Updating video driver FAIL.

Thought I’d update the 3 year old video driver on my 3 year old ATI HD3650, which does the job on my business-use VM running machine under XP64.

Less than delighted to find VMs crashing the host when I switch them to full screen console mode, as I do when I use Emacs.

Clearly a case of minimal testing of the latest shiny drivers (11-5_xp64_dd_ccc_ocl, April 2011 build) with older hardware/OS configurations.

Back to the old driver. The new driver hadn’t fixed a minor mouse issue with VMWare anyway.

WebScheduler and Windows 7 x64

Having achieved good video playback on my new ION 330 HTPC, the next step was to get WebScheduler running. Being a 32 bit app that had to work with 64 bit BDA drivers for the TV tuner cards, I was concerned about it all working properly.

I started with WebScheduler 4.0.14, the last “free” version before it had gone commercial a few years back. This version has served me well under XP32. It installed without a hitch and ran fine, the only glitch I noticed was that the taskbar icon didn’t reflect when WebScheduler was recording.

It recorded files fine and I left it running about a week while I worked on other stuff. It wasn’t until the weekend that I discovered that the recordings it had made were unplayable!

I determined the recordings were not total trash, just Haali Media Splitter didn’t want to know about them. Using the Gabest splitter or ProjectX to demux them both worked.

So it seemed something was different between the files recorded under XP32 and Win7 x64, both done with the same version of WebScheduler and the kWorld USB Dual DTV Tuner stick capture hardware.

Having determined I was on the latest driver for the KWorld tuner, I was resigned to having to try Win7 x86, but before I did, I thought I’d try the latest version of WebScheduler as it had become open source again. Its now known as TV Scheduler Pro, by Blue Bit Software.

To my total delight, the recordings came good. I was able to export and reimport my schedule match lists between the 2 versions, which happily co-existed on the same machine for a while while I transitioned. The new version has some nice new features and I’m glad its recording properly under the new operating system.

So now recording and playback is in place on the new HTPC.

Drive letters under XP

It can be a nuisance transferring windows to a new HDD for a system when that windows has previously seen that drive as an external.

It boots OK but then has the wrong drive letter for the user profile so it can’t login, since Windows “knows” the drive by the letter it was last used as, not the default C:.

In this case I was updating a laptop with a bigger drive, which I’d previously used as an external USB drive for it.

In this past I’ve used a BartPE XP boot CD to get into it but in this case I got it to allow login by plugging in the original drive as the USB drive. It could now find its profile and let me in.

Then in regedit I removed everything in HKLM\System\MountedDevices (maybe best to just remove the \DosDevices only) then it booted fine from the internal drive.

JQS.exe – huge I/O counts

On my HTPC and work VM, I enabled monitoring I/O Reads and i/O Writes in task manager and noticed that a process ‘jqs.exe’ had something like 30GB against it after just a day of running!

This illuminated why:
http://forums.java.net/jive/thread.jspa?threadID=36505&tstart=-1
it keeps rereading some files to keep them in the system cache. What a hack!

Control panel – Java – Advanced – Miscellaneous has the checkbox to disable the quick starter.

Cant believe all these apps and their ‘quick starters’ that want to bog your system.

Laptop With Virtual Issues

My 5 year old Tecra M2’s hinge gave out rendering it useless for travel and meetings so I started sourcing a new laptop.

I was surprised how complex buying a laptop can be these days, companies like Toshiba have dozens of models with minor differences in CPU speed and feature set.

I finally settled on a 2GHz Satellite Pro A300 with 4GB RAM, 320GB HDD and 15.4″ 1280×800 widescreen. I would have preferred something physically smaller but the high resolutions and smaller screens just wont work for me and my vision.

All seemed nice; I duly wiped Vista and “downgraded” to the provided XP32. I know I’m wasting 1GB of RAM with a 32 bit OS but I plan to take it to Windows 7 64 bit which Toshiba had better support!

Running Steve Gibson’s securable revealed something nasty. The 7350 Core2Duo CPU doesn’t support hardware virtualisation. ARGH! Something I never thought to check for, having assumed any recent Intel chip would support it.

Performance wise it didn’t worry me much, VMWare’s binary translation works fantastically for me. But it meant I would not be running 64 bit VMs on it.

Without hardware virtualisation support, you cannot run a 64 bit VM on any host – 32 or 64 bit – with either VMWare or VirtualBox.

The binary emulation layers simply haven’t been written for the 64 bit opcodes, not with the availability of hardware virtualisation on most modern CPUs. Except mine.

Fortunately I don’t need to do much in 64 bit apart from test 64 bit builds of Planimate®. I can compile them from a 32bit OS at least. I had partitioned the drive so I had a handy alternate OS partition. I did a test-install of XP64 on it and it works fine. So… if I’m on the road and really need a 64 bit OS, I have one available though none of the Toshiba tools will run under it.

Apart from that and the weight of the 15.4″ screen (and the chunky power supply) I’m happy with the laptop. It was well priced so I was able to get a second screen and extended warranty for it.

