The collected works of ClassicHasClass - Page 12

gcc's double-double on PowerPC will work on 32 and 64-bit PPC, so that should be fine. Conversion to and from PPC double-double to 80-bit long double should not be hard, and in your code should just involve changing fpu_bytes_to_long_double() and x87_to_motorola(), if I understand your program flow correctly.

I just use a video digitizer with a built-in TBC. My Canopus ADVC-300 works with my Fuel over FireWire.

Just install a USB one. That's what I use. Virtually any Bluetooth dongle and most 802.11g or lower dongle will work.

You know what they put on French fries in Holland instead of ketchup? Mayonnaise. I've seen 'em do it, man. They f*ckin' drown 'em in that sh*t.

Sugar on popcorn, too. Then they claim that fast-food is bad for you ...

But that's just, essentially, kettle corn, isn't it? (Though I'd argue that's even worse than salty buttery popcorn.)

I recognize the quote, but man you make my mouth water

Some areas of the USA have something called fry sauce, which is roughly mayonnaise and ketchup mixed in equal parts. I rather prefer it to regular old ketchup, and I've found just plain mayo with my fries is pretty damn good too.

My Solbourne S3000 needs Perl, but Perl 5 will not build with the built-in ancient K&R-only cc. Does someone have a binary around of this? I don't need the whole distribution, just the main perl binary for SunOS 4.1 (to be specific, OS/MP 4.1C, which is binary-compatible with SunOS 4.1.2 and 4.1.3).

Any version of Perl 5 accepted, 5.00504 or later preferred.

I do have gcc around here somewhere, but I don't want to install that whole pig just to build Perl if I can avoid it.

I'm looking forward to the "whitebox" POWER8s. I might even think about picking one up in a year or two when the prices drop, though my POWER6 is doing just fine, thanks.

AIX does virtualization very well (see: LPAR), or you can use something like PowerVM on Linux.

I'm mostly an AIX thug though ("loves the jackboots," if you get the ancient reference ).

That's a good choice, actually.

I haven't had cycles to look at this recently (darn Master's degree) but I'm going to pull the new release when I get a spare minnit.

Yup, the z/Machine is hardly dead.

I'd like one in my house, but I'll settle for a POWER8.

So the POWER6 threw a rod (anyone know service request code 11002624? I suspect the RAID card, because the RAID array is not coming up) and of course on a Friday afternoon when my sales rep is on Eastern time and has gone home for the weekend.

The ANS is temporarily doing server duty, just like it used to. Fortunately, the POWER6 had done its usual backup automatically, so at most I lost an hour or two of E-mail.

It's really not as revolutionary as it sounds. PPC has technically always been bi-endian though some hardware adjustment was required.

Besides, even on POWER8 AIX and IBM i are still big-endian, and I can't imagine Linux won't still run in big-endian mode even after the "transition" if you still want to.

ASMI says it could be either: the sidecar Ethernet, the RAID card(s) or the system backplane. Gonna be an expensive week.

I'm consoling myself waiting for spare parts for the POWER6 by attempting to study for a final exam and setting up my brand-new-to-me-and-you Amiga Technologies A4000T! It's not the original, unbelievably hard to find Commodore A4000T, but the Escom-era A4KTs are hardly growing on trees, and because this is one of the later Quikpak units it is substantially more powerful. From the factory, it came with a Quikpak 68060 CPU card (50MHz part overclocked to 56.75MHz), SCSI HD and ATAPI CD-ROM, plus 6MB of fast RAM and 2MB of chip RAM. This one was upgraded to 16MB of fast RAM and also came with a Picasso IV combination video card and flickerfixer, and a Hydra NIC; it also included its manuals, AmigaOS 3.1 disks and 3.5/3.9 CDs, and cost an ungodly amount (but about the going rate for a system of this spec). I had to fix the drivers for the CD-ROM and NIC because the previous owner apparently wrecked them, but it all works now!

And here it is, running AmigaOS 3.9, surfing important websites!


And here it is, after I correctly reprogrammed the horizontal frequencies for the scandoubler, freaking out Sysinfo! (SysInfo 4.0 still gets some of it wrong. They didn't call it Mis-Info for nothing.)


It's really very nice now that it's all working properly, so the A3000 got bumped from its slot in the KVM. I still intend to put Amix on the A3000 one of these days, since I do have an A3070.

In the meantime I'm going to install a Jaz or ZIP drive and some more RAM and hard disks (a 128MB Amiga sounds tasty!), and maybe the Video Toaster 4000 I have rotting in a box, and then get some WHDLoad games running on it for serious businesses!

