zeusk

5 ms·

1.
by zeusk 18d ago
Hey, just don't store your pagetables in PIM capable sections :)
2.
by zeusk 18d ago
That really doesn't make sense because your system is already using memory that needs to be physically contiguous but mapped virtually. Also, would you say pinned memory is not part of the virtual memory system?? You're only restr
3.
by zeusk 18d ago
why would it? the parent OS can already handle physically contiguous allocations so these should be no different (with the exception that a separate interface can be used to do compute over these buffers/pages).
4.
by zeusk 22d ago
PS4 is as close as it gets to the colloquial computer; it's a AMD APU running BSD with homebrew linux builds available. And if not, would you consider a raspberry pi a computer? or your iphone a computer?
5.
by zeusk 22d ago
also https://velaura.ai/
6.
by zeusk 2mo ago
Isn't the whole point of attention, some context comprehension in the stochastic parrot machine.
7.
by zeusk 2mo ago
Well their whole model is a stolen art collection :)
8.
by zeusk 2mo ago
Hey, Silicon Valley is back to it's roots :)
9.
by zeusk 3mo ago
You can get the user's head position using WorldTrackingProvider, that's enough for xeyes to follow you across the room. There's even a sample app close enough https://developer.apple.com/documentation/vi
10.
by zeusk 3mo ago
what's the context behind this?
11.
by zeusk 5mo ago
> copper can barely support 10G and is terribly power hungry when it does that. AFAIK, thunderbolt cables are also copper - so what trickery do they use for supporting USB4-80? i believe both connectors use differential pair wires for si
12.
by zeusk 5mo ago
Half life alyx and their push for openVR has made a big impact in that part of the gaming world. But yeah, their games are just as filled with lootbox, crates, skin garbage as other low effort money grabs; saving grace being its all cosmeti
13.
by zeusk 5mo ago
ordering lock acquisition is a tested strategy to avoid deadlocks; so locking the cache lines sorted by PA would cover that?
14.
by zeusk 5mo ago
This is very spot on in my experience.
15.
by zeusk 6mo ago
> My personal dream is that vms would support pci pass through and so you can just spin up a Linux vm and let it drive the gpus. SR-IOV is just that? and is well supported by both Windows and Linux.
16.
by zeusk 8mo ago
They’re used by the internal register renamer/allocator so if it sees you’re storing the results to memory then reusing the named register for a new result - it will allocate a new physical register so your instruction doesn’t stall fo
17.
by zeusk 8mo ago
This has already been tried :) iirc, in the 2016 a quadcore intel cpu ran the original crysis at ~15fps
18.
by zeusk 9mo ago
Get the DGX Spark computers? They’re exactly what you’re trying to build.
19.
by zeusk 9mo ago
the TB5 link (RDMA) is much slower than direct access to system memory
20.
by zeusk 10mo ago
Nvidia has been investing in confidential compute for inference workloads in cloud - that covers physical ownership/attacks in their thread model. https://www.nvidia.com/en-us/data-center/solutions/confid
21.
by zeusk 11mo ago
Yes, but parent comment definitely implied they weren't talking about people running on the latest and best out there. Even the middle-grade smartphones today are leaps and bounds better than the atom from 2010.
22.
by zeusk 11mo ago
> And yet, if I go to Youtube or just about any other modern site, it takes literally a minute to load and render, none of the UI elements are responsive, and the site is unusable for playing videos. Why? I'm not asking for anything
23.
by zeusk 11mo ago
What do you mean by that? Most syscalls are still interrupt based.
24.
by zeusk 11mo ago
Double buffering a 4K 4bpp framebuffer itself is 64mb
25.
by zeusk 11mo ago
> same hardware as the higher end models but needs a firmware bit flip Is this firmware bit flip known? couldn't find anything off google.
26.
by zeusk 1y ago
AUM is not theirs to keep; and market cap is a very deceitful metric especially for banks where liabilities dwarf the market cap.
27.
by zeusk 1y ago
Well so is the snapdragon X elite, including the older snapdragons (anyone remember scorpion cores on QSD8x50?)
28.
by zeusk 1y ago
The real issue with packed structs and bitfields happens when concurrency gets involved. Majority of modern CPU caches are private and only allow one core to hold the cache line - so it creates more false dependencies when cores are trying
29.
by zeusk 1y ago
I've used kd/windbg at Microsoft, lldb at Apple, gdb at Intel; WinDbg(Next) and kd are still my absolute favorite. Coolest feature of windbg is time travel debugging - https://learn.microsoft.com/en-us/windows
30.
by zeusk 1y ago
but afaik, we still don't have a break on xbox one released 2013 - 12 years later.

More ›