A1200 Acceleration - which route?

[masterchief]

Originally Posted by grond

And where does the "almost as fast as a 060 accelerator" come from? No 040 accelerator is anywhere close to an 060 accelerator in speed.

In any case, Matze is developing an 060 card which I would expect to hit a very sweet price/performance ratio.

https://forum.icomp.de/index.php?thr...=6346#post6346

"The old Apollo and Blizzard 060 accelerators clocked at 50MHz gave about 37000 Dhrystones in Sysinfo, which is of course awesome: It's about 3 times faster than the fastest 68030 that we can offer. However, even with our first prototype, which does not have all the optimizations that the next prototype will have, we get well over 30k Dhrystones for a 40MHz 68040, and memory access performance that has never been shown on a 68040 before."

Who is Matze, really interested in that card you mentioned.

[grond]

Originally Posted by masterchief

https://forum.icomp.de/index.php?thr...=6346#post6346

"The old Apollo and Blizzard 060 accelerators clocked at 50MHz gave about 37000 Dhrystones in Sysinfo, which is of course awesome: It's about 3 times faster than the fastest 68030 that we can offer. However, even with our first prototype, which does not have all the optimizations that the next prototype will have, we get well over 30k Dhrystones for a 40MHz 68040, and memory access performance that has never been shown on a 68040 before."

Well, as everyone knows SysInfo is a bullshit pseudo benchmark that has a completely nonsense "work loop" that doesn't do anything real. The one thing it does is prohibit any superscalar execution of code so that the 060 drops down to one instruction per cycle which is the maximum the 040 could do. So yes, a 40 MHz 040 will just give the same result as a 40 MHz 060. However, the 060 would still be faster in real-world applications because it can put its second pipeline to use and execute some instructions in parallel. (EDIT: plus the 060 has double the caches of the 040 which doesn't make a difference in SysInfo either but in real world applications)

Hence, the above is just marketing BS from Jens. He should use the updated SysInfo version that allows superscalar execution of the nonsense "work loop" and compare again. But this would make his 040 card look much worse. Jens obviously wants to sell 1240s because he can't source 060s (just like everyone else) and to make people upgrade to 060s losing their warranty. His 1260s will be priced prohibitively.

Who is Matze, really interested in that card you mentioned.

Matze is a local hero at the German a1k.org forum who has made many accelerator cards and add-ons that usually end up being open-source hardware products.

[Chucky]

I also heard that upgrading his 040 to 060 will require a 060 license..

[AMike]

Originally Posted by Chucky

I also heard that upgrading his 040 to 060 will require a 060 license..

Yeap - a licence for the memory controller is needed and you loose your warranty. I will go for a Warp1260 - a great card with big potential.

[grond]

So even if you have that lovely Rev 6 in your drawer, you would have to pay 499€ for the 1240, some unknown extra money for the 060 upgrade and end up with a card without any warranty. Hardly an attractive or even budget option.

[Chucky]

"yupp"

[lmimmfn]

Lol, the license requirement is a handy automated function to invalidate the warranty.

[Chucky]

And a way to squeeze extra money of the situation

[Storm]

I love my Blizzard 1260 in my A1200, even though it is a rev1 heat tile. It's all about the old smell and feel. Authentic, beautiful
My A4000 is lucky to have a rev6 in it's CSMKII@72MHz. I must be an 060 fanboy lol.

[E-Penguin]

Originally Posted by Chucky

I also heard that upgrading his 040 to 060 will require a 060 license..

It's a paid product ; if the software / firmware is different between the two then a fee for the other one isn't unreasonable. They're a business, can't fault them for wanting to make money.

[Higgy]

Help us Matze, your our only hope

A1200 connector availability - sorted
CPU socket - a must
Memory - 64mb min
5v header - for DIY fan hacks
68060 - leave it to us

[aeberbach]

Having to pay for a license to plug in a 060 seems silly... definitely makes me lean toward the competition.

[E-Penguin]

You can plug it in for free but if you want the compatible firmware it'll cost. No different to buying 3.1 4 separately for each Amiga model. It's reasonable for developers to get paid for their work, no?

[klx300r]

@ Chucky


any more testing on the TF1260 for us future users patiently waiting

[Chucky]

for the moment I do not know the future of the TF1260 unfortunally :-(

[klx300r]

Originally Posted by Chucky

for the moment I do not know the future of the TF1260 unfortunally :-(

but I thought it was going to be sold by 'trusted' Amiga people/ companies moving forward...please don't tell me the actions of one idiot will ruin it for the rest of us

[Chucky]

as told.. for the moment I do not know more.. I can say I had some sleepless nights about this..

[klx300r]

Originally Posted by Chucky

as told.. for the moment I do not know more.. I can say I had some sleepless nights about this..

well I'm sure many like me appreciate the work you've done and do for us amigans so hopefully things get worked out in the end

[aeberbach]

Originally Posted by E-Penguin

You can plug it in for free but if you want the compatible firmware it'll cost. No different to buying 3.1 4 separately for each Amiga model. It's reasonable for developers to get paid for their work, no?

It certainly is, but it is silly to price it so that the developer isn't being paid until you need the firmware for the better chip. It's also silly to pretend that any number of people are going to buy it and only ever use it with the 68040. This is just an unnecessary complication to the product.

[E-Penguin]

On the other hand, why make people who are happy with an 040 pay for the 060 firmware they won't use?

I don't know what the cost difference would be, maybe it's negligible compared to the cost of the card so they might as well bundle it in but then people would complain that they're paying an extra $(whatever) for functionality they don't want/need.