There's something to be said for the last "classic 68K Amiga," even as a PPC-o-phile.

Yeah, this was ... an expensive purchase. Not the most expensive vintage computer I've ever bought, but I've passed kidneystones easier than this.

You could get an '060 in an Atari ST, I think. There was never one for the Mac, though: the missing instructions would have meant a lot of software trapping.

Some of the Alpha Micros had them too, my particular favourite obscure computer system. Wikipedia mentions they turn up a lot in TV chyron systems.

I got ADoom running on it tonight. Combined with the Picasso's speedy video, the integer performance of the '060 really shines (remember, Doom doesn't use the FPU). It is noticibly faster than my 133MHz 486-class Am5x86 "P75," which is running Doom in DOS and doesn't have to multitask!

Serious businesses!

Very nice. I hope the gremlins stay away. Skins look pretty good in that photograph.

Yeah, that's all the IBM tech docs say too. They recommend repair procedure PWR1918 which on the x20 units is "replace everything."

If you guys are gonna drool like that, I'm never inviting you over.

Btw, just discovered this: http://aminet.net/package/comm/www/NetSurf-m68k
128MB RAM ordered!

Having not experienced ElectroPaint in the flesh (I should go put it on the Fuel now), how close is StonerView, since that has both Mac PPC and Classic versions?

It's probably already on your Fuel - try running "ep" at the command line (if you like that, try "bongo").

Bongo like candy!

I'll try it tonight after the parts have come in for the server.

I've been tempted a lot by the Tez but my limitation is space, heat and noise. The G5 is already pretty damn hot and noisy when it wants to be, and the Tez is much fatter.

I have a p520, but it's a POWER6 (currently awaiting a new system backplane, sigh -- and the RAID array may be b0rked).

When you say licensed options, do you mean AIX? If you own the hardware, you are entitled to run AIX. Any CD set will do. The POWER5+ p520 can run 5.3 onwards and some very late TLs of 5.2. However, you'll be running it straight on the metal, no LPARs.

No, I'm not using LPARs; it's "AIX on the metal." I don't have an HMC.

So the POWER6 (which the Apple Network Server 500 is subbing in for) did indeed blow its system backplane. Unfortunately it appears to have taken the RAID 5 array with it. There is data still in the auxiliary battery-backed cache, but the cache directory is apparently hosed and SRN 2505-9050 popped up in the logs, indicating it does not recognize the cache data as belonging to the array. The associated MAP 3131 to resolve it strongly suggests I'm more hosed than a fire truck at a gay pride parade, with ugly words like "data loss" and "delete then recreate array."

Diagnostics shows the disks are there, and recognized as belonging to the array, but the array itself is listed as Failed.

My IBM tech friends aren't sure what to do with it either, but I thought I'd ask here in case anyone is a POWER Systems god. The system was relatively quiescent the day it went bad, so there shouldn't be a LOT in that write cache, mostly log files (the 57B7/8 card pair has a modest 175MB of write cache). This is AIX, so these are all JFS2 file systems.

The way I figure it, /, /usr and /opt haven't seen much action since Obama's first term. They should be clean of writes; there shouldn't be anything in the write cache for those partitions. /tmp is expendable. /home is backed up daily, so I can restore it. The logs in /var are almost certainly toast, but the database hasn't been written to in several weeks, mail is backed up several times a day, and the web server and gopher server are backed up daily and weekly respectively. I don't care about the journal volume or the paging volume, since I already assume those are fried.

So, given this, my thought is to just reclaim the write cache and wipe it, and fsck and hope for the best. The drives still organize themselves into an array, just a failed one. The journaling should keep the file system in a sane state, even if I've lost some writes.

Or do you think I'll be rebuilding the server from scratch and backups?

I have a mksysb image of the OS, patched up the way I like it, though it probably got some minor modifications that I don't have in that .iso. So I'd rather not rebuild from scratch if I can avoid it, though it does look like my backup strategy is at least adequate ...

What I'm trying to find out is if anything other than write cache is in the RAID, and how often it gets written out. I'd hate for it to queue up a full 175MB of data before it tries to emit any, and I'd really hate for it to be caching things like the superblock.

On the p520 and I think p720, the planar can be either "flat" or RAID. There is an enablement card set that you can install as an option (that's the 57B7/8 in mine) that enables the planar backplane RAID and provides the cache.

