Login
Back to forumReply to this topicGo to last reply

Posted By

Krill
on 2011-09-11
15:40:27
 Please help testing: New loader release imminent

Hi,

as you might know, i have recently ported my C-64 IRQ loader to C-16/116/+4.
An earlier version was used in States United by TEK and Bauknecht.

Before releasing it, i'd like to make sure it works alright on random actual hardware (64K needed).

Basically i only need to know on which hardware combination it fails.
Some information about exotic hardware which works with it (anything that is not 1541/71/81) is also appreciated.

I'd be grateful for all reports.

The test images can be found here: http://krill.earthtomoon.net/loadertest-v114.zip.

Thank you!

Posted By

Degauss
on 2011-09-14
11:14:21
 Re: Please help testing: New loader release imminent

I try to dig out my 81 and 71 to do some testing. Might take a while since i don't know where i stashed those...

Posted By

Krill
on 2011-09-14
13:14:22
 Re: Please help testing: New loader release imminent

Degauss: Thanks! Try 1551, too, if you have one. happy Well, any drive, but that's clear anyways.. happy

Posted By

Gaia
on 2011-09-14
15:23:14
 Re: Please help testing: New loader release imminent

Does it support the 1551? That sounds intriguing! Unfortunately I am yet left with no parallel port and I have yet to order one of those nifty new USB transfer devices...

Posted By

Degauss
on 2011-09-14
15:43:48
 Re: Please help testing: New loader release imminent

Thing is: Me and my brother have a rather huge bulk of commodore stuff. Like 2x 1581, 6x 1551, 2x 1541 II, 2x 1541 and so on. That obviously occupies a considerable amount of space and at some point we had to put all of it into boxes to stash it in the cellar of our parents house. If you could "hint" on "preferred hardware combinations" it would be easier for me to dig the desired devices out of those boxes.

p.s.: Did you say you wanted to create a webinterface for customized builds? That'd be great (as i had some trouble building from the sources).

Posted By

Luca
on 2011-09-14
16:00:48
 Re: Please help testing: New loader release imminent

OMG that means in the heart of Germany lies the hidden treasure of the Jache family? happy
Ok, I quickly can run (I guess it's the only test I should perform: load'n'read on screen) the .d64 on a 1541-II.
I don't remember where my 51/71/81 are, but if I'll get free from some situations, I could find time enough to test those too.

Posted By

Degauss
on 2011-09-14
16:23:37
 Re: Please help testing: New loader release imminent

OT: Nah, i wouldn't say we were hardcore-collectors. I cannot even remember buying any of this stuff at all. I think i bought one +4 and one amiga500+. the rest just "arrived" - or got dumped - at my place in short intervals from the 1990s on. So we didn't collect, but we never gave that stuff away. When i put these stuffs into boxes i had to think "was that really necessery to keep ALL of this?" wink

Posted By

Chronos
on 2011-09-14
16:24:14
 Re: Please help testing: New loader release imminent

ok, i have 1541-2, 1541-1 and a 1541-2 clone.. also i need a few days to dig it out from my boxes but you can count on us grin

Posted By

Luca
on 2011-09-14
16:27:37
 Re: Please help testing: New loader release imminent

The .d64 is an extended one. Must I respect and preserve this format in copy phase, or can I simply copy the files into a normal 664 blocks one?

Posted By

Degauss
on 2011-09-14
17:29:20
 Re: Please help testing: New loader release imminent

From my experience with cbm4win/opencbm: Disks need to be formatted in extended format first, otherwise the d64-transfer won't work.

Posted By

Luca
on 2011-09-14
17:53:32
 Re: Please help testing: New loader release imminent

Oh... :o
Should I use 759 Blocks on the real iron, as an option?

Posted By

Degauss
on 2011-09-14
19:19:18
 Re: Please help testing: New loader release imminent

umm, no. if i remember right i just did this:

cbmformat -x -c 8 bla,01

then

d64copy bla.d64,8

Posted By

Luca
on 2011-09-15
02:31:13
 Re: Please help testing: New loader release imminent

Degauss: eh, but I can't use cbm4win/opencbm, that's why I'm opting for a real iron solution.

