on Intel than they have for AMD and Nvidia. They know what the majority of users use. N
324.
by DiabloD3 1y ago
Most things are developed for standardized APIs, not any specific vendor. For things that are hyperoptimized for a specific vendor, it is usually for AMD, not Nvidia, because most of that time spent optimizing is not done for enterprise com
325.
by DiabloD3 1y ago
Nvidia "working" on Linux requires binary drivers that often have severe issues, especially if your mainline stable kernel is too new. The new open source drivers that replace Novaeu aren't ready yet, and the closed source dr
326.
by DiabloD3 1y ago
I don't know a single person that bought a Framework and put Windows on it as the primary OS. I wish Framework would actually find out what their customers are using, maybe with a survey, because I'd be highly surprised if it was
327.
by DiabloD3 1y ago
Not really. Until modern times, Intel was the largest GPU manufacture, unless you include phones, then it was Qualcomm. Now it's AMD, between computers, consoles, and the datacenter. DGPUs for the desktop aren't really all that re
328.
by DiabloD3 1y ago
We don't, but TikTok and other pro-Russian/pro-Chinese/anti-West propaganda platforms sure as hell do. They made up a word for it, "manlet". People are, on average, fucking shitty people.
329.
by DiabloD3 1y ago
>"preponderance of evidence," Yes, which as I said above, is the act of distribution in most cases. If you don't distribute it or sell a service around it, it isn't worth my time to sue. >Napster cases, the RIAA ha
330.
by DiabloD3 1y ago
Kinda wish you guys didn't have an Nvidia product at all. It's only really useful for Windows users, but openly hostile and offensive to Linux desktop users.

More ›