zznzz

7 ms·

1.
by zznzz 3y ago
jax and whisper-jax support running on GPUs in addition to TPUs, this particular demo just runs on a TPU.
2.
by zznzz 4y ago
what kind of lasers / lidar technology is this that is not reflected by trees?
3.
by zznzz 4y ago
altra is also > 2 years old, so the age of the apple chip doesn’t matter Neoverse N1 is hardly a dog. It’s a decent enough core in a server processor that has 80 of them. Altra is not intended to compete core for core with any laptop&#x2
4.
by zznzz 4y ago
A comment on some of the presentation of the history (‘staunton has already noted the strangeness of bellyaching in 2021 about renormalization without mentioning the renormalization group) In section 4.3 the author incorrectly describes Fer
5.
by zznzz 4y ago
What you describe (multiple cores sharing the same lock) sounds like true sharing, not false sharing. False sharing means cache line contention is when cores are using separate structures (i.e. not actually shared) which happen to be on the
6.
by zznzz 4y ago
Accelerating Standard C++ with GPUs Using stdpar https://developer.nvidia.com/blog/accelerating-standard-c-wi...
7.
by zznzz 4y ago
Yes, this only works for user space stacks, but that is sufficient since with ORC kernel stacks are solved (IMO) and it avoids all the issues with trying to mlock debuginfo of all processes that you mentioned. The NMI handler would still un
8.
by zznzz 4y ago
I don’t understand your concern - what about this would involve one process sniffing another process’s memory? The kernel would still be doing the unwinding, just not in the NMI handler.
9.
by zznzz 4y ago
because this is about PMU based sampling, which involves triggering interrupts at some interval and doing the sampling while handling the interrupt
10.
by zznzz 4y ago
It is not required to unwind the user space stack in the NMI handler. It can be done later before returning to user space in a context that can handle faults.
11.
by zznzz 4y ago
See also https://lists.fedoraproject.org/archives/list/devel@lists.fe... - the botan phoronix results with frame pointers were probably measuring debug builds.
12.
by zznzz 4y ago
I cannot comment on whether “everyone” is oblivious but yes, this is still the case - frame pointer based unwinding sometimes skips the caller when the IP is sampled before the callee sets up a frame. This is also common for samples in leaf
13.
by zznzz 4y ago
IIUC Windows implements in-kernel unwinding using FPO debug data embedded in every executable.
14.
by zznzz 4y ago
Register renaming doesn’t significantly address the impact of reducing the number of architectural registers available to the compiler. With fewer register names available, the compiler will spill locals to stack more often, and register re
15.
by zznzz 4y ago
Can you give more detail? AFAIK frame pointers work fine for unwinding on aarch64. And on aarch64 the gcc default is not to omit frame pointers and IIRC when the default was switched at some point it was treated as a bug and reverted (not
16.
by zznzz 4y ago
> Is DWARF unwinding so slow that that is actually faster? No. The only reason it works like this is because the upstream Linux kernel has thus far rejected in-kernel dwarf unwinders, but copying the stack is simpler and available /