Posted By

Krill
on 2011-09-15
03:59:19
 Re: Please help testing: New loader release imminent

Chronos: Thanks!

Degauss: Thanks, too! One specimen of every model would be nice. If you cannot retrieve all of them, old 1541 is preferred, as at least one known (but yet unfound) bug exists with certain 1541 devices. And yes, i will put a loader build service on the server some day, but so far the priorities were strongly on this port, the features and the bugfixes, which is also why there is no documentation to speak of yet.

Gaia: 1551 is supported through the KERNAL fallback feature. The loader simply uses KERNAL calls for all drives it doesn't have custom drive code for. This means that loading isn't faster than without the loader, but at least it works. If enough people request native 1551 support, i might add it some day - could re-use some code for parallel 1541 connections, too. happy

Luca: Yes, extended format and transfer needed. The upper extra tracks have more read restrictions than the normal ones, so this must be tested, too. Thanks for testing!

Posted By

Luca
on 2011-09-15
13:39:49
 Re: Please help testing: New loader release imminent

I fought a whole afternoon, but now we can comment the very first results happy

Before all, let's say that Krill opened the upper and lower borders in a test stuff, and I didn't do it even in a normal release, there's too much genie in this world happy

First, I checked the .d64 on 1541 and 1541II: the test loader loads continuously, if I've seen good the end should be when the border INC, then the border color itself is increased by +1 and the cycle restart. I did several cycles that way, and the loader has run with no defect for both the drives.

After that, I checked the .d64 on a 1551: the test loader appears to stuck in a certain point (PAL #08,GENERIC $0A3E 769 kB/s GETCHUNK/CMP), printing a message on the first line (@$15ED/$FC($FA,$01):DEVICE NOT PRESENT.)

It's 1571 time. At first, it used to stuck in the initial installing phase, then I changed the IEEE hole and it worked. I checked the .d71, it runs like hell till the 280kB/s GETCHUNK/CMP, where it stucks and at a certain point with the anime gal part hires and part MC, and occasionally it crashes after some time.
I also checked the .d64 on 1571 to see what happens, and it stucks in the same point but at 518kB/s, and acts the same way.

Posted By

Krill
on 2011-09-16
03:54:06
 Re: Please help testing: New loader release imminent

Luca: Thanks for testing!

I could reproduce the 1551 issue, it simply fails to load stuff from the extra tracks, which is expected without custom drive code. The error message is wrong, though. I will simply add code to the test portion (not the loader itself) to ignore illegal track or sector errors when using KERNAL fallback.

I'm a bit puzzled about the 1571 issue, though. Does the loader correctly detect it (CBM1571 instead of generic in the lower border)? The speed figures are strange, should it be .280 kB/s instead of 280 kB/s? How does it crash exactly? The test program loads 3 files for GETCHUNK/CMP: Sheba the cat-girl, then the misspelled Panta Rei logo, then both in one file. Does it freeze/crash on the first or the third file? And with chainging the IEEE hole, you mean use the other serial port on the drive? Which one is the one that works?

Again, thanks for the help!

Posted By

Krill
on 2011-09-16
04:15:46
 Re: Please help testing: New loader release imminent

Oh, and only one drive can be connected on the serial bus with the currently-used protocol. If there is more than one drive on the serial bus, the loader will resort to KERNAL fallback.

Posted By

Luca
on 2011-09-16
05:12:12
 Re: Please help testing: New loader release imminent

Yes I avoided cascade linked drives wink One by one!

The loader perfectly recognizes the 1571, naming it on screen. Beg your pardon for the speed reading, probably I missed to read the dot in the speed values because it has been hidden in the corner of the extended lower border.
I see the whole test normally runs with the upper screen slice in hires and the rest in multicolour, hence that's normal and don't need to be reported. I remember it stucks at first file of GETCHUNK/CMP, I'll check again in the afternoon to be absolutely sure.

Posted By

Krill
on 2011-09-16
07:34:10
 Re: Please help testing: New loader release imminent

Okay, so 1571 seems to be detected alright, and loads fine at good speed and without IRQ disturbance, just that it freezes/crashes at some point, right?

