Oh yes, good idea. Forgot about the split!
+Lucene Dev
On Tue, 1 Aug, 2023, 6:17 pm Uwe Schindler, wrote:
> Maybe ask on Lucene list, too, if there are some bug people like to have
> fixed in Lucene.
>
> Uwe
>
> Am 01.08.2023 um 11:10 schrieb Ishan Chattopadhyaya:
> > Hi all,
> > There have
>
> Wow, this looks very relevant to Lucene! Could this index be used for
> faster implementation of our skip lists? Even though they are static
> (computed once at segment-write time) vs dynamic/online that these learned
> indices are also able to handle, it looks like learned indices are still
On Sun, Jul 30, 2023 at 11:17 AM Bruno Roustant
wrote:
Interesting coincidence, I'm currently working on a learned index on sorted
> keys that can advantageously replace binary search.
> It is very compact (additional space of 2% of the sorted key array, e.g.
> 40KB for 200MB of keys), and it is
On Fri, Jul 28, 2023 at 7:04 AM Dawid Weiss wrote:
> Actually this is exactly the same for Java:
>>
> Yup, I know (we all know by now, I guess). People (including me) evidently
> crave this low, iron-level control, while at the same time mostly try to
> dodge writing any software in languages
Yeah +1 on waiting/asking/expecting CMOV to be properly utilized by
Hotspot, instead of trying to target the instruction ourselves. This is
all more of a curiosity / exploration. I am curious whether "branchless
binary search" is something Arrays.binarySearch already does / compiles
to. Even in
On Thu, Jul 27, 2023 at 9:19 AM Uwe Schindler wrote:
See Shipilevs blog:
> https://shipilev.net/jvm/anatomy-quarks/30-conditional-moves/
>
Really interesting! This is an awesome, quick explanation of the tradeoff
CMOV is making (pre-computing both paths) vs branching (have to predict,
with
On Thu, Jul 27, 2023 at 8:44 AM Uwe Schindler wrote:
> Hi Mike,
>
> actually Hotspot is using CMOV. Some nodes after bytecode analysis are
> converted to CMoveNode and it has implementations to generate machine code
> (as far as i see) for x86, s390 and arm.
>
> The generic code is here:
>
