I’ll post the journey I’ve had with Amithlon, from hardware I’ve used, installing issues and tweaks and eventually to compiling the kernel based on Milan’s “Kernel 4” tree. Make sure to visit http://amithlon.snkbitten.com/ for downloads of patches, kernels and miscellaneous files. Lots of internet searches and hardware experimenting to make your journey easier 🙂 – SnkBitten
Reading through the posts I noticed my system had changed slightly since the last time I posted, so a little updating…
Gigabyte Gigabyte GA-P35-DS3L motherboard
Intel Core2Duo @ 4.00 Ghz (e8600 @ 3.33 GHz)
(2) 512 MB ram modules (1 GB total)
64 MB nVidia 5200 PCI graphics card
32 MB Matrox G550 PCIe graphics card
Catweasel MKIV PCI card
Soundblaster Live 5.1 PCI
Realtek 8169 10/100/1000 Network card (PCI) Intel Pro-1000 10/100/1000 Desktop Network card (PCIe)
Motherboard 10/100/1000 Realtek RTL8111 chipset
8x DVD-RW(-/+) Slot loading drive
20 GB SATA dos boot drive (freedos)
256 GB SATA AmigaOS drive
Multi-card USB reader
3.5″ PC Floppy (Catweasel allows for r/w of Amiga DD/HD and PC DD/HD disks)
Nothing spectacular but for those of you with the common RTL-8111 motherboard NIC (10/100/1000) this is a very nice addition.
Doing some linux investigating we’ve found a 2.4 kernel module for the RTL8111 series chips, common on motherboards (10/100/1000 nic).
I have a new module file for the following network chips used on motherboards, it doesn’t build with the kernel so had to be manipulated a little to get it to ‘make’ for kernel4.
It’s based on version 1.07, no changes to the driver just modified the make file.
Download this for your Amithlon system.
Extract the files to ram: then copy the r1000.o file to your devs: pcidrivers/net/ folder and copy the other three files to your S: directory.
It will overwrite your pci_modules, vendors.txt and vendors_pci.txt files in S: so you may want to back those up. The new ones were updated and should give more useful info as well when doing a pcilist or pciscan command (much less “unknown device”).
I can remove the Intel Pro 1000 Nic from my system now and use the motherboard built-in Nic.
I replaced the kickstart in Amithlon with a stock A500/A2000 3.1 kickstart. The system booted just fine and runs with no issues that I’ve noticed so far. It looks like only exec is different between the two. I’m going to experiment with how much the shape.rom can be patched without creating any issues.
Okay, this is a much easier process to setup a compiling environment in Virtualbox to build the Amithlon kernel based off Milan’s Kernel 4 tree.
I’m working on a new guide for compiling the kernel using an older Ubuntu distro, though it’s newer than the previous guide using a really old Centos distro.
It seems to be a quicker and easier setup as well and I hope to have it posted/hosted over the weekend.
This doesn’t add anything new though it does make the process easier and hopefully help some start rolling their own kernels.
I feel the installation guides have reached a point of being complete, at least as much as I’ve been able to make them. So I exported them as PDFs and archived them together and posted the archive to Aminet.
I’ve started a new ‘Miscellaneous Notes’ of items that I’ve found useful or information I’ve found that doesn’t really involve the installation process. Once that has a bit more info in it, I’ll host it as a live Google Doc like I’ve been doing with the installation guides.
Honestly there isn’t a lot I have going on with Amithlon (or Amiga) at this time. I’m getting more involved in Lightwave 2018 (and 2015.3) these days but I’m hoping to do some more hardware testing this year if funds permit. Still wanting to build up an i7 based system, just need to find (and purchase) the right motherboard/CPU and use the best combination of PCI and PCIe cards I’ve tested.
Doing some more reading/researching and I found some nice information on the init string to launch Amithlon and including certain settings from the ‘startup-sequence’ in AmigaOS in the init string to potentially eliminate an additional OS restart.
First off, the init string….
Don’t change this one, it is needed for Amithlon to start
Set this to 4 (instead of 0), and the boing ball will no longer bounce, but rather sit in the top left corner, and you will get kernel and emulator messages during startup. Do this if things fail to start!
