| Pages: [1] :: one page |
| Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Uhr Zylex
Ginnungagaps Rymdfarargille
Blade.
   |
Posted - 2009.01.14 15:09:00 -
[1]
When you log in to the character selection screen, EVE reads in cached stuff from c_program_files_ccp_eve_tranquility\cache\MachoNet. Caching is great, but for the love of god, stop putting everything into tiny files! It causes massive fragmentation and NTFS isn't known for dealing with these kinds of files in the best possible way.
CCP: Please consider storing the cache in a better way so I don't have to clear it every few days. |
Sir Substance
Minmatar
MagiTech Alliance Inc.
MagiTech Corp
|
Posted - 2009.01.14 15:16:00 -
[2]
the devs will not read this post just because you ask them to in the title.
my understanding of NTFS was that it was explicitly designed to handle high fragmentation environments, and to minimise disk fragmentation in the first place
this thread is rated 2/10. good idea, but poorly executed, with extra whine in the title. |
Uhr Zylex
Ginnungagaps Rymdfarargille
Blade.
|
Posted - 2009.01.14 15:19:00 -
[3]
 Originally by: Sir Substance
my understanding of NTFS was that it was explicitly designed to handle high fragmentation environments, and to minimise disk fragmentation in the first place
This reply is rated 0 / 10. |
Joe Starbreaker
Starbreaker Frigateers
|
Posted - 2009.01.15 00:15:00 -
[4]
I don't know about the file system, but I know that login is a serious serious problem. And the devs should take a look. |
Astria Tiphareth
Caldari
24th Imperial Crusade
|
Posted - 2009.01.15 16:00:00 -
[5]
Originally by: Uhr Zylex
Caching is great, but for the love of god, stop putting everything into tiny files! It causes massive fragmentation and NTFS isn't known for dealing with these kinds of files in the best possible way.
It is? A small write is less likely to suffer fragmentation than a large write. A quick C test program on a heavily fragmented NTFS partition produced consistent results that any file over 4096 bytes will be fragmented by NTFS without additional code to create the file space in a single operation, even if a larger free cluster area can then be used to defragment the file. Do you perhaps mean locality of reference, i.e. having the various cache files close together on the physical disk?
Originally by: Uhr Zylex
Please consider storing the cache in a better way so I don't have to clear it every few days.
I cannot even begin to think why you consider this necessary - as a person who has cleared my cache, but for most part does not, the performance difference to me seems virtually impossible to measure. I'd enjoy seeing some statistics as to how much this cleansing improves your FPS or lag.
There is no good alternative to defragmenting your hard drive.
___
My views may not represent those of my corporation, which is why I never get invited to those diplomatic parties...
Environmental Effects
|
Gunnanmon
Gallente
UNITED STAR SYNDICATE
|
Posted - 2009.01.15 18:38:00 -
[6]
Originally by: Joe Starbreaker
I don't know about the file system, but I know that login is a serious serious problem. And the devs should take a look.
Out of interest, what is the problem? |
Roozu Valentine
Gallente
University of Caille
|
Posted - 2009.01.15 18:40:00 -
[7]
Originally by: Sir Substance
Edited by: Sir Substance on 14/01/2009 15:21:12
the devs will not read this post just because you ask them to in the title.
my understanding of NTFS was that it was explicitly designed to handle high fragmentation environments, and to minimise disk fragmentation in the first place.
this thread is rated 2/10. good idea, but poorly executed, with extra whine in the title.
You are rated troll, with almost no helpful information you are clearly ****ed at life and need someone to take it out on. |
Astria Tiphareth
Caldari
24th Imperial Crusade
|
Posted - 2009.01.15 19:06:00 -
[8]
Originally by: Gunnanmon
Originally by: Joe Starbreaker
I don't know about the file system, but I know that login is a serious serious problem. And the devs should take a look.
Out of interest, what is the problem?
I believe he was being sarcastic. From Joe's previous posting style, to me the above implies that actually there are vastly more important things to worry about than how long it takes to log in. I tend to agree. |
Clansworth
Blackwater USA Inc.
|
Posted - 2009.01.15 19:11:00 -
[9]
As for file fragmentation, if this is a concern to you, you could turn on File Compression for the eve cache folders, which on modern processors is a negligible performance hit, and in fact clusters multiple small files together as basically a single larger file on the disc. This will end up reducing seek times, and overall speed up the process, as windows caching mechanism will have a much better idea of where on the disc to read next (as the whole directory ends up being a single continuous file). |
Astria Tiphareth
Caldari
24th Imperial Crusade
|
Posted - 2009.01.15 19:22:00 -
[10]
Edited by: Astria Tiphareth on 15/01/2009 19:24:30
Originally by: Sir Substance
my understanding of NTFS was that it was explicitly designed to handle high fragmentation environments, and to minimise disk fragmentation in the first place.
It's better than FAT32 for certain. I must confess I've never entirely understood why NTFS didn't adopt the same methodology as EXT3, which optimally has zero fragmentation because it never normally allows any fragments. In practice I've found both architectures to end up needing the same basic rule - don't let your free space drop below about 30%. In the case of EXT3 it's to stop the file system being forced to fragment (and because everyone claims EXT3 doesn't normally fragment, defragmenters don't exist), and in the case of NTFS it's to have room to defragment.
Edit: whilst it might be the equivalent of King Canute facing the waves, if we could avoid a Linux vs Windows flamewar, that would be great - I was not attempting to claim one is better than the other, merely highlighting the difference between the two.
___
My views may not represent those of my corporation, which is why I never get invited to those diplomatic parties...
Environmental Effects
|
| |
|
| Pages: [1] :: one page |
| First page | Previous page | Next page | Last page |