Message-ID: <c4baae3f14a809cdea5a0d04d07d447124fa0caf.camel@j-davis.com>
Subject: Re: Trouble with hashagg spill I/O pattern and costing
From: Jeff Davis <pgsql@j-davis.com>
To: Tomas Vondra <tomas.vondra@2ndquadrant.com>
Cc: pgsql-hackers@postgresql.org
Date: Thu, 21 May 2020 12:40:23 -0700
In-Reply-To: <20200521191318.gw3xjivrxracvzmg@development>
References: <20200519151202.u2p2gpiawoaznsv2@development>
	 <88dd39d782178539d7ee46f98b98579940e2a12e.camel@j-davis.com>
	 <20200519175320.tl4qm6le5dbg2pcf@development>
	 <acad6f2ba8d560b9dfa9a564c7b7d3162abf8347.camel@j-davis.com>
	 <20200521191318.gw3xjivrxracvzmg@development>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Thu, 2020-05-21 at 21:13 +0200, Tomas Vondra wrote:
> 1) Instead of assigning the pages one by one, we can easily extend
> the
> API to allow getting a range of blocks, so that we don't need to call
> ltsGetFreeBlock in a loop. Instead we could call ltsGetFreeBlockRange
> with the requested number of blocks.

ltsGetFreeBlock() just draws one element from a minheap. Is there some
more efficient way to get many elements from a minheap at once?

>  And we could keep just a min/max
> of free blocks, not an array with fixed number of elements.

I don't quite know what you mean. Can you elaborate?

Regards,
	Jeff Davis