Posted By

Krill
on 2011-09-19
14:51:08
 Re: Please help testing: New loader release imminent

New build here: http://krill.earthtomoon.net/loadertest-v118.zip.

This fixes a known issue with 1541-family drives, and also some "adventurous" optimization with 1571.

Luca: Maybe this one works better with your 1571. If not, maybe you could help me with a remote debugging session? happy

The 1551 (and generally "generic" drives) issue hasn't been tackled yet.

Posted By

Luca
on 2011-09-23
09:13:09
 Re: Please help testing: New loader release imminent

I'm not able to run the V118
I copied the .d71 disk image with a 1571, then I tried to install the test prg on my naked plus/4+1571: this time it stucks at the install screen and does not go further, unregarding the IEEE port I use

Uselessy I checked the .d71 with a 1541II: the loader installs but, as expected, after the screen colour setting, no gfx loads since the very beginning.

Posted By

Krill
on 2011-09-23
14:27:57
 Re: Please help testing: New loader release imminent

Luca: This sounds strange, i haven't changed anything in the 1571 code that would explain this new behaviour. Can you read all files using plain KERNAL load from BASIC? Nothing is stored on the extra tracks in the .d71, so everything should load alright without the loader (after switching to 2-sided mode, though - open15,8,15,"u0>m1":close15).

The 1541 can't read the second side, of course, so the 1571-format disk can't be read properly there, which you expected correctly.. happy

You're the only one with these problems so far, so would you be in for some remote debugging? If yes, just send me a mail to <mynick>@<mygroup>.de. I'll then just send you images to test to first home in on the bugs and then dispatch them.

Posted By

Chronos
on 2011-09-24
10:03:32
 Re: Please help testing: New loader release imminent

hi!

you can see it for yourself

http://www.youtube.com/watch?v=6HWEAOb-qK0

only tested with the blue chip drive, looks like everything is fine.. thanks ingo for the hint to make an extended disk

Posted By

Krill
on 2011-09-25
20:31:20
 Re: Please help testing: New loader release imminent

Chronos: Thanks for the video. Didn't know the border flickers on the real thing. Will fix this.

Does anyone have similar 1571 problems as Luca?

Posted By

siz
on 2011-09-26
03:40:37
 Re: Please help testing: New loader release imminent

I'm trying to make half an hour free to test on real iron. I've a plus/4 with a 1541-II, a 1551, a 1571 and a 1581 at hand but had no time to turn on them.
I wanted to make the test yesterday but couldn't make it.

Posted By

Krill
on 2011-09-26
03:53:35
 Re: Please help testing: New loader release imminentK

siz: No worries, there is time happy And i will have to fix that 1551 bit first anyways. Didn't do that yet because it seems like an easy fix, and so i cared for other problems in the meantime. I will fix that asap and upload the next test image.

Posted By

siz
on 2012-02-09
12:41:34
 Re: Please help testing: New loader release imminent

Today I'm working at home and took a bit of worktime to test the loader. grin

The test hardware:
- plus/4 (more or less standard)
- 1541-II (SpeedDOS 2.7 switchable between 35 and 40 tracks)
- 1551
- 1571 (old, 3.0 ROM)
- 1581

(And my laptop with zoomfloppy and an USB screen grabber as monitor)

The 1541-II is detected as 1541C, parallel mode is not detected (I don't know if it's implemented) but serial mode works fine.
1551 works in generic mode with the same error Luca posted. (Took a screenshot http://siz.hu/external/krill-1551.png )
The 1571 is recognized but does not start (screenshot : http://siz.hu/external/krill-1571.png )
Finally I had some tough time to copy the image to the 1581. (I've never done this before, waiting for tips on how to do this well). Finally I've extracted the files from the D81 image with total commander and the dircbm plugin and copied the files one-by-one with the dos command
for %i in (a*.*) ; do cbmcopy -w -o "%~ni" 8 "%i"
(the files were in a subdirectory called "a")
After the copy was finally done I could test it: works flawlessly.

Posted By

Luca
on 2011-09-27
05:05:34
 Re: Please help testing: New loader release imminent

About 1571: yes, I had the same with v118, but I got no error message at the top.

