[ Miles Barr ]
The content of this website is licensed under a Creative Commons Attribution-NoDerivs 2.0 England & Wales License.
Carson Workshops ran a Future of Web Apps summit on Wednesday. Sadly I couldn’t go because I was too busy at work, which is a shame because one of the speakers from Google was Steve Crossan who was one of the people who started Runtime Collective, where I work now.
Anyway this post is more of a placeholder to remind myself to read some of the write ups when I have some time. This looks like a good summary of what went on:
What’s a reliable video card for Linux? Google says: Matrox G550. Let’s see if I can find one on the cheap.
Update: Damn my greed. I found two G550s on eBay with a Buy It Now of £12, but there was also an auction ending this morning so I bid on that instead. It looked like I could get one for £4, but of course I got outbid at the last second and the two other auctions for £12 have been snapped up. Now the cheapest one is hovering around £30, maybe it’s a sign.
Just put the old card back in and the box failed to boot and spat out one long beep and three short beeps, which means “No video card or bad video RAM”. The video card is seated fine and I reseated it to make sure. How much do I hate computers, let me count the ways…
Turns out I actually have a GeForce2 MX-400, the GeForce2 Ti was living in a Windows box. I switched the cards and got the same response from the Ti. So that confirms it’s a driver issue not a hardware one. No I really regret chucking out that Matrox…
NVRM: loading NVIDIA Linux x86 NVIDIA Kernel Module 1.0-6629 Wed Nov 3 13:12:51 PST 2004 ------------[ cut here ]------------ kernel BUG at arch/i386/mm/pageattr.c:137! invalid operand: 0000 [#1] PREEMPT Modules linked in: nvidia snd_seq_midi snd_emu10k1_synth snd_emux_synth snd_seq_virmidi snd_seq_midi_emul snd_pcm_oss snd_mixer_oss snd_seq_oss snd_seq_midi_event snd_seq snd_emu10k1 snd_rawmidi snd_seq_device snd_ac97_codec snd_ac97_bus snd_util_mem snd_hwdep rtc sd_mod snd_bt87x snd_pcm snd_timer snd_page_alloc tuner bttv video_buf firmware_class v4l2_common btcx_risc tveeprom videodev tulip CPU: 0 EIP: 0060:[] Tainted: P VLI EFLAGS: 00213082 (2.6.15-gentoo-r1) EIP is at __change_page_attr+0×191/0×1b0 eax: 0000a420 ebx: c15d8a20 ecx: c100a420 edx: c1000000 esi: c0521eec edi: 00000163 ebp: eec51000 esp: ef039d68 ds: 007b es: 007b ss: 0068 Process X (pid: 20492, threadinfo=ef038000 task=ef020a70) Stack: eec51000 eec50000 2ec001e3 c15d8a20 00000011 00000000 00203296 c0116e36 c15d8a20 00000163 f5d94940 f8e80000 f6d1b800 ef039dec c01169d5 c15d8800 00000012 00000163 00011000 f8e80000 f937ee56 f8e80000 f9193180 f8e80000
Just upgraded to kernel 2.6.15 and got this message on boot up:
nvidiafb: nVidia device/chipset 10DE0110 nvidiafb: EDID found from BUS1 nvidiafb: Unable to detect which CRTCNumber ...Defaulting to CRTCNumber 0
After which it stalled. I’m guessing it’s something to do with the new nvidiafb driver option I saw in this kernel’s config. Why did I throw out that old Matrox G400?
I was at Heathrow last night and as we were taxing to our gate I saw Concorde out on the tarmac. I first I didn’t take much notice, then I realised it’s been decommissioned for over two years and supposed to be in Bristol. No one else seemed to notice. Very odd.
The power supply arrived. Switching it was simple enough. I figured I’d switch to round IDE cables at the same time to improve airflow only to be thwarted by the immense number of cables attached to the Tagan. It was nice to be able to boot my PC from cold without having to wait for the voltage to settle and it looked like everything was going smoothly…then X wouldn’t start, I just got a blank screen. This has happened before but a reboot got things going, I always assumed it was something to do with the lack of power. This time rebooting had no affect.
ssh’ing into the machine showed everything was normal, no errors in the logs. X wouldn’t die though and needed a ‘kill -9′ to go away. Switching from the ‘nvidia’ driver to ‘nv’ got X up and running but after a few minutes it freezes and can’t be killed. I did have it up long enough to test the TV card and was glad to find out the static problem I was experiencing was another power issue.
The nv freezing problem might be an old one. It was commented out in my XF86Config (running xorg) so obviously I decided it wasn’t good enough sometime in the past, but why? Probably because it froze.
Given that the blank screen is only constant after I switched power supplies I would have to suspect that, but everything else is running smoothly, and Google thinks I’m not the only one suffering this problem:
http://www.google.co.uk/search?q=xorg+nvidia+blank+screen
There must be an answer somewhere out there.
It looks like the power supply in my desktop is on its last legs. It’s been struggling with the 5V line for a while. Normally I’d have to turn it on, leave it for a bit until it settled on 4.6V and then boot up, now it’s struggling to make 4.5V, and that means a lot of crashes. The old one was an Antec TruePower 380 that I got when the cheap that came with the case blew up a couple of years back. It’s done the job admirably, but as I’ve moved hardware from my Windows desktop to this one, it’s become just too much.
380W from a good quality power supply should be enough for anyone but at last count the hardware in this box is:
I tried unplugging various things last night but there was no change in the voltage output, so I suspect it’s permanently damaged. After some research I’ve decided to go for a Tagan TG480 U01. I considered some higher wattage PSUs but this thing can deliver 48A to a maximum load of 240W on the 5V line, that’s 18A more than most, so I figure that must be enough for whatever hardware is sucking up all the juice. The Antec claimed to supply 35A but the total power was shared amongst all lines. As a bonus the Tagan is quite and cheap so it’s a no-brainer. Let’s hope it does the trick.