Leave this as it is. Amithlon needs it like this (kernel 4 uses ram1 instead)
Leave this as it is for ‘bigird.gz’ or use 2310 if using ‘smallird.gz’
Don”t reduce this number. You might want to increase it to give Amithlon more memory for disk caching (only read-caching, no write caching, so it’s safe!).
Each page stands for 4k; If you increase the number by 1024, you are adding 4MB of disk cache (but take 4MB away from the Amiga). Reducing the number below the values given will result in weird and hard to track hangs!
Only used if you are using the VESA driver for graphics support. In that case, this decides the one (and only) mode you will have available. See the following table for the meaning of the numbers. Installing amithlonupdate131 will give you screenmode choices in AmigaOS for additional VESA modes.
Resolution 640×480 800×600 1024×768 1280×1024
8bit 769 771 773 775
16bit 785 788 791 794
24bit 786 789 792 795
So an 8bit 800×600 screenmode would be vga=771
If you are using the setconfig options in your startup sequence (e.g. to set cachesize, or monitor frequencies), you can also put those onto the kernel command line. The advantage is that those options are available before AmigaOS starts up, and will thus not require a reboot to come into effect. To set, for example, the cachesize to 16384kB from the kernel command line, simply add the option “cachesize=16384” after the “leavepages=xxxx” entry. A word of warning, though — the various boot loaders all have limits on how long the kernel command line can be; So if it looks like your settings didn’t make it into Amithlon, check with “dumpconfig”… They might just have been truncated.
My init string for loadlin.exe is named tsmall and contains the following:
nvitel.gz init=/linuxrc console_level=0 root=/dev/ram1 initrd=smallird.gz ramdisk_size=2310 leavepages=7800 cachesize=65536 vesa_defaults=0 hsyncmin=24 hsyncmax=82 vsyncmin=48 vsyncmax=76 clockmax=300 asynchronous_io=1
So my autoexec.bat file reads
nvitel.gz is my compiled kernel for my nVidia cards and the IntelPro Nic. The rest set all the config options and the setconfig settings that were previously in my s:startup-sequence file. Since I can comment those out of the startup-sequence now, on boot the “setconfig reboot if changed” no longer forces a reboot. Running dumpconfig confirmed all my settings were still present (after a complete power cycle and restart).
Nothing really going on currently for me and Amithlon. I’d like to continue testing new(er) hardware, perhaps an early i7 motherboard (ICH10R chipset) and maybe a PCIe Soundblaster type card….. However that will all need to wait until I have some spare funds.
I’ve been busy lately playing with Lightwave 3d 2015 on my Win10 PC, so much that I have barely touched anything AmigaOS related (A4000T, WinUAE, Amithlon….). I was a Lightwave fan back in the ’90s on my A2000 and A4000D Toaster setups and I was curious to see how Lightwave had evolved over the years.
Make sure to leave any comments or questions on any previous posts as I always try to answer any questions and assist. Or add posts to the Google+ Group, maybe some of the other members have something to add.
A few of the animations I’ve done in Lightwave 3d 2015 using only Lightwave 3d features. All models downloaded from online sources. Trench run was an .lws (scene file) with X-Wing model added and time scale doubled.
I swapped out the Matrox G550 PCIe card, going back to the nVidia FX5200 PCI card. Main reason would be it is significantly faster and I’ve been spoiled with starting with nVidia cards when I first began working with Amithlon. The nVidia cards seem to have the highest level of hardware accelerated features, besting the Matrox in SysSpeed’s graphics test sometimes 12x faster. Typically it’s 1.x to 2.x faster overall, a few 3x and higher. The Matrox did win a few, I would assume from running on the PCIe bus. The second reason would be Matrox cards are limited to 4mb for Picasso96 while the nVidia card presents 16 mb. This allows higher resolution/higher color depth screen modes.
I am generating SysSpeed “modules” for the cards I’m testing, running 800×600 in 8/16/32 bit color screenmodes. I’ll post the modules soon, adding my A4000T with CyberVision 64/3D as a comparison base.
I’ve used nVidia FX5200 64mb, FX5200 128mb, FX5200 256mb and a FX5500 256mb cards, the 256mb FX5200 showing the fastest results. The 128mb card failed to initialize the framebuffer once loadlin was run (open framebuffer…..no such device) and surprisingly the FX5500 was much slower than the FX5200.