On Mon, Jun 9, 2025 at 4:03 AM John Naylor wrote:
>
> On Sat, Jun 7, 2025 at 4:34 AM Masahiko Sawada wrote:
> > BTW I found that the constant 'maxblkno' in test_tidstore.sql actually
> > equals to InvalidBlockNumber, but not MaxBlockNumber. I think it
> > doesn't make sense that TidStore uses Inv
On 06/06/25 18:34, Masahiko Sawada wrote:
>
> Thank you for the report. With the 0001 and 0002 patches, I got a
> SEGV. I've fixed this issue in the attached updated version patches.
> I've confirmed that the patches pass CI tests but I'm not sure it
> fixes the shared memory segment leak problem y
On Fri, Jun 6, 2025 at 4:28 PM Peter Geoghegan wrote:
>
> On Fri, Jun 6, 2025 at 6:58 PM Masahiko Sawada wrote:
> > Agreed. Given the above test results, it's unlikely always sorting the
> > array helps speedups.
>
> Did you try specializing the sort? In my experience, it makes a big
> differenc
On Sat, Jun 7, 2025 at 4:34 AM Masahiko Sawada wrote:
> BTW I found that the constant 'maxblkno' in test_tidstore.sql actually
> equals to InvalidBlockNumber, but not MaxBlockNumber. I think it
> doesn't make sense that TidStore uses InvalidBlockNumber as the key.
> The attached 0001 patch fixes i
On Sat, Jun 7, 2025 at 8:46 PM Junwang Zhao wrote:
>
> Hi Masahiko,
>
> On Sat, Jun 7, 2025 at 5:34 AM Masahiko Sawada wrote:
> >
> > On Tue, May 13, 2025 at 2:26 PM Matheus Alcantara
> > wrote:
> > >
> > > Hi,
> > >
> > > On 30/04/25 19:36, Masahiko Sawada wrote:
> > > > Here are the summary of
Hi Masahiko,
On Sat, Jun 7, 2025 at 5:34 AM Masahiko Sawada wrote:
>
> On Tue, May 13, 2025 at 2:26 PM Matheus Alcantara
> wrote:
> >
> > Hi,
> >
> > On 30/04/25 19:36, Masahiko Sawada wrote:
> > > Here are the summary of each attached patch:
> > >
> > > 0001: Introduce TIdStoreIsMemberMulti() w
On Fri, Jun 6, 2025 at 6:58 PM Masahiko Sawada wrote:
> Agreed. Given the above test results, it's unlikely always sorting the
> array helps speedups.
Did you try specializing the sort? In my experience, it makes a big difference.
--
Peter Geoghegan
On Sun, Jun 1, 2025 at 11:01 PM John Naylor wrote:
>
> On Thu, May 1, 2025 at 5:36 AM Masahiko Sawada wrote:
>
> > HEAD PATCHEDDIFF
> > case-1: 3,021 ms 2.818 ms93.29%
> > case-2: 5, 697 ms 5.545 ms97.34%
> > case-3: 2,833 ms 2.790 m
On Tue, May 13, 2025 at 2:26 PM Matheus Alcantara
wrote:
>
> Hi,
>
> On 30/04/25 19:36, Masahiko Sawada wrote:
> > Here are the summary of each attached patch:
> >
> > 0001: Introduce TIdStoreIsMemberMulti() where we can do IsMember check
> > for multiple TIDs in one function call. If the given TI
On Thu, May 1, 2025 at 5:36 AM Masahiko Sawada wrote:
> HEAD PATCHEDDIFF
> case-1: 3,021 ms 2.818 ms93.29%
> case-2: 5, 697 ms 5.545 ms97.34%
> case-3: 2,833 ms 2.790 ms98.48%
> case-4: 2,564 ms 2.279 ms88.86%
> case
Hi,
On 30/04/25 19:36, Masahiko Sawada wrote:
> Here are the summary of each attached patch:
>
> 0001: Introduce TIdStoreIsMemberMulti() where we can do IsMember check
> for multiple TIDs in one function call. If the given TIDs are sorted
> (at least in block number), we can save radix tree lookup
11 matches
Mail list logo
