In Reply to: The art of building Computer Transports posted by cics on April 5, 2007 at 08:00:10:
I want to clarify this subject further on why they should be avoided. This also applies to Windows on USB flash drives. My insights are as follows:1. Best design is when soundcard has lions share of PCI bus use.
2. All soundcards (USB, Ethernet or PCI) operate of PCI bus.
3. USB drives and Network Drives (which also connect to PCI bus) goes against this design principle as data traffic now competes for PCI bus use.PCI read or write transactions take place between master and slave devices (soundcard, CPU…). All devices connected to PCI bus can act as master or slave. During a read or write transaction, master takes control of bus and STOPS all other devices from transacting. This is where a conflict arises – soundcard wants more data for output BUT PCI bus is busy with a data transaction (from USB drive or Ethernet) thus forcing soundcard to wait. This all happens at frantic speeds but for ultra low latency playback, this conflict will have impacts.
Hence, I prefer internal laptop SATA drives where mobo maintains dedicated bandwith for data flow. A further improvement is to use SATA RAID 0 which does away with IDE – this is my current setup using 2 Seagate 5400.3 160GB drives.
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- Don’t play music from USB drives and Network - cics 10:52:57 04/06/07 (17)
- Re: Don’t play music from USB drives and Network - swarmer 22:04:19 04/10/07 (0)
- Re: Don’t play music from USB drives and Network - artcheng@netvigator.com 22:55:40 04/06/07 (1)
- Yes. - cics 02:04:46 04/07/07 (0)
- Re: Don’t play music from USB drives and Network - artcheng@netvigator.com 22:45:28 04/06/07 (2)
- Could be - not sure. - cics 02:21:59 04/07/07 (1)
- Re: Could be - not sure. - artcheng@netvigator.com 02:32:01 04/07/07 (0)
- Not necessarily - Christine Tham 15:31:34 04/06/07 (10)
- Re: Not necessarily - cics 20:28:27 04/06/07 (9)
- Re: Not necessarily - Christine Tham 21:11:53 04/06/07 (8)
- Re: Not necessarily - cics 03:19:40 04/07/07 (7)
- I'm not sure I understand you - Christine Tham 19:37:45 04/07/07 (4)
- Re: I'm not sure I understand you - cics 07:55:02 04/08/07 (3)
- Re: I'm not sure I understand you - KeithC 13:09:03 04/08/07 (2)
- Design goal is to elliminate all IO overheads - cics 23:04:53 04/08/07 (0)
- Hardware RAID is implemented - cics 22:49:12 04/08/07 (0)
- Re: underclocking - aljordan 16:09:02 04/07/07 (1)
- Re: underclocking - aljordan 16:12:23 04/07/07 (0)