[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [cp-patches] Speed up java.nio.ByteBuffer
From: |
Tom Tromey |
Subject: |
Re: [cp-patches] Speed up java.nio.ByteBuffer |
Date: |
13 Aug 2004 17:31:20 -0600 |
User-agent: |
Gnus/5.09 (Gnus v5.9.0) Emacs/21.3.50 |
>>>>> "Florian" == Florian Weimer <address@hidden> writes:
Florian> On GCJ, this patch results in a 18% improvement in speed for
Florian> loading databases with Berkeley DB Java Edition.
Excellent.
Florian> Further improvements will require a new class for a read-only
Florian> byte buffer, so that the read-only flag can go.
It would be nice to hear what performance impact this has.
Florian> 2004-08-12 Florian Weimer <address@hidden>
Florian> * java/nio/ByteBufferImpl.java (put): New methods to speed up
Florian> ByteBuffer put and array put operations.
Do we want/need similar changes to all the other *BufferImpl classes?
I hate to mention this, but would it make sense to generate these
classes by running sed on a template file, or something similar? It
seems strange to keep patching 7 copies of the same code.
I know, this approach does make life harder for some folks,
e.g. Eclipse users (if any). And perhaps Jeroen...?
Tom