Individual Computers - New accelerator cards for A600 & A1200

[Schoenfeld]

Originally Posted by Vlad

I have the heat related issue with ACA 1231. My magic number is 55C CPU surface temperature, crashes start then. At 57C I have endless crash/reboot loop. 100 % repeatable.

1D1 mainboard, 3.0 kickstart, NTSC, A500 power supply, stock machine, only a laptop harddrive. Those temps are reached at idle. During the reboot cycle temp reaches 60C.

The temperature you're measuring is NOT the junction temperature, which is specified to be a lot higher for 68030 CPUs. The trouble with this small package is that junction temperature gets too high even with package temperature within limits (60 deg. C is not really hot). Like I wrote yesterday, a tiny heat-spread piece of metal makes sure that generated heat is moved away from the chip, making it 100% stable.

Originally Posted by Vlad

Jens, couple of questions:

1. Is it possible to downclock the CPU at all?

Yes, it is, but it requires soldering, so you'll lose warranty. The easiest way is to exchange the crystal. CPU frequency is crystal frequency/2*3.333. Memory refresh period is independent of the crystal frequency, because it's generated from the A1200 frequency, so you're free to choose any frequency below 25MHz and won't violate RAM timings (there's even a little headroom for RAM timing - the calculated-but-never-verified max. number for the crystal frequency is 28.1MHz).

Originally Posted by Vlad

2. Do you have official support forum?

I don't - I only give support via eMail. Contact me through my web form on icomp.de for that. However, please note that the ACA1231 is an OEM product, so Vesalia is your first port of call for support.

Originally Posted by Vlad

3. This heatsink kit from Vesalia:
a. Is there a timeframe for its delivery?

Later this week (shipping starts friday or saturday).

Originally Posted by Vlad

b. There is a piece of tape with something that looks like serial number on the CPU, does it have any special meaning or can I just get rid of it?

It's your serial number, and you must leave it on for warranty reasons. It can also stay on when the heatsink is placed on top of it. The source of heat is the center of the chip, and that is not covered by the serial number.

Jens

[Bamiga2002]

Does this overheating issue only affect 1231's or do 1230's need some cooling too?

[Schoenfeld]

Originally Posted by Bamiga2002

Does this overheating issue only affect 1231's or do 1230's need some cooling too?

ACA1230s are not affected, because they are using PGA processors (the package with "pins down, through the board"). These dissipate heat a little better than the QFP package that is used on the ACA1231.

Cooling never hurts, so you're free to add some if you feel like it. However, it may be more effective to reduce voltage on your power supply: We found during our tests that 5.3V instead of 5.0V increases heat by far. Your thought should be "eliminate the source of the problem, not the effect" (if there's a problem at all).

Jens

[Schoenfeld]

Originally Posted by Mr B

Sooner or later these things will start changing hands, and at that point, having to guess if yours will be more or less finicky with what drives it accepts...

There will be no guesswork. The changed logic has a new card ID, so ACAtune will tell you in clear text if the startup-wait is in or not.

Jens

[mfilos]

My own ACA1231 gets really hot as well (can't bare touching it after some minutes).
I haven't experienced any Gurus yet, but some instability was noticeable after some time. I was about to go buy some low profile heatsink when I saw Jens comment about Vesalia shipping the heatsink kit. Nice gesture tbh.

So far, ACA1231 is running pretty great kicking an 1MB custom kickstart of mine and with "cache" and "burst" options on.
If I add any other option that works on my ACA630 (like fastchip, fastz2 or z2cache) makes the ACATune not kicking the ROM and not using the cache/burst options.

Without using ACATune I get ~4.5MB/s transfer speed on my 1GB Sandisk CF, while using cache/burst on I get ~5.4MB/s (both speeds using Idefix Express without software installed).

Will have to give more test with different CF's (I have 4GB Transcend 133x and 8GB Sandisk Extreme) to see how it's going or else I'd have to send Vesalia my adapter for a boot-delay firmware change.

