aris
14th February 2002, 14:50
Hi peeps!
Here's the setup:
1. Athlon TB 850 WinXP client, 10Gig UDMA66 disk, and Sony DDU-something (x16 DVD drive). SmartRipper 2.40 as comes in latest GKnot. UDMA enabled on both, both using 80-pin flats, so don't send me reading about the traditional "smartripper reads slowly at x1.8 and what does DMA mean?" threads. It's not that.
2. 100MBit fully-working switch & network, ftp cables, all by the book.
3. FreeBSD4.4 Server, with big UDMA66 harddisks.
1. RIP to local harddisk on WinXP works like charm, yielding more-or-less sustained rate of 10-13MegaBYTES/sec.
2. Copying a large (>1gig) file to the bsd server using either a network drive mapped to a samba share or by just using FTP yields about 8-9MegaBYTES/sec (as is to be expected of a fully-functional 100M network). This seems to prove network functionality.
And here's the bummer:
Ripping directly to mapped network drive (samba share) yields 2.5MegaBYTES/sec. Neither machine is utilizing 100%CPU or 100%RAM (both have a lot of both).
So a rip takes 30-40 minutes instead of 10-15. Ouch.
Any ideas?
Possibly the way SR is written that forces a delay on writing VOBs - that rurns out minimal on HD but takes much longer when done through network..? An integrity-check on what it wrote maybe? Tron?
Here's the setup:
1. Athlon TB 850 WinXP client, 10Gig UDMA66 disk, and Sony DDU-something (x16 DVD drive). SmartRipper 2.40 as comes in latest GKnot. UDMA enabled on both, both using 80-pin flats, so don't send me reading about the traditional "smartripper reads slowly at x1.8 and what does DMA mean?" threads. It's not that.
2. 100MBit fully-working switch & network, ftp cables, all by the book.
3. FreeBSD4.4 Server, with big UDMA66 harddisks.
1. RIP to local harddisk on WinXP works like charm, yielding more-or-less sustained rate of 10-13MegaBYTES/sec.
2. Copying a large (>1gig) file to the bsd server using either a network drive mapped to a samba share or by just using FTP yields about 8-9MegaBYTES/sec (as is to be expected of a fully-functional 100M network). This seems to prove network functionality.
And here's the bummer:
Ripping directly to mapped network drive (samba share) yields 2.5MegaBYTES/sec. Neither machine is utilizing 100%CPU or 100%RAM (both have a lot of both).
So a rip takes 30-40 minutes instead of 10-15. Ouch.
Any ideas?
Possibly the way SR is written that forces a delay on writing VOBs - that rurns out minimal on HD but takes much longer when done through network..? An integrity-check on what it wrote maybe? Tron?