Viewing profile — zeusk
zeusk
HN member- Joined
- Mon, Jul 15, 2013, 8:09 AM UTC
- HN karma
- 2,382
- Public activity
- 1,088 items
- HN profile
- View on Hacker News ↗
About zeusk
No profile information was provided.
Recent public activity
-
comment
Comment #48931297
Isn't the whole point of attention, some context comprehension in the stochastic parrot machine.
-
comment
Comment #48866388
Well their whole model is a stolen art collection :)
-
comment
Comment #48866124
Hey, Silicon Valley is back to it's roots :)
-
comment
Comment #48615544
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.ap…
- comment
-
comment
Comment #48591566
what's the context behind this?
-
comment
Comment #47993282
> 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 b…
-
comment
Comment #47988997
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 othe…
-
comment
Comment #47731369
ordering lock acquisition is a tested strategy to avoid deadlocks; so locking the cache lines sorted by PA would cover that?
-
comment
Comment #47671237
This is very spot on in my experience.
-
comment
Comment #47538807
> 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 …
-
comment
Comment #46714201
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…
-
comment
Comment #46522478
This has already been tried :) iirc, in the 2016 a quadcore intel cpu ran the original crysis at ~15fps
-
comment
Comment #46338958
Get the DGX Spark computers? They’re exactly what you’re trying to build.
-
comment
Comment #46320306
the TB5 link (RDMA) is much slower than direct access to system memory
-
comment
Comment #45841906
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…
-
comment
Comment #45814220
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 …
-
comment
Comment #45813955
> 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 …
-
comment
Comment #45796610
What do you mean by that? Most syscalls are still interrupt based.
-
comment
Comment #45723964
Double buffering a 4K 4bpp framebuffer itself is 64mb
-
comment
Comment #45541759
> 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.
-
comment
Comment #45427856
AUM is not theirs to keep; and market cap is a very deceitful metric especially for banks where liabilities dwarf the market cap.
-
comment
Comment #45368059
Well so is the snapdragon X elite, including the older snapdragons (anyone remember scorpion cores on QSD8x50?)
-
comment
Comment #44740399
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 - …
-
comment
Comment #44246516
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://lear…