|
Chianti and cousins
Status
- The Chianti computer now runs as a storage server
- The 2.6.12-rc1 kernel has drivers for the pcHDTV card, but we have no signal
- Installation history
Updates summer 2006
- Modified BIOS on Chianti and Prato
- In the opening menu, press Ctrl-F1 to get Advanced settings
- Advanced BIOS Features
- Advanced Chipset Features
- set AGP aperture to 64 (was 32)
- Somewhere else in the menu
- disable serial and parallel ports
- Power Management Features
- Resume on AC on -- set to Full on (power on when power returns)
To do in spring 2006
- Test if you can capture from all three card to the same drive
- Disk I/O could be a limitation
- Does it matter if I use the sata_sil drives vs the nforce3 drives?
- Seagate vs Maxtor?
- Tradeoff on journaled file systems -- see overview
To do in fall 2005
- Prepare to move SATA drives into enclosures
- get SATA to USB enclosures
- if you want to add a PCI card (you have no space), get stackable SATA enclosures ($30)
- get an eSATA (SATAII-compliant) external sata connector PCI card
- Avoid the sata_sil PCI cards as they don't work with Seagate
- By fall a suitable card may be available
- Test the digital part of pcHDTV
- Get the latest software from pcHDTV
- You may want a more recent kernel first
- See the March 2005 tarball in ~/software/tarballs
- Check out the
pchdtv-pvr project for hints
- See /usr/share/doc/xine/README.dvb.gz for using xine to watch digital tv
Inventory
- BIOS
- CPU
- AMD Athlon dual-core 4600+ (Chianti)
- AMD Athlon dual-core 4800+ (Prato)
- NIC
- Marvell Yukon takes the skge or sk98lin (all)
- hardware supports Wake-On-LAN, but the kernel's sk98lin or skge doesn't
- marvell.com's own Linux driver has magic packet support, but the driver doesn't work
- try "ethtool -s eth0 wol g" and "ethtool -i eth0"
- nforce3 250 onboard takes the forcedeth (Chianti only)
- misleadingly identified as a 100/10 ICS1883 PHY LAN in the manual
- supports wake-on-lan
- Sensors
- it87: Found IT8712F chip at 0x290, revision 5 (from dmesg)
|
Gigabyte K8NSC-939 |
Gigabyte K8NS Ultra-939
|
Chipset
|
nForce3 250 |
nForce3 250 |
Socket
|
939 |
939 |
CPU
|
Athlon-64 4800+
|
Athlon-64 4600+
|
PATA
|
2 x ATA-133 - 40 pin IDC |
2 x ATA-133 - 40 pin IDC
|
SATA
|
2 x Serial ATA-150
(nforce3 250)
|
4 x Serial ATA-150
(sil 3512 and nforce3 250)
|
RAID
|
sata_nv
|
sata_sil and sata_nv
|
|
GigaRAID ATA 133 RAID |
Max RAM
|
Dual channel DDR400/333/266 - 184pin
Up to 4GB by 4 DIMM slots |
Dual channel DDR400/333/266 - 184pin
Up to 4GB by 4 DIMM slots
|
NICs
|
Marvell Yukon 88E8001
|
Marvell Yukon 88E8001 ICS1883 (forcedeth)
|
USB2
|
4 x Hi-Speed USB |
4 x Hi-Speed USB |
Firewire
|
T.I. IEEE 1394a 6 pin |
T.I. IEEE1394a and b
|
Data bus
|
1600MHz |
1600MHz |
Sound
|
Realtek ALC850 Audio AC'97 Codec
7.1 channel surround |
Realtek ALC850 Audio AC'97 Codec
7.1 channel surround |
|
|
|
The nForce3 250 chipset has a built-in ICS1883 nforce NIC which takes the forcedeth driver. This didn't work under Debian-Installer, but it works fine in 2.6.12-rc1. Cf. image of PCI devices seen by the BIOS at boot. And it turns out to be a gigabit card! Very cool.
The
Silicon Image 3512 controller needs the sata_sil driver, which has
problems with several Seagate drives, which means they are deliberately
slowed down -- see http://marc.free.net.ph/message/20050310.044844.559a9ee7.en.html. My working Seagate drive is a blacklisted ST3200822AS. The current BIOS is F10; you can update here.
Hard drive speed
22 May 2005, updated 3 June 2005: benchmarking SATA and PATA drives
hdparm -tT /dev/sda:
Timing cached reads: 3616 MB in 2.00 seconds = 1807.37 MB/sec
Timing buffered disk reads: 64 MB in 3.03 seconds = 21.12 MB/sec
hdparm -tT /dev/sdb:
Timing cached reads: 3624 MB in 2.00 seconds = 1811.37 MB/sec
Timing buffered disk reads: 184 MB in 3.01 seconds = 61.22 MB/sec
hdparm -tT /dev/sdc:
Timing cached reads: 3624 MB in 2.00 seconds = 1810.46 MB/sec
Timing buffered disk reads: 186 MB in 3.03 seconds = 61.38 MB/sec
hdparm -tT /dev/hda:
Timing cached reads: 3692 MB in 2.00 seconds = 1845.36 MB/sec
Timing buffered disk reads: 186 MB in 3.00 seconds = 61.91 MB/sec
Looks
like /dev/sda is the sata_sil chip with the crippled Seagate drive --
really not ideal as a capture drive. The new PATA drive is just
as fast as the SATA drives -- in fact a bit faster. Booting should be a
bit faster now that it's on the PATA drive.
1 April 2005: I ordered an Abit AV8 K8T800 (Socket
939), but they sent us a Gigabyte K8NS Ultra-939. This is a more recent
board (Q4 2004 vs Q2 2004), with a few more features. In brief, it
looks OK.
System monitoring (ITE IT8712F-A)
- Super I/O: ITE IT8712F chip (driver it87)
- CPU and memory voltage, +3.3V and +12V
- RPM of 3 fans
- CPU temperature (by the embedded CPU sensor)
- Board temperature (by the on-board sensor)
The system supports the NMI watchdog timer -- cf. the output of cat
/proc/interrupts, which shows a non-zero value for NMI:
NMI: 666 783
See
/usr/src/linux-*/Documentation/nmi_watchdog.txt for information
about how to enable it -- you basically add the boot parameter
nmi_watchdog=1
to grub or lilo; cf. detailed instructions. The machine will then reboot if it gets seriously stuck.
See Netcell comments here and here -- superfast RAID!
|
|