Pegasos “Non-April” – Does it run or not?
The Pegasos computer without the famous april2 fix is a machine that everyone avoids to buy and it is obvious in the informal online “exchange” market where a Pegasos 2 G4 or G3 was being sold on an average of 450 up to 600 euros, a Pegasos 1 april2, was sold between between 200 and 300 euros while the poor Non-April was being sold at the max 150 euros or even less than 100 euro! And all this when in absolute numbers, the computing power of a Pegasos non-april compared to a Pegasos 2 G3 is essentially the same! For all that though, there is a guilty and this is the bugs and general issues left on motherboard by manufacturer company, Mai. Problems serious but not completely insurmountable. But what exactly is happening
Throwback
Because the intention of this article is not throwback of Pegasos, will be limited to the early stages of creating machine wherever occurring problems are under discussion.
Problems occurred in the first boards manufactured, on the distant 2002. In the first months of 2002, Thendic-France (later Genesi) receives the first motherboards from bplan which were meant for developers. The Thendic discovered hardware problems mainly related to the manufacturing chips ArticiaS from Mai. Nevertheless the machine was sent to developers and test-boards were sent in the market. But because of the problems that can not be solved, Thendic baptized the machine as Betatester, announced a program for foreign betatesting and bplan begins manufacturing the first 150 motherboards for this reason. All this until September 2002, from where bplan met with representatives of Mai (who manufacturesdthe defective chips), they recognized the problems and agreed to produce a hardware patch that will fix all problems. The remaining of 2002 was dedicated to the manufacture of the patch. In December 2002, appeared the famous April Fix and Thendic started exchanging old motherboards, with 200 new ones in April but immediately appears another problem and Articia started manufacturing the April2 fix!. The name april, rumored to have originated from the phrase «there is no Mai without April” which is a pun of the word May (May). Mai indicated that the Articia chip has no physical value without the April fix.
In 2003 Thendic disappeared as the company merged with bplan and formed Genesi. In February the april2 is ready and quoted by Genesi an exchange program for the betatester motherboards with April2 Fix. All these problems, delays and inconvenience, leading Genesi eventually abandoning the production of the Pegasos and launched the project, Pegasos II with promissory that once the new machine is available, holders of the old will be able to upgrade to a charge of only 200 euros. And that’s what happened.
But what about Pegasos Non-April?
That’s a good question. There were few Pegasos Non-April produced and there was the exchange program and patching for all boards of developers. Nevertheless, there were many who did not exercised their option for various reasons, perhaps indifference, leading to the still today, existance in the market of some used Pegasos Non-april.
How to recognize them if they fall into my hands;
To declare a non-april Pegasos, one must visually observe the board. Specifically in the centre Northbridge chip called Articia S are obvious differences between Pegasos and April2 or April1. Diversification has been brought about in two more chips on the 2 of the 4 sides of Articia in a “ear” shape. The chips do not exist in the version given to Beta Testers.
What problems has the Non-April?
Unfortunately officially we have no idea what problems the machine experienced as by the 2002 Thendic, when it began sending boards to the betatesters, forced them to sign an NDA or non-disclosure agreement which protected the results and committed the testers fto reveal to any third party, that is, except Thendic, the problems facing the motherboard. Nevertheless, we know of course that the main problem was the chip of Pegasos, Articia type Northbridge. The type of event the problem was basically the degeneration or corruption of data which manifested by the functions controlled by the northbridge, ie memory, the processor bus and the PCI / AGP port. The subject of course was that Pegasos had a non functioning DMA (Direct Memory Access) which roughly copies and reads data to and from the RAM of the machine, to and from a device which uses DMA for example a hard disk. Using the DMA, normally released from the central processor and data transfer functions efficiently undertake other tasks concurrently when data transfer becomes faster. In Pegasos this function often corrupted files and especially when requested to transfer large files. This problem resulted in the Pegasos to be a very unstable and problematic machine.
Solutions??
The main solution is known. AprilFix. But a current holder of Pegasos non-april can not do anything because the machine is no longer being produced or patched. The fix is only on a hardware level so there is no substance whatever of software on any operating system or Linux-MorphOs either to correct the problem or to bypass it. But even with these data a Pegasos non-april can be fully functional and stable. It would take some small sacrifices for this purpose. Specifically:
Because the main problem is the data transfer of DMA hard drive to and from RAM, the only solution is not using DMA! The first problem is that the analog BIOS machine (open firmware) does not support disabling the DMA mode. So the DMA should be bypassed. The solution proposed is simply not to use hard drives with DMA support. But to do this given that all modern hard drives support DMA mode should be used old technology and specific capacity of 4GB and under whIch HD’s can be operated in PIO mode.
In this way you might avoid data corruption and the system could be somewhat stable but you will lose the capacity to load the main processor by reducing the performance and of course will substantially reduce switching speeds, data speeds will be 5 to 6 Mb / s . This speed however is greater than the speed data switching a hard drive on Efika which is known to be in the same format as the Pegasos. Moreover such speed levels are not unknown to the Amiga community as often found in classic Amiga 1200 machines etc. which often use hard of this size due to limitations in the ide controller.
As an extra measure of stability of the machine is the correct choice of RAM. The component will provide additional assistance to the system is the use of non-DDR main memory and main memory and Registered ECC, ie with Error Correction Code which function finds and corrects errors in the data avoiding corruption-degeneration.
Alternatively one could use a CF2IDE adaptor for 3.5 “device with one or more CF cards which until recently were not using DMA mode from the manufacturer specifications. The latest generation in many cases are DMA enabled (with however many complaints from consumers that DMA transfers are not proper). Nevertheless with this solution, the user has the opportunity to substantially obtain new cards to use as hard drives and not expended in finding old hard drives and used, which after so many years of use will certainly be in a questionable condition and would often fail.
Furthermore it should be noted that some hard disk manufacturers provide tools to control DMA transfers or disabling, such a tool might be been tested on a Pegasos in the future to see if it performs what it promises!
By combining the above solutions (according to common sense) a relatively stable system than one would expect for Pegasos non-april, can be achieved, with the possibility to install and use the new operating Morphos without major problems. However, in general, the community of Morphos does not propose the use of MOS 2 and above for such machines and is widely acknowledged that a Pegasos Non-April clearly works better with the previous version of Morphos 1.4.5.
Published in Greek by “Npanag” and translated (probably poorly… 😦 ) on 2009 on Amigaplanet.gr
Amigaplanet.gr Blog 