Re: [Firebird-devel] Lock manager's hash calculations

On 12/29/2015 07:05 PM, Dimitry Sibiryakov wrote: > 29.12.2015 11:43, Alex Peshkoff wrote: >> That I do not expect to get something better than this bell curve: > > I was unable to create a good load on database with my weak > notebook, but oltp-emul with 5 windows gave me these numbers for

Re: [Firebird-devel] Lock manager's hash calculations

29.12.2015 11:43, Alex Peshkoff wrote: That I do not expect to get something better than this bell curve: I was unable to create a good load on database with my weak notebook, but oltp-emul with 5 windows gave me these numbers for lock manager using CRC32: Hash slots: 8191, Hash lengths

Re: [Firebird-devel] Lock manager's hash calculations

28.12.2015 16:03, Dimitry Sibiryakov wrote: Yes. But in this case the hash is only used internally and not stored anywhere. Because of this, different platforms can use different algorithms for it. >> Of course, but then it should be encapsulated and moved to /common to >> avoid

Re: [Firebird-devel] Lock manager's hash calculations

I stumbled upon quite evesome Hash-algorithm xxHash. There are lightning fast implementations. https://github.com/Cyan4973/xxHash Noted that usually the Hash algorithm is not the bottleneck, but still pretty interesting... -Tee- On Tue, Dec 29, 2015 at 6:05 PM, Dimitry Sibiryakov

Re: [Firebird-devel] Lock manager's hash calculations

29.12.2015 18:08, Dmitry Yemanov wrote: > Really? Yep. Look into Hash.h. It is much different (and much slower). -- WBR, SD. -- Firebird-Devel mailing list, web interface at

[Firebird-devel] march switch in posix build for x86 architecture

Hello, All. Isn't it a good time to drop support of 20 years old processors and raise -march value from i386 to i686? -- WBR, SD. -- Firebird-Devel mailing list, web interface at

Re: [Firebird-devel] march switch in posix build for x86 architecture

On Tue, Dec 29, 2015 at 10:04:52PM -0200, Carlos H. Cantu wrote: > > Let's try to see the case from another angle: > > On Windows, Firebird 3.0 uses VC 2010 runtime. From MS site, the > minimum requirements of this runtime are: > > Windows XP SP3 (SP3 was released in 2008) > Computer with 900

Re: [Firebird-devel] march switch in posix build for x86 architecture

On Tue, Dec 29, 2015 at 07:43:37PM -0500, James Starkey wrote: > Could you explain to us what optimization a are possible on 686 that > are not on the 386? >From the top of my head, 386 doesn't even have any atomic cmpxchg or xadd instruction which rather complicates any lock or semaphore

Re: [Firebird-devel] march switch in posix build for x86 architecture

Could you explain to us what optimization a are possible on 686 that are not on the 386? On Tuesday, December 29, 2015, Dimitry Sibiryakov wrote: > 29.12.2015 21:37, James Starkey wrote: > > Why? For all practical purposes they're the same architecture. There > is no

Re: [Firebird-devel] march switch in posix build for x86 architecture

DS> 29.12.2015 21:37, James Starkey wrote: >> Why? For all practical purposes they're the same architecture. There is >> no upside to >> the project and only down side for users. DS>Using of i386 command set limits optimization possibilities for compilers. DS>And, frankly, can you

Re: [Firebird-devel] march switch in posix build for x86 architecture

29.12.2015 21:37, James Starkey wrote: > Why? For all practical purposes they're the same architecture. There is no > upside to > the project and only down side for users. Using of i386 command set limits optimization possibilities for compilers. And, frankly, can you imagine that

Re: [Firebird-devel] march switch in posix build for x86 architecture

Why? For all practical purposes they're the same architecture. There is no upside to the project and only down side for users. Go ahead and drop Apollo/Domain, with my blessings. On Tuesday, December 29, 2015, Dimitry Sibiryakov wrote: >Hello, All. > >Isn't it a

Re: [Firebird-devel] march switch in posix build for x86 architecture

Carlos H. Cantu wrote: > Anyway, I think the real question is: how much performance > increase this would bring, in real world environment? Do you > have this number? If you tell me +25%, I would vote to drop > i386 support :) Yep. This is it exactly. Like all optimisations: Work out what it

Re: [Firebird-devel] Lock manager's hash calculations

On 12/29/2015 01:23 PM, Dimitry Sibiryakov wrote: 24.12.2015 18:29, Alex Peshkoff wrote: If we have one chain with 12 objects, three empty chains and chains with 1, 2 or 3 objects - it's not bad. I.e. existing stat-s is not enough for analysis. I've got following data from Igor

Re: [Firebird-devel] Lock manager's hash calculations

24.12.2015 18:29, Alex Peshkoff wrote: > If we have one chain with 12 objects, three empty chains and chains > with 1, 2 or 3 objects - it's not bad. > I.e. existing stat-s is not enough for analysis. I've got following data from Igor Valchenko: > Hash slots: 65521, Hash lengths