BRL-CAD and hardware

[unixronin]

A few months back, Gentoo added a BRL-CAD 7.16.6 package, and after a few tries I got it to builkd and install.  I worked my way through the tutorial, then bame back to it next week and ... it wouldn't run.  Updates had broken it and it would neither run nor rebuild, and hasn't since.

Until yesterday, that is, when the Gentoo package was updated to BRL-CAD 7.16.8.  Yay!  It builds again!  So I have a working BRL-CAD again.  Unforetunately it's been not working for so long I've forgotten much of what I learned in the tutorial, and will have to get back up to speed again.

Next project:  Use BRL-CAD to model the planned rebuild of my computer desk.  I still don't have the money to spare to actually DO it, but with luck one or the other of us — maybe even both — will have a job again in another month or two.  I'm planning to construct a new top out of two solid-core doors, and new kick panels out of 3/4" plywood.

---

On the hardware front, I should have whitestar's disk back Monday, and babylon5's LCD monitor back next Friday.  Hannspree failed to surprise me at all by finding a fault on the main board, and replaced the board.  I got babylon5's third RAM stick back last night ... but the brand-new module exhibited the same problem.  I plugged it in, powered back up, ran a heavy compiler, and immediately started getting gcc internal compiler errors and kernel oopses.

Fortunately, I think I now know the REAL problem.  It's only a theory as yet, suggested to me last night by a Bitminer, but thinking about it, it seems highly probable:  babylon5's power supply simply isn't adequate to drive the full 3GB of RAM.  I've never had a problem with the power supply before, but on the other hand, it's a no-name power supply that could well be significantly optimistic on its output rating, and it's only rated 420W to start with.  According to Thermaltake's power requirements calculator, babylon5 may be drawing as much as 390W when it's working at full load, which means if that rating plate is even as little as 5% optimistic, I'm right on the ragged edge.

Unfortunately, I don't have a spare power supply — let alone a bigger one — to test this theory with, right now.  But when I can, I'll drop in a current-tech high-efficiency 600W-650W power supply and see if the problem goes away.  Doing so will also give me enough extra power to replace babylon5's (excellent at the time, but now rather antiquated) Matrox G450 graphics card with something several generations newer and more capable, with DVI or HDMI out, which should dramatically improve desktop responsiveness and video playback.

There's just that pesky money thing...

Comments

[tomatoe333.livejournal.com]

I've seen issues where the memory bus drivers in the Northbridge chipset on a motherboard cannot drive all memory slots to maximum capacity. There's simply too much load from the actual memory locations in the DIMMs. Load the box one slot short of full with the largest DIMMs, works fine. Load all slots with smaller DIMMs, works fine.

Back in the old days, they had buffered DIMMs, and this was why. Not sure if the same technology is still out there on newer DIMMs.

[unixronin.livejournal.com]

Yeah, this is the other possibility, a memory controller issue at maximum memory, but I have found no references online to such a problem occurring with the Asus A7V333.