I'm not sure I'd call 1.5min processing compared to 3-5 min processing as "interactive".
>I can't overclock it either Except you can overclock v3 :)
How do I connect it to a local llama.cpp instance?
Note that EOD is both rare (of all dementia cases) and highly inheritable.
https://github.com/sdbds/flash-attention-for-windows/release... - FA2 binaries for you
Please refrain from posting an AI slop.
Still, in the year 2025, the only offline player that stays in menu bar and doesn't clutter the dock is Vox.
Believing in hints from Apple about software? Sweet summer child.
Keeping internet locked, you mean?
Nah. I want fixes in macos, not "boring" nor "shiny updates".
Shitty shits. Literally. As for metabolic acidoses, rates are extremely low (<10 cases per 100,000 patient-years, per https://pubmed.ncbi.nlm.nih.gov/26773926/), so it's almost a non-concern.
https://bugs.php.net/bug.php?id=53632 Never, for sure.
Hydras tend to disagree.
>Hopefully the last breaking change. Every year's cope with systemd.
I've been living with sysvinit up until Debian 11. Then it became unusable with lxc containers :(, so I had to bite the bullet. But for the basic system it indeed works.
Let me decide for myself what must I love.
Just use double Ctrl+A to pass it down: bind C-a send-prefix
Memory bandwidth is only relevant for comparing LLM performance. For image generation, the limiting factor is compute, and Apple sucks with it.
Grab one from Aliexpress while you stil can)
I have a 100TB array setup with LVM. I've thought (twice, each time I expanded this array) about using ZFS. It was ruled out for three things: - there had been no raidz expansion (but now is, so this point is crossed) -…
It doesn't matter to the public's perception.
`xhost +` Problem solved
For anyone wondering, have fun reading responses to those: - https://github.com/openssl/openssl/issues/20286 - https://github.com/openssl/openssl/issues/18814 - https://github.com/openssl/openssl/issues/16791 -…
Their maintainers have a rather dubious anti-user history with opt-out analytics and forced auto updates to said analytics.
>I wish you didn't have to buy Xeon But that's the whole point of Intel's market segmentation strategy - otherwise their low-tier workstation Xeons would see no market.
I'm not sure I'd call 1.5min processing compared to 3-5 min processing as "interactive".
>I can't overclock it either Except you can overclock v3 :)
How do I connect it to a local llama.cpp instance?
Note that EOD is both rare (of all dementia cases) and highly inheritable.
https://github.com/sdbds/flash-attention-for-windows/release... - FA2 binaries for you
Please refrain from posting an AI slop.
Still, in the year 2025, the only offline player that stays in menu bar and doesn't clutter the dock is Vox.
Believing in hints from Apple about software? Sweet summer child.
Keeping internet locked, you mean?
Nah. I want fixes in macos, not "boring" nor "shiny updates".
Shitty shits. Literally. As for metabolic acidoses, rates are extremely low (<10 cases per 100,000 patient-years, per https://pubmed.ncbi.nlm.nih.gov/26773926/), so it's almost a non-concern.
https://bugs.php.net/bug.php?id=53632 Never, for sure.
Hydras tend to disagree.
>Hopefully the last breaking change. Every year's cope with systemd.
I've been living with sysvinit up until Debian 11. Then it became unusable with lxc containers :(, so I had to bite the bullet. But for the basic system it indeed works.
Let me decide for myself what must I love.
Just use double Ctrl+A to pass it down: bind C-a send-prefix
Memory bandwidth is only relevant for comparing LLM performance. For image generation, the limiting factor is compute, and Apple sucks with it.
Grab one from Aliexpress while you stil can)
I have a 100TB array setup with LVM. I've thought (twice, each time I expanded this array) about using ZFS. It was ruled out for three things: - there had been no raidz expansion (but now is, so this point is crossed) -…
It doesn't matter to the public's perception.
`xhost +` Problem solved
For anyone wondering, have fun reading responses to those: - https://github.com/openssl/openssl/issues/20286 - https://github.com/openssl/openssl/issues/18814 - https://github.com/openssl/openssl/issues/16791 -…
Their maintainers have a rather dubious anti-user history with opt-out analytics and forced auto updates to said analytics.
>I wish you didn't have to buy Xeon But that's the whole point of Intel's market segmentation strategy - otherwise their low-tier workstation Xeons would see no market.