Welcome everyone to this week’s TWiP. I’m going to start this week with some serious stuff before I get into the fun stuff. Please excuse the lack of my own screenshots this week, my Amiga 2000 is awaiting a part for the PSU before I can turn it back on again.
A Word on Community
There are quite a few accelerator projects out there right now, each with a thriving community. Not just PiStorm but Vampire, Buffee, and Terrible Fire to name a few. Each one of these accelerators focuses on different use cases and they all have various crossovers. Being a project leader for public projects like these is very difficult, some decisions made by those leaders may not seem right without the view and context they have.
We want to make the PiStorm community a safe place and we urge people not use our community to say bad things about other projects, leaders or communities. We also urge people not to do this on other communities either. You may not agree with a project’s goals, costs, or the way it is being run, but attacking them in a bad way will could damage to several Amiga communities, harming things for everyone.
Know Your PiStorm
In the beginning there was a PiStorm revision A. This had a minor design flaw in it, but a workaround was possible. Soon after revision B came into existence, this fixed the design flaw and added support for more versions of the CPLD. We have seen a few new revision A variants suddenly being made and what is more concerning is the workaround for the known problem isn’t being applied, which is causing them to not work. I happen to now have both boards so I figured I would show the difference and why it is important (ignore the flux on my rev B 🙂).
On the top side the most obvious difference is revision B has an extra place for a voltage regulator. This is to support the “GT” variant of the MAX II CPLD (EPM240GT and EPM570GT) and potentially the MAX V CPLD in future (the MAX V does not work right now). Both of these have different voltage requirements to the regular CPLDs used.
The bottom here is the key way to tell and where the problem we see lies. The most obvious different here is the image on the silk layer on revision A which is not on revision B. But, if you have revision A, your PiStorm must have pins 1 and 17 removed from the GPIO header. These are the 3.3V lines from the Pi and they were incorrectly hooked up on the revision A. If these pins are not removed then the Pi will not function and it could cause damage to either the Pi or the PiStorm board. Alternatively they can be cut from the Pi’s own GPIO header.
Revision B does not have this problem as they were disconnected in that design.
We have seen several instances recently from more than one vendor of revision A being sold and the GPIO pins not cut. We highly recommend that:
- You stop making revision A boards, revision B’s Gerbers are much easier to obtain anyway.
- If you must make revision A, please pull pins 1 and 17 of the GPIO socket before soldering, this is very easy to do and saves a lot of hassle later.
Contrary to something I’ve seen a few people say, both boards can take both the EPM240T and EPM570T MAX II chips. The first PiStorms were constructed with the EPM570T but we have seen both used on both boards, both of my boards have an EPM240T for example. Your revision does not dictate which of the two you have.
The Great Merge of June
For those of you who follow this blog, you’ll know that almost everything I talk about here is in the “beeanyew” tree under the branch “wip-crap”. Whereas the “main” branch is in the “captain-amygdala” tree. The “main” branch has been getting increasingly behind all the new cool stuff that is in “wip-crap” so this week a snapshot of the commits were merged up. The “main” branch now has all the new features I have been talking about up until last week.
To update your “main” branch please see the section on “Getting the Changes” in last week’s blog post. Follow on from the paragraph starting “We recommend”.
_Bnu mentions special thanks to:
- ChaosPif for helping out with the A314 stuff and writing the excellent documentation/setup instructions for it.
- LinuxJedi for making the amazing PiStorm tool for the PiStorm Interaction Device and doing various PiStorm device testing
- runehol for the Musashi CPU emulator speedup.
And of course thanks to everyone who helped out testing wip-crap to make sure it didn’t explode in people’s faces the moment they start it up with these change.
Even More Speed!
It seems like every time I post one of these things get even faster! There have been a few developments in “wip-crap” (after the great merge of June, so not in “main”) that make things even faster still!
First up is the changes to the firmware made by shanshe. Who has managed to shave off some of the performance hit the Amiga’s Chip RAM takes with the “proto 3” firmware. In SysInfo 4.4 the chip speed increases from around 0.97 to around 1.13. Some games will run smoother now. _Bnu has also done a few tweaks on software side around this. We are observing right now that this has actually improved compatibility for some machines (we saw a CDTV fire up) but there have been a couple of reports of stability issues for others. Consider this very-beta right now if you want to try it out.
I should note at this point, this is unrelated to the “proto 4” firmware currently in development right now, which should improve the situation even more.
To apply a firmware update you’ll need to use
./nprog_240.sh as you did during the initial setup when you have updated the “wip-crap” branch and shut down the emulator.
In addition to the above, Rune Holm has been at it again. A patch this time which gains another 5-10% performance boost from the emulator!