diff options
| author | Frank Gevaerts <frank@gevaerts.be> | 2009-05-20 16:57:17 +0000 |
|---|---|---|
| committer | Frank Gevaerts <frank@gevaerts.be> | 2009-05-20 16:57:17 +0000 |
| commit | fba4be213f399f19933aadcce25498ea1fb0c176 (patch) | |
| tree | c74268b3625e31948cd5cac75bdbe831af2ddeb9 /firmware/usbstack | |
| parent | 106d2ebd072648482c739d35f6a9f9a0cdfa18e5 (diff) | |
| download | rockbox-fba4be213f399f19933aadcce25498ea1fb0c176.tar.gz rockbox-fba4be213f399f19933aadcce25498ea1fb0c176.zip | |
Don't force double-buffering for sd devices. They apparently are not faster with it, ans possibly actually slower
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@20999 a1c6a512-1295-4272-9138-f99709370657
Diffstat (limited to 'firmware/usbstack')
| -rw-r--r-- | firmware/usbstack/usb_storage.c | 8 |
1 files changed, 7 insertions, 1 deletions
diff --git a/firmware/usbstack/usb_storage.c b/firmware/usbstack/usb_storage.c index 0d06221750..b498311014 100644 --- a/firmware/usbstack/usb_storage.c +++ b/firmware/usbstack/usb_storage.c @@ -53,10 +53,16 @@ * wouldn't buy us anything. * Due to being the double-buffering system used, using a smaller write buffer * ends up being more efficient. Measurements have shown that 24k to 28k is - * optimal + * optimal, except for sd devices that apparently don't gain anything from + * double-buffering */ #define READ_BUFFER_SIZE (1024*64) + +#if (CONFIG_STORAGE & STORAGE_SD) +#define WRITE_BUFFER_SIZE (1024*64) +#else #define WRITE_BUFFER_SIZE (1024*24) +#endif #define ALLOCATE_BUFFER_SIZE (2*MAX(READ_BUFFER_SIZE,WRITE_BUFFER_SIZE)) |