Posted By

Krill
on 2011-09-27
17:04:39
 Re: Please help testing: New loader release imminent

http://krill.earthtomoon.net/loadertest-v120.zip fixes the 1551/generic drive issue. Files on the extended tracks are simply skipped.

Luca, siz: Can you confirm that this works with 1551?

The 1571 issue is still not solved, will investigate further.

siz: Thanks for testing! Will also look at the 1541-II/SpeedDOS issue. Can you maybe send me a ROM dump?

And yes, parallel isn't supported yet, but might be in some future version, both for native 1551 support and 1541/71 with parallel extension.

Posted By

Krill
on 2011-10-08
13:19:47
 Re: Please help testing: New loader release imminent

http://krill.earthtomoon.net/loadertest-v126.zip fixes the SpeedDOS issue.

This leaves the Plus/4+1571 as the last remaining known bug.

Anyone willing to help me debug that?

Posted By

MMS
on 2011-10-09
15:20:05
 Re: Please help testing: New loader release imminent

Hi, I have both, but no tool to transfer the file to real HW and floppy from PC...
(to be honest I got my 1571 almost 2 weeks ago (together with 6 Plussy happy ), but no time yet to fire up the iron. It did well the start-up sequence without a computer, so 1571 probably OK, but not sure yet)

Posted By

siz
on 2011-10-10
04:17:44
 Re: Please help testing: New loader release imminent

Hi Krill!

As I wrote in my e-mail, I'm willing to help in debugging. happy

Posted By

Krill
on 2011-10-19
17:59:33
 Re: Please help testing: New loader release imminent

I found something which might resolve the 1571 issue.

http://krill.earthtomoon.net/loadertest-v132.zip

siz, Luca: Can you see if this works now? Thanks!

Posted By

siz
on 2011-10-20
01:20:13
 Re: Please help testing: New loader release imminent

I haven't had the time to test it on real hardware yet but gave it a try in VICE. The old version you sent me (126M) produces the same result in VICE as on my plus/4 and this new one does not work either. At starting POLLBLOCK without other messages it just flashes border in brown(?). I'm not sure about the color in VICE.
I'll give it a try on real hardware in the evening.

Posted By

Krill
on 2011-10-20
02:14:15
 Re: Please help testing: New loader release imminent

The assumed problem is with the drive-side watchdog IRQs. These are supposed to time out on single-sided communication break-ups, meaning they shall reset the drive if the computer is takingmuch too long to respond, which happens when it is reset or crashes unexpectedly.

I used a wrong timer-mode, which resulted in a mere 256-cycle time-out on real hardware while all emulators seem to run the time-out at the full 65536 cycles.

The result was that the drive decided to reset while the computer was in the midst of communication, and depending on where in the protocol the drive reset, the computer would do one of various things, including loading faulty data, crashing, freezing, or returning a device not present error.

Now, VICE/xplus4 suffers spurious VIA timer interrupts, which means the watchdog IRQ is triggered for no reason, producing the same results.

The watchdog can be disabled at the loader's compile time.

As it's still perfectly possible that real hardware will work just fine, i'm eagerly awaiting your test results... happy

Posted By

Luca
on 2011-11-10
12:05:41
 Re: Please help testing: New loader release imminent

Tested right now.
It stucks in the installing loader screen with the counter locked at 00:00:00

Posted By

crock
on 2011-11-16
07:49:24
 Re: Please help testing: New loader release imminent

Are you still soliciting feedback on this? I have a +4/1541-II combination that its failing on.

rob

Posted By

KiCHY
on 2012-01-13
11:02:57
 Re: Please help testing: New loader release imminent

Hi Krill,

Is there any progress with this loader?

[SOMEHOW OFF]
What are the requirements of the loader after the driver code is uploaded? Memory requirements? Is it relocatable to anywhere? I plan to use your loader in my next project. Where can I find these infos (or current source)?
[/SOMEHOW OFF]

Posted By

siz
on 2012-01-14
03:33:14
 Re: Please help testing: New loader release imminent

There were some problems with 1571 loading for Luca and me. I tested a bunch of test revisions for Krill, it seems that finally the problem is solved. He wrote me that after some cleanup he will make a new full build. He will release the loader if that works well.

