René Rebe

Did not even imagine to find the CPU spec of Apple’s hobby on their website, but they indeed list it: Apple A4!

You need to add Linux system (router, server, etc.) directly onto an 802.1q VLAN trunk? Well, for one there is this old (read: deprecated) vlan, the new way is simply via iproute2’s ip:

ip link add link eth0 name eth0.1234 type vlan id 1234

As usual: just substitute the ethernet interface name, alias, and VLAN ID as needed

Normally the performance gain of going from 32bit (i386, aka i686 etc.) code to 64bit amd64 (aka x86-64, EM64T, or the likely to mismatch x86, …) is not soo huge. However, unlike classic RISC CPUs -which usually loose performance when comparing their 32 and 64-bit- code due to bigger instructions and thus data bus saturation. For amd64 AMD did a great job defining the ISA (Instruction Set Architecture) in a way that 32 and 64bit operations can be freely mixed. Thus normal programs that do not benefit from 64bit operations remain compact, while single instruction can freely utilize 64bit by adding a REX prefix (similar to the 0×66 prefix since the 386). So normally only data moving loops benefit from going 64bit, e.g. multi media codecs, encryption and such.

Doing some performance analysis on a new AMD board I spotted an rather extreme case: openssl’s RSA 2048 bit encryption shows an over 300% boost while going 64bit:

From:

rsa 2048 bits 0.007128s 0.000212s 140.3 4722.2

To:

rsa 2048 bits 0.002380s 0.000068s 420.1 14716.7

Yes, that the same machine (AMD Phenom II), exactly same software stack and such, a whooping 312% increase!!!

So we got some new PC server mainboard, and the latest greatest AMD Phenom II X6 CPU. Of course the BIOS would not recognized the CPU, just list an “unknown” CPU, and let it run at a bare 800MHz. Apparently no real OS wanted to boot in protected mode either. So I thought it’d be a good idea to update the BIOS to the latest, greatest. However, turned out that wasn’t so much of an good idea: after the update the board would not boot up anymore. No sign of the BIOS at all, …

The board vendor, however, was so nice to handle our support inquiry nicely and sent out a new 8-pin, serial EEPROM last Friday, and it even arrived the following day, that is Saturday!

So with the new EPROM in the board it actually booted again (puh!), and re-reading the BIOS Release Notes I found that it indicates running the DOS flash EXE with some special arguments, that I obviously did not include when I just run it intuitively the fist time:

PFUDOS.EXE FILENAME /p /b /c

How nice. If I would author some BIOS flash utility, I would rather write it in a way that a run with default, that is without fancy parameters would produce a reasonable, good outcome, …

Anyway, another note: After booting I removed the second, new, good ROM and injected the old, bad flash, and then used above run with fancy parameters to flash it again, and voila: I finally had an BIOS ROM with the latest version that worked !

And yet another note: Flashing from an bootable USB stick with FreeDOS worked just fine.

Sooo, with the latest AMD V105, Nile platform come the AMD Netbooks. 64bit, and such, nice, finally! If, well, if the manufactures would just learn that 1024×600 (or even worse x560) doesn’t really cut it, anymore. Never did. Heck, even the iPad got more pixels real estate! Without Intel’s silly Atom screen resolution restriction at least 1024×768 should be in order (while even that is so last millennium). There is so much screen bezel to cut down.

Personally, though, I find the Nokia Booket’s 1280×720 pretty usable abroad. Wouldn’t want to miss a single pixel of it. It’s Just it’s (Intel Z-series Atom) CPU could be a little more performant, …

Is that they apparently are often only not so well integrated facades, with poor error handling. When they work, they are nice and pretty, but woe some system-call returns an error, or backend program quits unexpectedly. Then those shiny facades become annoying, hard to control monsters. Two cases in point:

As I write this I’m trying to sync my father’s in law iBook files to a brand new MacBook. Thanks to the dismiss of Firewire I may do so by the means of the Migration Assistant via Ethernet. So far so good. Until you hit the “an system call returns an error” condition: Turned out the assistant just did not want to run thru yesterday. Hanging at 99%, 1 minute remaining the whole afternoon. One could not even cancel or quit it (the “woe something goes wrong and your are lost” condition). I’d thought I give it a second run over the night, maybe it completes. Of course it didn’t, … The iBook system log revealed, i/o error, certainly some sectors gone bad. How nice. It certainly could have skipped some unreadable files. As resolution I unpacked my Unix skills and rsync‘ed the whole user’s home directory to our office file server. Turned out it where just 2 files (images in the photo library) rsync spotted with i/o errors and skipped over gracefully, warned about at the end. Actually rsync was coded with some extra intelligence, to re-try files with errors a second time at the end! Btw, from the system log it looked that the Migration Assitent uses Racoon/IPsec internally to encrypt the transfer internally, over the air or on the wire. I somehow get the feeling the errors handling in Apple’s assistant is not so well tested after all.

Which brings me to the second example. Actually also a bad hard disk: My brother recently got a brand new Mac mini. He unpacked it and wanted to apply his Time Machine backup from the sold MacBook (Pro I think). He tried all, again and again, but it just did not want to run thru. So without his files restored he started to work on the machine, which inhabited a pretty odd behavior, even I had not seen before: every now and then, like every few minutes, or just once an hour the whole machine would lock solid, with just the colorful mouse cursor ball spinning around, no audio playback, nothing, for a could of minutes. And then, suddenly, as if nothing happened, it would continue to operate normally. My brother was already totally perplex and mad on the just new Mac mini when I came over the other day (it’s 300km, IIRC it was around CeBIT 2010). I’d also not know what to make of it. Running too hot? Noisy PCB wires resulting in stray, random memory content? Though the later would shure rather result in kernel panics. Anyway, my first loved click on the Console to browse the famous system log an there it was: i/o errors every now and then, … (on a just onboxed Mac mini!). The world’s most advanced operating system, that just got even better really could have pop’ed up some nice information dialog, or put some yellow exclamation mark on the hard disk icon, or the top menu, whatever, ….

Somehow I see a repeating pattern here: disk quality issues paired with notorious bad habit not to check for, and handle system call error conditions gracefully.

Finally. So it’s only about 1½ years until we “might” actually get our hands on them, …

ExactCODE just released another major ExactScan product family update: the new version 2.10 brings the teased major surpise Most notably is the new and novel TWAIN Bridge. It allows to utilize the many built-in drivers built into ExactScan from other, third-party TWAIN applications.

Another magical feature is the new flatbed de-skew. While previous versions of ExactScan already came with, what I would call, industry leading auto-crop and de-skew for scan with the built-in drivers from the ADF (Automatic Document Feeder), recent advancements in our in-house R&D (Research & Development) allowed us to add truely revolutionary de-skew even for flatbed scans! It allows to intelligently track objects on the flatbed glass and auto-crop and de-skew sufficient rectangular objects, such as: letters, receipts, post-cards, CD covers, coaster, etc.

Of course we also continue work as usual on the next maintenance update.

Read more: ExactScan homepage

So Android 2.2 adds JIT, Just-In-Time compilation of the Dalvik Java VM. This must be a joke. They tried to compete on the mobile space with that slow apps? Against other, native, frameworks. No wonder Android was that lagging. And it took them over 5 years to get to JIT? I’m feeling lucky I didn’t got an Android phone