Re: [Gc] Alignment/padding of small objects on 64-bit platform (x86-64)

It would be a definite bug if it didn't work with 8 bytes. Some support
data structures might use a bit more space.

On Sunday, October 12, 2014, Stefan Kral <[email protected]> wrote:

> That is okay with me... At the moment the GC allocates memory in chunks of
> 16 bytes, so I lose 8 bytes everytime I allocate a 24 byte block.
>
> Would lowering the granule size to 8 bytes likely cause any problems? (If
> most of my allocations are small...)
>
> Regards, Stefan.
> On Oct 11, 2014 11:31 PM, "Bruce Hoult" <[email protected]
> <javascript:_e(%7B%7D,'cvml','[email protected]');>> wrote:
>
>> The GC needs to store a pointer for a free list, so needs at least 8
>> bytes per object. And aligned on most CPUs.
>>
>> On Sunday, October 12, 2014, Stefan Kral <[email protected]
>> <javascript:_e(%7B%7D,'cvml','[email protected]');>> wrote:
>>
>>> Hello.
>>>
>>> My app operates on a lot of 3-word objects and I wonder if it would be
>>> reasonable to adapt the granule size accordingly.
>>>
>>> The application requires 64-bit pointers and has no double-word
>>> alignment requirements, so I would be keen on reducing the memory
>>> requirement.
>>>
>>> I'm grateful for any advise and/or tips!
>>>
>>> Best Regards,
>>> Stefan.
>>>
>>> --
>>> This message has been scanned for viruses and
>>> dangerous content by *MailScanner* <http://www.mailscanner.info/>, and
>>> is
>>> believed to be clean.
>>
>>
> --
> This message has been scanned for viruses and
> dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
> believed to be clean.

_______________________________________________
bdwgc mailing list
[email protected]
https://lists.opendylan.org/mailman/listinfo/bdwgc