Posted By

bubis
on 2012-04-19
11:14:48
 Re: Please help testing: New loader release imminent

Wow, Krill loader on C=+/4, supporting 1551 as well?
Are you telling me that we have a proper irq loader on plus/4 finally? happy
Is this much fasted on 1551 compared to 1541?

Posted By

siz
on 2012-04-19
13:19:44
 Re: Please help testing: New loader release imminent

It was a long time ago when I tested it for the last time but as far as I can remember it uses KERNAL fallback on the 1551.

Posted By

bubis
on 2012-04-20
07:36:51
 Re: Please help testing: New loader release imminent

Checked it on YAPE, and it seems really slow, so you must be right.
The 1541 performance is very impressive though, and most ppl don't have 1551 anyway. happy

Posted By

MMS
on 2012-04-20
10:58:01
 Re: Please help testing: New loader release imminent

As i learned, 1551 housing is just too britte, and tends to break during transport. Some people even do not ship them any more.
So there will a limited and decreasing number of collectors of 1551 (it was never too famous or widely available, as if I remember, my parents also checked in Vienna more than a dozen Computer shops, and finally gave up and came back with an 1541/II).

If the loader covers 1541, 1541/II, 1571 (in 1541 mode) and falls back to the original 1551 mode if detected (1551 is anyhow ~4x faster), it will cover 80% people speed increase requirement, and the 1551 owners remain as happy as they were. 1581 I do not cover here, I do not have, so I do not care

(what a selfish guy, haha! happy )

Posted By

siz
on 2012-04-20
11:43:18
 Re: Please help testing: New loader release imminent

The loader supports
* 1541 (and clones) with serial interface
* 1571 (native, 2MHz mode), serial
* 1581 (native, 2MHz mode), serial
Parallel cables are not used. 1551 and other unidentified drives are used in kernal fallback mode.

Posted By

bubis
on 2012-05-02
09:16:12
 Re: Please help testing: New loader release imminent

Apologies for my stupid questions in advance! I have very limited time to look into these things, so I just want to leverage on your experience. wink

1) It seems to me that you need to refer to the file you desire to load by its file name. Does this mean that the dir records are checked on the disk every time before loading a file?

2) Is there any way that allows dir art with this loader? Like, please load next file , or n-th file, or something like that?

Posted By

siz
on 2012-05-02
13:19:09
 Re: Please help testing: New loader release imminent

Hi bubis!

If you meant me then I've to disappoint you: I've never seen the loader itself and I don't know anything about it. Krill sent me a couple builds of the loader with a test application and I've tried those.

I don't know anything more about it.

Posted By

bubis
on 2012-05-03
05:10:58
 Re: Please help testing: New loader release imminent

Thanks, I will ask Krill.

Posted By

Krill
on 2012-07-12
07:46:05
 Re: Please help testing: New loader release imminent

Hi all,

sorry for being absent for a while.

In the meanwhile, I'm close to finishing implementing a new, faster protocol which allows for multiple drives on the bus, and this hopefully without the 1571 issues experienced with the 2bit+ATN protocol.

I plan to release the loader within the coming week(s), and will prepare a new Plus4 test build soon.

crock: Can you check if your 1541-II setup still fails with http://krill.earthtomoon.net/loader-v136.zip? Keep in mind only one drive is allowed on the bus in this version. Thanks!

KiCHY: In the minimal Plus/4 configuration (loading raw, no built-in decompression), the loader uses slightly more than $0100 bytes once installed. It comes as source (see link above), so yes, you can relocate it.

bubis: Native 1551 support is planned for the future. I really want to see that drive's blazing speed unleashed in an IRQ-loading setup some day. I expect it to be about twice as fast as 1541, so roughly 7 kB/s.
The loader uses a dir buffer with 40-something entries (sliding window ring buffer, file names hashed to two bytes), so it rarely needs to access the dir track.
Dir art is supported by setting the maximum file name length to e.g. 2, then loading "01*" and similar - all extra characters in the dir entries are ignored and can be used for dir art. Alternatively, you can hide a second directory on any track and make the loader use it, or load by track/sector, so you can do whatever you want in the normal directory (both alternatives break 1551 compatibiliy, as they contradict with KERNAL fallback). An actual "load next file" call is planned for some remote day in the future.

