Sorry if offtopic

Besides that your not the only one who asked if Apollo would be a good name.

james stewart wrote:

Just out of curiosity ARM has announced its up & coming 64bit ASIC based on ARMv8 architecture to be named after greek mythology namely Atlus and APPOLLO.   Will the softcores naming APOLLO be a problem ?

Not a problem if Gunnar's Apollo core is dead :/.

james stewart wrote:

The future Apollo cpu softcore looks very interesting, being rich in features and speed and a very much improved 68k instruction set. I wanted to ask, does the team now have an idea of how the core will perform on the Natamis Cyclone 4 fpga Vs the Stratix 4 in testing ?

Some of the hardware features like the cache unit and OoO integer division unit looked very good. You saw the new instruction set/ISA and liked it? Gunnar said the Apollo core ran at about 100MHz in the Cyclone 4 which was about 1/2 the speed of the Stratix 4 running at 200MHz. He thought the bottleneck was the decoder which he was planning to rewrite to achieve closer to 120MHz in the Cyclone 4 as I recall. Jen's has brought up some doubts as to Gunnar's skill at improving the fpga core. It's possible that his skill has improved but that's a mute point if he has quit the Apollo Project.

james stewart wrote:

Will it perform a little better than a rev6 060 at the same clock due to most instructions completing in 1 clock cycle any info would be great.

I can only guess that the Apollo core would have been a little better performance with optimized code as the rev6 68060. The Apollo core supposedly had a helper/simple integer unit able to process immediate and register effective addresses while forwarding them for the main integer unit to use in the same cycle. This is almost as powerful as the 68060's superscaler while missing some improvement where the reordering of instructions is not possible. I got the idea that some code fusion (combining instructions) that Gunnar had talked about was complicated and either not possible or would wait until the rest of the decoder was finished. Branch folding (0 cycle loops) may not have worked as well as the 68060 yet. Some instructions were slower than 1 cycle in fpga like 32*32 multiply and bit field instructions (usually require a 32 bit shift). Muxes are very costly in fpga and often not practical. A modern 68060 with larger caches, a link stack and a few other modern improvements would probably be about as fast if not faster.

Apollo
+ larger caches (writethru caching only)
+ large instruction fetch (but no decoder instruction queue)
+ more cache and register ports for reading
+ superscaler with simple/helper integer unit
+ more powerful instruction set/ISA
+ link stack for faster rts
+ OoO division so other instructions can be processed in parallel
+ branch cache with history based prediction and hint bit
- big shift, multiply and add/sub are slow in fpga
- muxes like the decoder instruction queue are not practical

There's a reason to believe that Gunnar at some point grew impatient and decided to earn some money on his own work, or maybe there's simply been some disagreements between team members not known to the public.

So, perhaps the Natami is ignored by Gunnar, but this doesn't mean the project is dead. Hopefully there are other people who can help Thomas further on his mission to make the project a reality (if he needs it).

I wish Thomas and Gunnar could speak out so assumptions like this could be avoided. Unfortunately they don't seem to be willing to do so. :(

To be honest, I think they owe their loyal supporters (after so many years) some well-chosen words. To withdraw completely from the limelight instead of talking to people creates more damage than one may think. It creates, at worst - false rumors, etc.
Although we cannot demand anything from them, I seriously believe they have a great responsibility as the project's prominent figures.

... and it doesn't help what the other team members tell us. The bosses themselves have to get on the path, otherwise discussions like this will never end.

This is very probable.  He know what is actually going on.

Even more probably - he was given some money to work for someone else and forget natami. Others too just we don't officially know.

If "team" would be serious they will not just leave without words, and explain everything.

Wojtek Wojtek Wojtek...

Gunnar was nothing. He waltzed around thinking he was wearing the big boy pants when the reality is he did nothing. He was also a liar. (Hint: Neither Starwars nor 194x Deluxe were Natami software. They were generic SDL based games).

I cannot remember him (or anyone else) saying that these two programs were "Natami" software. They were aimed to be for demonstration purposes and SDL is easy to port. So in this context you cannot tell him a liar. Or show me the text passage where he (or anyone) claimed that these programs are real Natami software.

I don't think it matters. I believe Gunnar's intentions were to make the games Natami games but the developer tools were not available at that time to do it. I expect his intentions were to have the games available for the Natami eventually. I will not and have not called Gunnar a liar based on what I know. I think he was well meaning and legit and then gave up. We all know people who have given up on the Amiga and moved on. They aren't bad people. They just don't have the grit that us stubborn persevering Amigans have :P. The only thing Gunnar did wrong was not saying goodbye to his friends before he left. That was kind of rude :/.

I do not say he is a liar, but there is a time and a place when to come forward with information.
He came too early with his, and this project has lost some of it's steam because of it.
There are more things that went wrong from a Public relations standpoint, and no I shall not go around pointing fingers.

BTW Olaf your doing a fantastic job on your end.