SMS (and for that matter, Open Firmware) does not know how to manage RAID controllers; a crashed array will in fact be completely invisible to SMS and won't even show up when you list devices. You have to boot from the AIX diagnostics CD to view the status of the RAID array and attempt repair. This is not the same as an AIX install set. Incredibly, you can download an ISO of that from here, at least until IBM puts that under support-contract-only too: https://www-304.ibm.com/webapp/set2/sas ... /home.html

The component drives show up as RWProtected in AIX diagnostics, not Failed -- only the array itself appears as Failed. This is the designation the SAS RAID uses for a drive that is still part of an array, but the array has failed "safe," so the controller will lock it down until repairs are effected. See, for example, http://pic.dhe.ibm.com/infocenter/power ... ration.htm

I'm going to test the individual drives, too, but I very much doubt that's the problem. Near as I am aware, you can't set the individual status of the drives manually, at least not in AIX or System i; the controller has to be "happy" first before it will release the protection lock.

Well ...

IT WORKED!

It doesn't look like I lost much of anything. MySQL is unhappy that it didn't shut down cleanly, but other than that the system looks functional. Hooray for JFS!

At least it's good to know that wiping/reclaiming the controller cache is a recoverable operation, if it has to be done. That's not at all clear from IBM's technical documentation. JFS' resiliency as a file system is probably a big part of that.

Oh well, guess I remain the resident AIX dweeb.

Well, there probably was at the time, and I imagine their current non-POWER enterprise RAID offerings are much the same. But I appreciate the kind thought

I'm less worried about the computer and the cost and more worried about a leak in the tank.

cron job that does osascript -e 'tell application "Finder" to sleep'. T,FTFY.

First, I have administered various AIX machines going back to 3.2.5 and the real RS/6000s with RSC CPUs, even prior to the PowerPC 601. I don't claim to have any experience with the z/Machines and their other really big iron, but I know my stuff with the small-to-medium size POWER boxes. I've rebuilt some of those servers from operating system tape on up. Later, we administered AIX 4 systems, and I even owned one personally (the Apple Network Server).

So when the time came to upgrade I bought this POWER6 because I like the AIX jackboots, and when it's configured right, it's a freight train. It's not very flexible, you really need to know what you're doing to get it started and it doesn't do well if it's not on a secure set of rails, but if you set it up properly and point it in the direction you want it to go, it's tough to stop.

But I'm sick of IBM's new licensing crap. It wasn't this way before. You bought the server and AIX came with it. You could download patches and APARs for free. If you have to swap out the hardware, you got what you put in it. If you wanted to upgrade it, you just bought the stuff and put it in. You didn't have to activate things, you didn't have to d*ck around with an LPAR not being licensed for however many cores.

When the POWER6 blew, the replacement board had an ancient firmware version on it that couldn't untangle itself from an HMC, and I couldn't configure it from ASMI to boot any operating system, and because I didn't have an HMC, I couldn't update the firmware because I didn't have permission from the HMC which wasn't there. So I reset the board to get control. I could update the firmware, all right, but I lost activation for the second core which I PAID FOR, DAMMIT. ASMI won't let me reactivate it because apparently the 8203 has to have it done through an HMC. So I'll be paying to buy an HMC to manage a single system EXACTLY ONE TIME to relicense the second core I already paid for! And you can bet that HMC I buy today (which is really a vanilla piece of crap 1U Xeon running the IBM software) won't work with future models.

Plus, IBM is now locking most of their website behind serial numbers, and if your serial number isn't one of their paying zombies, you're out in the cold. No more fixes and APARs for free. No more firmware downloads. IBM already didn't talk to people looking for direct sales, and now they won't talk to anyone without a service contract. Because by golly, they really hate people like me who can do the work on their own hardware without calling Big Blue and being financially violated.

I'm looking forward to OpenPOWER. Really, really, really looking forward to it, because I can run my favourite architecture and tell IBM to suck it. Even if I have to run Linux and can never run AIX again, I am so sick of this bullsh*t.

Too late, it's already purchased. Besides, my vendor, who are not jerks like IBM, actually cut me a very nice deal and then I don't have to worry about any glitches if turns out the VBoxed HMC software doesn't behave exactly the same.

Hopefully OpenPOWER launches and hangs on for at least one more processor generation (POWER9 or so), and then I can jump to that last generation until the bitter end.

Have to ask: are you sure nothing's wrong with the DIMM or the socket?

Yes, I'm pretty sure they have to be matched pairs.

Gawd, those things infected my highschool days.

MD.

I pick Door 2.