[Riempie]

Originally Posted by Bamiga2002

Does this overheating issue only affect 1231's or do 1230's need some cooling too?

I can't explain why, but my A1200 with ACA 1230 56mh became stable when I placed some very small heatsinks on some chips on the mainboard itself (not on the CPU of the ACA-card)...

[Schoenfeld]

Originally Posted by mfilos

If I add any other option that works on my ACA630 (like fastchip, fastz2 or z2cache) makes the ACATune not kicking the ROM and not using the cache/burst options.

Fastchip, Fastz2 and z2cache have been removed from the ACA1231 because they caused too many support requests. The only options available in the ACA1231 are "overall cache", "burst" and "maprom". Switching Dcache on for Chipram has caused instabilities, and ICache for Chipram has always been enabled. I am not planning on making any accelerator that has these options any more, as they are pure time-killers in support.

Originally Posted by mfilos

Without using ACATune I get ~4.5MB/s transfer speed on my 1GB Sandisk CF, while using cache/burst on I get ~5.4MB/s (both speeds using Idefix Express without software installed).

Not sure if your scsi.device is already accelerated in the inner copy-loop with move.l instead of move.w. If it isn't, try installing IDE-fix express software for a little more performance.

Jens

[mfilos]

Originally Posted by Schoenfeld

Fastchip, Fastz2 and z2cache have been removed from the ACA1231 because they caused too many support requests. The only options available in the ACA1231 are "overall cache", "burst" and "maprom". Switching Dcache on for Chipram has caused instabilities, and ICache for Chipram has always been enabled. I am not planning on making any accelerator that has these options any more, as they are pure time-killers in support.

I see Jens, no problem with that. I was just curious why it didn't work that's all

Btw, are the"slowpcmcia/slowpcmciamem" options give any serious boost if turned on or should be avoided for now (in case they are still experimental)?
Also does the "resetpcmcia" option is equivalent to the CardReset command for fixing the reset problem of PCMCIA bus?

Originally Posted by Schoenfeld

Not sure if your scsi.device is already accelerated in the inner copy-loop with move.l instead of move.w. If it isn't, try installing IDE-fix express software for a little more performance.

My scsi.device is 44.2 from Doobrey inside the custom 1MB ROM. It works very well with SFS/2 and atm I'm not planning to install Idefix.
DVD-RW that I have as secondary master, used to work just fine as well under OS3.9 without Idefix installed.

For some reason Idefix never worked nice on my systems (maybe because I always used custom ROMS with other scsi.devices than the original).

Dunno what is the max speed that Idefix Express with Idefix installed should reach, but I will try other scsi.devices and other CF cards at some point to see if I can reach a better speed. Till then, I'm pretty satisfied with the 5.5MB/s for IDE

[Mr B]

Originally Posted by Schoenfeld

There will be no guesswork. The changed logic has a new card ID, so ACAtune will tell you in clear text if the startup-wait is in or not.

It's still going to be sold as a "1231/42" so the second hand buyer will be guessing, and / or hoping for it to work when he gets it. Don't get me wrong, i love your stuff to death, and i know you support your buyers, i'm just saying that even someone who right now has no issues with the fast boot, might want to get theirs updated to future proof it.

B!

[melbourneben!]

Hi Jens, great work producing this card, thanks. Will the next batch sold have the new firmware installed as well as the heatsink?

[Schoenfeld]

All cards are currently being reworked, so whatever is not yet shipped, will be the latest and greatest.

I'll ship the coolers with all mounting material (partly pre-assembled) to Vesalia today, so they will start shipping tomorrow. The heat transfer pads will be "fixed for shipping" with an Amiga-related item...

Jens

[Retrofan]

Beautiful cooler!. I would like one of those for my 1230/56.

[Schoenfeld]

Originally Posted by Retrofan

Beautiful cooler!. I would like one of those for my 1230/56.

