ACA500 tested

[Schoenfeld]

Originally Posted by Dejavu

For this beta update, does it have any improvement other than ACA1221 compability issue?

It may also fix problems with the recently unsupported ACA1230 and ACA1231 cards. That's a total of 4 models that were 64MByte-cards, and the fix for the ACA1221 included a fix for the 64MByte memory map that is shared between those old cards and the ACA1221. While neither the ACA1230, nor the ACA1231 will be supported by the ACA500 officially, chances have gone up that they will work now.

Originally Posted by Dejavu

Do you have any news about ACA500Plus for us?

Not really - only that it's unlikely to happen this year. One thing that took up quite some time was...:

Originally Posted by Dejavu

And it just caught my eyes, There are some infos about ACA1232@50 & 1233@55 on wiki page ACA12XX benchmark table. Are those prices and availability dates real?

yes, they are real. I've put quite some time into debugging the 50MHz design. After that was stable, calculations have shown that the new memory controller is within specs at over 57MHz (assuming a jitter-free clock source), so I decided to make a few cards with slight overclocking and run the whole test cycle again. Note that only the CPU is overclocked, but all other components are within spec (same approach as with the ACA1221).

Since the available quantity is tiny, but these few cards need to pay for this new SD-Ram controller development, prices may appear a little higher than prices for my previous designs.

Jens

[duga]

No ACA500Plus (this year) means no ACA2000 either?

What about new A500/A600/A1200 PSU:s?

[Schoenfeld]

Originally Posted by duga

No ACA500Plus (this year) means no ACA2000 either?

Right. The ACA2000 will be built on the ACA500plus, only removing the 68EC000 processor and adding the right connector. That one will *require* an A1200 accelerator to work.

Originally Posted by duga

What about new A500/A600/A1200 PSU:s?

They won't be PSUs, but DC-DC converters. I may have a prototype until Christmas (as that's one of the building blocks of the Amiga Reloaded, which is scheduled for prototyping in Q1/2016), but production units won't be available this year.

The good news is that the 50 and 55MHz cards have been finished sooner than expected. The units passed QC now (warranty ID, QC photo), and we'll finish packing tomorrow morning. Taking orders now -> scroll down in the product list and click on ACA12xx accelerators, then choose the variant.

Jens

[Dejavu]

Jens,
Your website (icomp.de) needs to be improved. It could not be seen from neither android nor ios web browsers.

[TjLaZer]

Tested the new firmware with my ACA1221 and ACA500 and it works great!

Thanks Jens!

[AEV]

Hi Jens, any chance of the new ACA1233/57 FPGA overclocking core being made available for the existing ACA1233/40 cards via an overclocking license file like the ACA1221?

Obviously there would be no guarantee of hitting 57MHz but any increase in speed would be welcome and it would increase your revenues on existing cards sold.

[Schoenfeld]

Originally Posted by AEV

Hi Jens, any chance of the new ACA1233/57 FPGA overclocking core being made available for the existing ACA1233/40 cards via an overclocking license file like the ACA1221?

Obviously there would be no guarantee of hitting 57MHz but any increase in speed would be welcome and it would increase your revenues on existing cards sold.

That would require the crystal to be exchanged and the CPLDs to be re-programmed (note that it's CPLDs, not FPGAs!). Both requires equipment that only very few "professional hobbyists" have at home. In other words: For such an overclocking "upgrade", the card would have to be sent to us for re-work, and EU laws would require me to guarantee that it works. Since I obviously can't do that, this is not an option.

Also, please note that the new top model is not running at 57MHz, but 55.55MHz.

Jens

[Retrofan]

Originally Posted by Schoenfeld

It may also fix problems with the recently unsupported ACA1230 and ACA1231 cards. That's a total of 4 models that were 64MByte-cards, and the fix for the ACA1221 included a fix for the 64MByte memory map that is shared between those old cards and the ACA1221. While neither the ACA1230, nor the ACA1231 will be supported by the ACA500 officially, chances have gone up that they will work now.

Jens

I suposse that what you say about the old -fastest*- ACAs also applies to the new A1200 reloaded ?

*I've just measured with SysInfo 12053 Dhrystones, 12.58MIPS with the latest ACATune you've uploaded. Edit: Yes, nobody has shown so much and I can post a picture if you don't believe me. ACA 1230/56 with Fast ATA MKIV, look my signature.

Hello,

I put my question here again:
Do anybody here has an ACA500 recovery disk? If yes, can you make me an .ADF from it and send it to me via e-mail?
(I just don't wanna disassembling my Amiga to make a backup)
Thanks!

[Apollo]

I can create an image of the recovery disk when I am at home.

[tom256]

How fast are 030/50MHz and 55MHz cards in Sysinfo??

Does 030/55Mhz faster than ACA1230/56MHz? Does it has faster communication with motherboard??

[Schoenfeld]

All performance data is available in the Wiki.

Sysinfo is a very bad indicator for speed. You should take the time and download the AIBB module, load it in AIBB and then start the "review mode" from the drop-down menu. This will give you good indicators of speed.

Both the ACA1232-50 and the ACA1233-55 have optimized Chipram performance, but they will never be able to compete with a syncronous design like the ACA1230. The old ACA1230 was fully syncronous, which is the fastest-possible way to communicate with the mainboard. However, it's also very piccy in regard to mainboard timing. We've had quite a few support cases where the ACA1230 could not communicate with the mainboard at all, and after I've changed over to an asyncronous design, the majority of these problems were gone.

This brings us back to topic: The ACA500 was also syncronous, but it was not compatible with all A500 boards. The new ACA500plus will be fully asyncronous, and the prototype has already shown that it's compatible with the problematic board revisions such as Rev.3 and the NTSC-Rev.5.

Jens

[CrashMidnick]

Works great here with ACA1221 + ACA500 and the new firmware on A1000

[Retrofan]

Originally Posted by tom256

Does 030/55Mhz faster than ACA1230/56MHz?

Nope. The ACA 1230/56 is the fastest ACA card ever made in MIPS with SysInfo.

[TjLaZer]

Originally Posted by Retrofan

Nope. The ACA 1230/56 is the fastest ACA card ever made in MIPS with SysInfo.

Are you sure?

ACA1230/56 left, ACA 1233/55 right

[Retrofan]

Wonderful. Thanks TjLazer. I'm glad to be wrong. Now you have a brand new wonderful accelerator to run AKReal.

It's true I get more MIPS, but I use a Fast ATA. You are comparing them with the same specs (well, I hope); same HD and same Amiga. Thanks again.

[Schoenfeld]

left picture shows "DBurst off" - the two can't be compared. Make sure that all caches and burst is on, install the latest setpatch, then run both tests again. Or better: Run AIBB and make modules, then use the "review mode" to compare the two cards in detail.

It's true that the SD-Ram controller of the 55.55MHz card is one clock cycle faster on writes, so it does have a chance to be faster on certain tests. However, the ACA1230-56 is syncronous, and that gives it advantages on chipram access. Further, it's even faster on NTSC machines, as the board frequency is multiplied: On PAL, it's 56,75MHz, and on NTSC, it's 57,27MHz.

Jens

[TjLaZer]

Turning it off did not change the speed.

Added PAL speed. It is slightly slower.

[Retrofan]

Originally Posted by TjLaZer

Turning it off did not change the speed.

Added PAL speed. It is slightly slower.

Make an acatune -status and see if the ACA1230/56 is recogniced.

[TjLaZer]

I don't have the ACA1230/56, only the new one.