my123

9 ms·

61.
by my123 10mo ago
The ABI was stabilised for backwards compatibility since Windows Server 2022, but is not stable for earlier releases.
62.
by my123 10mo ago
It works with SR-IOV instead of mdev afaik Still needs some host SW to drive it but actually does static partitioning IIRC it's usable through using the MIG-marked vGPU types
63.
by my123 10mo ago
Is https://github.com/Samsung/ENNDelegate enough or is it TFLite/LiteRT only?
64.
by my123 10mo ago
There's a MIG vGPU mode usable for this
65.
by my123 10mo ago
Notably, ConnectX-8 already has 200G PAM4 for 800G as a single port - and uses it for Infiniband. Looks like it is so because it pre-dates 800G Ethernet ratification...
66.
by my123 10mo ago
Not for prompt processing. Current Macs are really not great at long contexts
67.
Tile IR Specification(docs.nvidia.com)
19 points by my123 11mo ago|3 comments
68.
by my123 11mo ago
The specialty DRAM vendors (Nanya notably) will keep making DDR4 and earlier
69.
by my123 11mo ago
Regular Hyper-V VMs can use it too but it's not well documented. The keyword to search for: "GPU-P" https://learn.microsoft.com/en-us/windows-server/virtualizat...
70.
by my123 11mo ago
> It will be interesting to see what Google got from Qualcomm for the new Chromebooks (EL3 isn't even the highest level anymore, that's TME now). The new AL BSP target for Hamoa, which is what's going to ship on the new An
71.
by my123 11mo ago
AFAIK Google runs their own EL3 on the Snapdragon Chromebooks. (And KVM at EL2) Lots of this is customer dependent but what you say is true for the typical android phone config that most use
72.
by my123 11mo ago
_Current_ phone platforms still use Gunyah.
73.
by my123 11mo ago
See qebspil for filling the gap for the EL2 side
74.
by my123 11mo ago
Didn't ship externally/OEMs didn't take it, but the CRDs had the option. The prototype impl at the time broke some Windows functionality
75.
by my123 11mo ago
Fedora boots ootb
76.
by my123 11mo ago
not all true. And Qualcomm taking over EL2 is optional now
77.
by my123 11mo ago
> And don't even get me started on Gunyah and Gunyah is disappearing from new chips, slowly but surely. X2 doesn't have it anymore, the IoT range has it as optional now. And it's going to be deployed less from there
78.
by my123 11mo ago
No you can https://developer.apple.com/support/xcode/ the iOS 26 SDK can target iOS 15 or later
79.
by my123 1y ago
Maybe for gaming but at 24 inches for desktop monitor I see a sizeable difference for text rendering. Looks like good anti-aliasing for text to look better on lower DPI display is slowly getting the bitrot treatment...
80.
by my123 1y ago
Qualcomm IoT likes U-Boot with UEFI compat
81.
Qualcomm QRB2210 Datasheet [pdf](docs.qualcomm.com)
1 points by my123 1y ago|0 comments
82.
by my123 1y ago
The GPU on the RPi is a _lot_ slower
83.
by my123 1y ago
128GB is what they can ship using RAM chips available today, but the SoC supports more.
84.
Linux 6.18 Will Further Complicate Non-GPL Out-of-Tree File-Systems(phoronix.com)
13 points by my123 1y ago|0 comments
85.
by my123 1y ago
Blackwell DC (B200/B300)
86.
by my123 1y ago
Windows XP version 2003 is a thing, for Itanium (which got version 2002 but was short-lived) and AMD64
87.
by my123 1y ago
On Snapdragon X and later, native Vulkan and OpenCL UMDs are provided out of the box GLon12 is still used for OpenGL however
88.
by my123 1y ago
Yes, you can use first-party models on the Pixel NPUs or you're stuck with NNAPI which is self-admittedly deprecated by Google and doesn't work all that well. On third party model workloads, this is what you will get: https:&#x
89.
by my123 1y ago
Tensor is essentially shipping subpar hardware with not even taking care of software properly. https://ai.google.dev/edge/litert/android/npu/overview has been identical for a year+ now. In practice Qualc
90.
by my123 1y ago
arm64 macOS doesn't even allow statically linked binaries at all. on the windows side, syscall ABI became stable since Server 2022 to run mismatched container releases

More ›