Not necessary, and would not fit with the trapdoor (socket+processor is already filling the space to the trapdoor). As I explained a few posts back, PGA packages can dissipate much more heat than the ceramic quad pack, so a cooler is not required. And just for the looks? Who looks under your keyboard?

Jens

[Retrofan]

Ok with "It isn't necessary", "doesn't fit", etc, but not with

Originally Posted by Schoenfeld

Who looks under your keyboard?

Who looks under the bonnet of his car? The answer is... who has something beautiful or powerful to show . Yes, I'm rare. I find the beauty in an inox screw.
Ok, anyway and thanks.

[vroom6sri]

That does look sweet! I'm looking forward to receiving one but, when you say " heat transfer pads will be fixed for shipping with an Amiga-related item..." does that mean we'll be getting a bonus item of some sort??!?!

[melbourneben!]

Good work Jens, I`ll be ordering one as soon as they become available.

[Schoenfeld]

Originally Posted by melbourneben!

Good work Jens, I`ll be ordering one as soon as they become available.

...like today? Vesalia marked it as "available" (green) for maybe 30 minutes during the day today, but you guys were really fast ordering. I'll ship the next stock to Vesalia tomorrow, so it'll be back in stock on march 8th.

Here's the new version of ACATune V1.5e, which supports not only the ACA1231, but also all previous versions of ACA turboboards. New in this version:

- recognizes the new ID for ACA1231 with the coldboot-wait
- allows moving VBR to fastmem (gives about 0.3% performance boost on my harddrive performance... wow ;-))
- allows running even older versions of Kickstarts on A600: V1.2, V1.1 and V1.0 also work with the patch. It's just V0.7 that we could not get to work yet, but I guess we can skip on that.
- commandline options don't have to be entered with "-" in front of them (but are still case-sensitive).
- polished the help text a little
- added a massive readme file with explanations for each option

Still experimental:
VBR moving to fastmem is marked as experimental, but works pretty well on all our test-setups. We will most likely mark it "working" if you guys don't report anything different.

SKick patchfile support: We still need your feedback on this. If you want to help, please get the SKick package from Aminet and try out some Patchfiles: http://aminet.net/package/util/boot/skick346

Under development: Changing a maprom'd kickstart without disabling maprom first. This will save you a reboot if you want to change Kickstart roms. Also under development: Loading a kickstart from an A1000 Kickdisk or an A3000 Superkickstart disk. This will save you the hassle of converting those disks to a binary file, just to find out if it was worth being converted.

So.. we're still not at the end of development with ACATune, even if the performance-tweaks have probably reached an upper limit now.

Jens

[Claw22000]

Originally Posted by mfilos

Anyway... since this thread became really large and it's really frustrating for someone searching it for the ACATune versions I uploaded to my Dropbox folder all the versions plus a ReadMe file with Changelog plus some other info.

I hope it's ok with everyone

• ACATune ReadMe
• ACATune 1.0
• ACATune 1.02b5
• ACATune 1.02b6
• ACATune 1.2b
• ACATune 1.4

Keep up the good work Jens

I hope you get a sticky for this since it helps so much to have a place where all the versions and a description file is!!! Please keep up the great work!

[mfilos]

What a lovely update (at least from what is written).
At last a ReadMe file with thorough details per option! \o/
I'll definitely test all features and report back

Thanks Jens


@Claw22000
Thanks a lot for the good words mate. Ofc I'll upgrade the post of mine with every new ACATune upgrade just for means of completeness and since some find it useful
TCD or a mod can edit the first post to redirect the following post instead of the oldest version post that exist now.

Updated post here: http://eab.abime.net/showpost.php?p=...postcount=2316

[melbourneben!]

I managed to get one ordered yesterday ;-)

Jens, I run WB 3.0, will the new version of ACAtune(V15e) resolve the issues with assigning the memory in WB 3.0?