AmigaOne X1000 booting AmigaOS4.1 Update 3

[koaftder]

Wow, that was underwhelming, almost painful to watch. The top of the line, brand spanking new, uber expensive A1X1K trundling along like an underspecd PC fresh off the discount rack from a department store somewhere.

[koaftder]

I don't know about y'all, but if I just spent a few hundred thousand dollars and 2 years of time building "the dream machine", I wouldn't have videos of the thing running like crap floating around. If they don't have a release build that can up and boot quickly by now, they're in worse shape than I thought.

[koaftder]

I can see that some non-technical people may not understand the development cycle and may genuinely believe that.  Some people in this thread I'm sure DO know it, and realise that what they are saying is a nugget of purest FUD.

Remember, this is a beta test system.
It will have serial debug output, probably at 9600 baud.
It could easily be tracking every single call to every single library, and dumping the text at that speed, which could take about 6mS.

That means after just 1000 calls, you'd have a 6 second delay.

Now tell me again how that is at all indicative of the final non-debug speed of the system, please!

:roflmao:

If you had to hang on spooling out text to a serial port clocked at 9600 baud to print out a message *every* time a library call was made you would wind up with a completely unusable system.

[koaftder]

3. Hyperion or their partner should have already had a pro video out about this system. Especially after all that silly "What is X" hype.

I seriously doubt they have anything worth showing. The only seriously redeeming feature a NG Amiga system that would impress anyone would be an absurdly fast boot time. Press the button, wait a couple of seconds and bam, there's the desktop. It's not like there's a whole hell of a lot going on in there. If they could show that, they would have.

All these excuses about it being a debug build responsible for the machine running like utter crap is just lame. There must be some really atrocious code flung together to get this running on the pa6t.

[koaftder]

Exactly! so if you only log say 10% of your calls then you still end up with what appears to be a very slow system, which is what we see here.

Q.E.D.

I can't think of anything more useless than lobbing a bunch of printf's to the serial port for debugging. That's bush league stuff. The processor has a JTAG interface for that kind of thing.


That's probably what that 16 pin header in the middle is for.

Here's a product that makes use of it:
http://www.abatron.ch/fileadmin/user_upload/products/pdf/ManGdbPA6-3000.pdf

[koaftder]

A debug build will always be slower, that's not an excuse that's a reason.

These are BETA builds, not release. They have debugging kernels.

If you don't believe me I challenge you to ask any kernel or os developer how fast their beta, debug, serial output enabled machines ran.

The reason debug builds are slower is because they're compiled without optimization, but non optimized builds aren't that slow and this is a fast processor with gobs of ram and not starved for IO. The serial port has nothing to do with this. You wouldn't use the serial port to debug system code anyway.

[koaftder]

Not a good day for OS 4 users.

[koaftder]

The ebb and flow of the popularity of various processor architectures has more to do with IP licensing and business processes than their technical details.