Seagate update

Seagate got their act together and released new firmware for their drives. They even replied to my email. I applied it to both my 500GB drives. Both went OK but the original fault in one of my drives remains – bad blocks, chattering sounds even when its idle.

I ran SpinRite on level 5 (reclaim lost sectors) over it for a few hours. Then I restarted SpinRite to test how it had gone… the bad sectors were still there. So the drive is toast and worse still, its S.M.A.R.T reports the drive is imminent to fail.

I don’t know if I’ll bother returning it. I dont really want another Seagate! As for the other drive, I’m going to be testing it regularly and watching it very carefully.

[EDIT April 2009].

I’ve been following the Seagate forum discussion on my ST500320AS drives. One of mine is still working and the other is useless with bad sectors. Kind of given up on it being resolved by newer firmware but I might get around to returning the drive, hopefully at a time in the future when they’ll find it easier to replace it with an SSD đŸ™‚ (wishful thinking).

Seagate Drive update crash followup

I guess I was lucky the 500GB Seagate drive update crashed… had it worked it might have bricked the drive

“This will be firmware to fix the firmware (SD1A) that was issued to fix the firmware (SD15) that caused some 1TB Barracuda 7200.11 drives and others to stop working. The SD1A firmware caused 500GB Barracuda 7200.11s to stop working.”

Woah, exactly the version of my 2 drives..

Latest on the Linux GL824 driver

Its been a year since I started working with a small group on getting a Linux driver going for the GadgetLabs GL824 8 channel sound card We’ve got card syncing going so you can have 2 or 3 cards synced for 16/24 channels playback.

During the break I messed with it a bit more as Mike, the project leader, is working on getting a WFS system going – using a speaker array to project sound into free space.

32 channels is a minimum so we’ve been pushing the envelope and working on a system with FOUR GL cards. Up until recently we’ve had no luck getting the 4th card working but we’ve now achieved playback on 32 channels at 44.1k.

The initial blockers included a few driver issues and apparent PCI slot sensitivty but the main one appears to have been that the sync-clock signal (passed from card to card with link cables) degrades from card to card.

The sync signal is generated by the first card in the chain, the master. Its around 16MHz for 44.1K and 18MHz for 48K. Each card filters and buffers the signal before outputting it again, so its not a level problem.

I’m currently working on the theory that each card in the chain introduces jitter to the signal. So by the last card theres enough to upset the DACs on it which causes a continuous static.

After some messing about we’ve stumbled upon a combination of cards and sync cable order where at 44.1K its workable. I’m thinking we need to build a small clock buffer board  (eg with a 74Ls14) and then distribute the clock in parallel to the 3 slave cards.

Mike is already imagining syncing 5 cards in one system, or multiple PCs …

This is all for playback. For recording, we’ve found the GL has a very poor transfer rate for samples from the card to the host – 4 times a slow as transferring samples to the card. This makes the maximum practical number of inputs 16 (2 cards).

Also we’re finding that later Linux distros like Fedora Core10 (x64), Ubuntu 8.10 (x64) have made a bit of a dogs breakfast of the ALSA tools, for example ‘arecord’ just doesn’t work properly but the graphical apps are OK.

I’ve gone back to FC8 and get good realtime performance from both the oldish CCRMA realtime kernel 2.6.24.7-1rt3.2 and the lastest FC8 update 2.6.26.8-57. With these and using qjackctl (with its realtime option selected), my system is pretty free of XRUNs.

Windows just can’t take care of itself

I have an old TECRA M2 laptop which is still barely adequate for the occasions that I’m away from the desktop. Its was still on its original install of XP (mid 2004) and generally has run OK.

Occasionally I’d rolled it back to an image of itself if I didnt like some software I’d installed or the occasion or two where it had messed itself up. These images were generally less than a few months old.

In the last few months I’d noticed it was taking longer and longer to launch applications. I’d always tried to keep it “clean”, not installing unnecessary junk, using a registry cleaner (Registry Mechanic) as well as manually preening out junk files, settings and registry settings.

Alas it still had this huge delay. I determined that it wasn’t the launching of the application that was the problem, it was resolving the file association. It happened with a couple of types (PDF,JPG,TS) and un/reinstalling the apps (FoxIt,ImgView,ZoomPlayer) didn’t help.

I also noticed it wasn’t really disk-bound, it just sat around for a while before finding what to launch. The usual malware scans turned up nothing.

Ah well, I decided to take it back to an image I’d taken shortly after getting it, more than 4 years old. Fortunately this avoided a complete reinstall, which is a pain with a laptop and its many support utilities and drivers.

A couple of service packs and 30 or so critical updates later, its now nice and fast again.

It really does prove that Windows systems do not age well; somewhere something had become bloated, fragmented, unbalanced or plain old corrupted beyond what defragging, cleaning and uninstalling would fix.

Whilst I’ve taken a new image of the refreshed system, I wonder if I’ll have it long enough to ever use it.