my123

14 ms·

151.
by my123 3y ago
> Which got me to wonder: what if Apple were able to introduce a "MacOS subsystem for Windows", which could run most x64 binaries for the latter platform? Was attempted by Apple but abandoned ages ago, around Leopard times. Wha
152.
by my123 3y ago
Yes, it does still apply
153.
by my123 3y ago
> Perhaps some GPUs do, but many (most?) don't See Qualcomm Adreno
154.
by my123 3y ago
The UM part, KM is OSS
155.
by my123 3y ago
A number of GPUs use a standard Arm SMMU instead of an IOMMU already. The problem with those GPUs in general is driver issues, the hardware is fine.
156.
by my123 3y ago
The 30x number is for a really narrow scenario tbh. Running a GPT 1.8T parameters (w/ MOE) on one GB200
157.
by my123 3y ago
All. Every single arm64 desktop machine supports disabling UEFI Secure Boot.
158.
by my123 3y ago
This runs Cortex-X1C and A78C, which are of the generation _just_ before the one that got SVE
159.
by my123 3y ago
Qualcomm disables SVE (masked at the hypervisor level) on all their silicon. If you want SVE on a phone roday, your options are Tensor G3, Exynos 2200/2400 or MediaTek phones with ARMv9 CPUs. Or if you have hypervisor execution level c
160.
by my123 3y ago
It's not just about a given microbenchmark - that platform just sucks tbh. Hopefully something better appears.
161.
by my123 3y ago
TH1520 with the XuanTie C910 core. Avoid. That particular platform is really incredibly slow, Cortex-A5x level. And with RVV 0.7.1 instead of 1.0 too.
162.
by my123 3y ago
https://www.arrow.com/en/manufacturers/qualcomm/microcontrol... - has plenty of newer parts than A72 available
163.
by my123 3y ago
Rosetta supports 32-bit code execution via a very basic (segment-level) facility, designed for use by Wine.
164.
by my123 3y ago
CUDA itself only got bumped from LLVM 5 to LLVM 7 in CUDA 11.2 ( https://developer.nvidia.com/blog/boosting-productivity-and-... ). LLVM 7 opt-in for OpenCL happened some time later (available since r510). What changes n
165.
by my123 3y ago
IEEE
166.
by my123 3y ago
The problem with this is licensing: the macOS SDK license says that it's only to be used on an "Apple-branded computer". That said I wonder what Apple would do if this is used outside of that case...
167.
by my123 3y ago
I hoped so but nope. Exynos with RDNA has closed source user-space, with AMD/Samsung not making public the differences, and no attempt at upstreaming the kernel-side ones too. The best mobile platform on Arm with open-source GPU driver
168.
by my123 3y ago
Hyper-V Server stops at the 2019 release - post 2029 it's dead (no Hyper-V Server 2022)
169.
by my123 3y ago
They financed the prior iteration of Zluda: https://github.com/vosen/ZLUDA?tab=readme-ov-file#faq but then stopped
170.
by my123 3y ago
Visual Studio had to wait for .NET Framework (not .NET Core) to be ported to ARM64 native. Quite a part of Microsoft wanted to avoid doing that outright, so it took quite a while to happen.
171.
by my123 3y ago
And same applies to AMD. A reputable retailer selling W7900s: https://www.bhphotovideo.com/c/product/1765536-REG/amd_100_3...
172.
by my123 3y ago
NVIDIA and AMD both have workstation PCIe cards with more than 24GB or VRAM (W7900 from AMD and RTX 6000 Ada both have 48GB). They just want more money.
173.
by my123 3y ago
NVIDIA changed this in 2021: https://nvidia.custhelp.com/app/answers/detail/a_id/5173/~/g...
174.
by my123 3y ago
Optimus proper power management for laptops on Linux (including fully turning the GPU down) only works on Coffee Lake (and AMD Renoir) and later with Ampere or later GPUs. Prior to that generation, standard power management mechanisms don&#
175.
by my123 3y ago
> has DirectX support, but doesn't have Vulkan The Vulkan implementation is still there. You just have to install the redistributable. (see https://vulkan.lunarg.com/sdk/home - Runtime) There's no separate
176.
by my123 3y ago
> nvidia drivers that have DirectX 12, but not Vulkan not a thing. at worst you get a situation where the ICD loader isn't installed
177.
by my123 3y ago
> AMD cards + ROCm are used in top supercomputers for (non-deep learning) HPC. Why is this the case? Public bidding processes are different and much more reliant solely on price. As such you can expect them to pick different choices than
178.
by my123 3y ago
SPIR-V for OpenCL and for Vulkan are substantially different, with the translation between the two being quite non-trivial. (note that rusticl + zink does deal with it _partially_ to some extent nowadays) + Vulkan memory management doesn&#x
179.
Guidance on Amortization of Research or Exp. Expenditures Under Section 174 [pdf](downloads.regulations.gov)
1 points by my123 3y ago|2 comments
180.
by my123 3y ago
That was the SD810. The SD820 was a lot better

More ›