virtuoso-opensource VS highway

https://en.wikipedia.org/wiki/Datalog#Evaluation

...

VMware/ddlog: Differential datalog

> Bottom-up: DDlog starts from a set of input facts and computes all possible derived facts by following user-defined rules, in a bottom-up fashion. In contrast, top-down engines are optimized to answer individual user queries without computing all possible facts ahead of time. For example, given a Datalog program that computes pairs of connected vertices in a graph, a bottom-up engine maintains the set of all such pairs. A top-down engine, on the other hand, is triggered by a user query to determine whether a pair of vertices is connected and handles the query by searching for a derivation chain back to ground facts. The bottom-up approach is preferable in applications where all derived facts must be computed ahead of time and in applications where the cost of initial computation is amortized across a large number of queries.

From https://community.openlinksw.com/t/virtuoso-openlink-reasoni... https://github.com/openlink/virtuoso-opensource/issues/660 :

> The Virtuoso built-in (rule sets) and custom inferencing and reasoning is backward chaining, where the inferred results are materialised at query runtime. This results in fewer physical triples having to exist in the database, saving space and ultimately cost of ownership, i.e., less physical resources are required, compared to forward chaining where the inferred data is pre-generated as physical triples, requiring more physical resources for hosting the data.

FWIU it's called ShaclSail, and there's a NotifyingSail: org.eclipse.rdf4j.sail.shacl.ShaclSail: https://rdf4j.org/javadoc/3.2.0/org/eclipse/rdf4j/sail/shacl...

The bf16 dot instruction replaces 6 instructions: https://github.com/google/highway/blob/master/hwy/ops/x86_12...

[0] for those interested in Highway.

It's also mentioned in [1], which starts off

> Today we're sharing open source code that can sort arrays of numbers about ten times as fast as the C++ std::sort, and outperforms state of the art architecture-specific algorithms, while being portable across all modern CPU architectures. Below we discuss how we achieved this.

[0] https://github.com/google/highway

[1] https://opensource.googleblog.com/2022/06/Vectorized%20and%2..., which has an associated paper at https://arxiv.org/pdf/2205.05982.pdf.

Thanks so much!

Everyone working on this self-selected into contributing, so I think of it less as my team than ... a team?

Specifically want to call out: Jan Wassenberg (author of https://github.com/google/highway) and I started gemma.cpp as a small project just a few months ago + Phil Culliton, Dan Zheng, and Paul Chang + of course the GDM Gemma team.

C++ users can enjoy Highway [1].

[1] https://github.com/google/highway/

At that point it is better to have some kind of DSL that should not be in the main language, because it would target a much lower level than a typical program. The best effort I've seen in this scene was Google's Highway [1] (not to be confused with HighwayHash) and I even once attempted to recreate it in Rust, but it is still distanced from my ideal.

[1] https://github.com/google/highway

Interesting, thanks for sharing :)

At the time we open-sourced Highway, the standardization process had already started and there were some discussions.

I'm curious why stdlib is the only path you see to default? Compare the activity level of https://github.com/VcDevel/std-simd vs https://github.com/google/highway. As to open-source usage, after years of std::experimental, I see <200 search hits [1], vs >400 for Highway [2], even after excluding several library users.

But that aside, I'm not convinced standardization is the best path for a SIMD library. We and external users extend Highway on a weekly basis as new use cases arise. What if we deferred those changes to 3-monthly meetings, or had to wait for one meeting per WD, CD, (FCD), DIS, (FDIS) stage before it's standardized? Standardization seems more useful for rarely-changing things.

1: https://sourcegraph.com/search?q=context:global+std::experim...

2: https://sourcegraph.com/search?q=context:global+HWY_NAMESPAC...

Thanks for the link. We were previously using GFNI for bit reversal and 8-bit shifts, and I just extended that to our 8-bit BroadcastSignBit (https://github.com/google/highway/pull/1784).

You could study Google's Highway library [1].

[1] https://github.com/google/highway

Runtime feature detection need not be rare nor hard, it's a few dozen lines of boilerplate. You can even write your code just once: see https://github.com/google/highway#examples.