diff options
Diffstat (limited to 'Documentation/video4linux/bttv/README.PCI')
-rw-r--r-- | Documentation/video4linux/bttv/README.PCI | 36 |
1 files changed, 0 insertions, 36 deletions
diff --git a/Documentation/video4linux/bttv/README.PCI b/Documentation/video4linux/bttv/README.PCI deleted file mode 100644 index 1ae8276f1..000000000 --- a/Documentation/video4linux/bttv/README.PCI +++ /dev/null @@ -1,36 +0,0 @@ -Because some people were asking about the bandwidth the Bt848 might use up -on the PCI bus I did a little benchmark. - -"bonnie -s 200" with a Fireball TM 3.8 Gb using Busmaster DMA on an ASUS P6NP5 - -without capturing: - - -------Sequential Output-------- ---Sequential Input-- --Random-- - -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks--- -Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU - 200 5353 76.6 5898 16.9 2363 12.1 5889 51.3 6416 10.2 37.8 0.9 - - -while capturing full screen PAL (786x576) with 24bpp: - - -------Sequential Output-------- ---Sequential Input-- --Random-- - -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks--- -Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU - 200 5619 69.3 5939 16.9 2334 12.0 5859 50.9 6441 10.5 37.9 0.9 - -The differences are small and probably within the normal error margin of -bonnie. -So, one bt848 card does not have much(any?) impact on the normal operation -of a Linux system. -If you have several cards running this will look very differently! -The same is probably true if your Linux box is used as a file server -with 15 (or 30) SCSI drives. - -I tested having 2 Bt848 cards grabbing in 32 bit mode (That's almost 100MB/s!) -while running bonnie. -The xtvscreen windows showed severe pixel errors. -After a while the ide driver failed to use DMA and switched DMA off. -It continued running but the results where bad. - - - |