Thanks and greetings!

Posted By

bubis
on 2012-10-10
13:11:34
 Re: Please help testing: New loader release imminent

@Kill: Dir buffer, lookup by hashcode, max file name length, all cool features!
Thanks!

Posted By

Spektro
on 2012-10-12
09:22:42
 Re: Please help testing: New loader release imminent

Hi, I tried versions 114 and 136 with my Plus/4 and 1541-III and they both stuck on the installing loader screen. The counter stays at 00:00:00 and there is no activity on the 1541-III.

Posted By

MMS
on 2012-10-13
15:31:53
 Re: Please help testing: New loader release imminent

hello Krill, as currently i have no data transfer tool, i could not test it on my 1541, 1541/Ii or 1571. On the other hand i went through on the source code (seems you had success with Cs65), but cannot really get how you could manage the small size (without compressor). As this loader may turn out to be a final multitalent, do you think you may help out the less talented guys with a version may work from BASIC, after loading the code with DATA to low mem, and call with eg. SYSxxx,"02.prg" . It would be a great help to easily integrate it and make stuffS made in Basic more professional (faster)? Thx in advance MMS

Posted By

Krill
on 2012-10-15
03:06:58
 Re: Please help testing: New loader release imminent

Hi MMS,

i am indeed currently working on a stand-alone wrapper for the loader, meaning you run an install program from BASIC and it returns back to the prompt, with the LOAD command now using the loader. Would this work for your purposes?

Posted By

MMS
on 2012-10-15
14:58:02
 Re: Please help testing: New loader release imminent

hi, thx for quick response! So it looks like DLOAD 1541 turbo loader. Sounds great. What will happen if i call it from a running BASIC PRG? May the prg will restart, or deleted? Is there a way to load the code to the proper place, start with Sys, then after that all Load will be blasing fast? You know, my aim is to load and start my prg with load"*",8,1 and without any extra user activity it should just work fast. Most of tester checks it in emu. If the turboloader is the first in D64,nothing will happen. If mine, and cannot call loader from running basic code, then the first experience will be: very slow loading. Compressor also sound interesting, as i want to load 10kb picture files, and we have no standard compressed file format like GIF.

Posted By

Luca
on 2013-05-07
16:08:40
 Re: Please help testing: New loader release imminent

Krill's Loader has reached today a stable release. Don't miss it.

Posted By

siz
on 2013-05-07
16:17:26
 Re: Please help testing: New loader release imminent

That's good news! I'll check it out! happy

Posted By

Litwr
on 2013-05-08
13:51:24
 Re: Please help testing: New loader release imminent

Fascinating! It may give 8-10 kB/s at 1551...

Posted By

MMS
on 2013-05-08
15:11:23
 Re: Please help testing: New loader release imminent

wow, 2-3 sec loading speed on an uncompressed pic with 1541. Unsure about 1551 support. I think there is no fastloader that automatically supports both drives...

Posted By

siz
on 2013-05-08
15:28:26
 Re: Please help testing: New loader release imminent

According to the release notes the 1551 support is only planned for future releases.

Posted By

Litwr
on 2013-05-08
15:52:57
 Re: Please help testing: New loader release imminent

This speed gives possibility to see films with the small raster (e.g., bw 160x50 or multicolor 80x50)! 1 kB of decompressed data per 1/24 sec may make the mentioned raster. So the film's length per one 170 kB disk without sound may achieve about 6 min. The best video should use interlace. I hope one day to see "Commodore +4 news channel"! happy

Posted By

Luca
on 2013-05-13
07:37:53
 Re: Please help testing: New loader release imminent

I tried and tried anything, I installed the most obscure tools as requested, I did any move for days, and now it's time to stop all this and uninstall all the shit. And accept that I haven't the knowledge to afford this compiling

EDIT: still fighting...

Posted By

Luca
on 2013-05-13
08:39:19
 Re: Please help testing: New loader release imminent

Krill gave me some good hints and I've been able to compile it, finally happy



Back to topReply to this topic


Copyright © Plus/4 World Team, 2001-2024