As soon as a N70 is out, i will buy it, even to "just have it in my hands" and even if it's pricer than planned at first.

I humbly dare to propose a name for the soft-core :

69k

It will put aside legal litigation with the name 68070 wich was used in the CD-I Philips.
And it's a more "sexy" name, as all the post i read on it make it realy realy hot on computational side.
It is too, very dreamy specifications. When i put my mind back in the 1993. I would have kill to have this kind of performance.

All my wishes and hopes for this project.

Welcome to the site.

Well, the name 69000 is not bad.
What I like about the 68070 name is that shows both heratige and improvement to the 68060 CPU and that the name shows the full compatibilty to the 68K line ...
But that's just my opinion.

LOL!!!!!!!!!

So disturbingly smutty,..... I likey. :-D

Maybe, we love computers toooo much?

Hmmmmm,..... naahhhh, that is IMPOSSIBLE! ;-) :-DDDDD

But really, NatAmi, the way it's envisioned, is the PERFECT computer!

BIG HUGE performance with mamoth ram availability and storage capacity, near INFINITE flexibility, sleek algorithm set, and low low low low internal OS footprint (low power and memory consumpution).

I still can't get over this:
AOS 3.9
~400K / 512,000K = .07812% of ram used!!!

(I don't really know how big AOS is, but I'd say I'm pretty close.)

Sometimes "new" < > "progress".

Numbers are not patentable.  The End.
Numbers are not copyrightable.  The End.
Numbers are not trademarkable.  The End.

A picture of a number in a specific color and font could possibly maybe be trademarked but it would be stupid to attempt such a thing and would end up costing millions of $$$ in lawyer fees.

We are free to use the number 68070 for any purpose that we want.

It will be able to simultaneously write to the cache and read from the cache in the same clock cycle as long as they are different locations.

This action can be repeated very rapidly but there may be a lot of heat build-up, possibly resulting in an eruption.

The 69xxx will be directly coupled to Agnus, Denise, Paula and Amber.  The 69xxx can poke Paula, Agnus, Denise or Amber at will.

When operating under the 69xxx, Paula and Agnus will obey every command given.  In the Amiga architecture, the 69xxx is the Master and Paula, Agnus, Denise and Amber are slaves.

;)

I think I should look it up in my big badass hardware book but i thought you could read and write to VRAM at the same time at the same address.
but that is totaly off topic, just a question why do we wander off?

It was just an idea put on the table.
Of course is up on the natami team to have the idea to be or not to be picked.

I knew there was risk of "smutty collateral damage" with this kind of name, but it was more a artillery barrage of them :-)
(did not think about 69xxx, doooh)

Is there a point where, more is less, for instance 32 KB of each is great, but 64 is interfering with too many misses, and therefore flushes vs. usable data being available. (Well, that's kind of the best I could word it.)

Hmmmmm, a secondary ram area inside the CPU that DOES NOT get flushed, because it doesn't have to..... where you could put often used SW, like ppmore, dir, list, lha, and other PURE commands,.... or the OS itself, DPaint, etc., if it was big enough!!!!!

This is unique, as the 1 MB cache in an intel gets flushed if what ISN'T NEEDED isn't THERE, while THIS segment of CPU ram cache would NOT get flushed, but utilized WHEN NEEDED.

BASICALLY, RESIDENT SW inside the CPU ITSELF!

Does ANYONE ELSE have this, or is ABLE to have THIS? (This is OUR computer.)

_ONLY_,..... AMIGA makes it POSSIBLE!!!!!!!!!!!!

(I believe there was someone else besides me that thought NatAmi could change computing? :-D :-D :-D )

You can NOT use LE for cache as LE are much to expensive to build a cache out of them.

1 LE gives you only 1 bit.

The 68060 did have 16KB of cache + tags this is ~ 20KB total.
This would cost you 20,000 LE.

The way to build a cache is to use the embedded build in SRAM for the cache. Many FPGA have certain amount of SRAM.

We have to share the SRAM of the FPGA for buffers for the SuperAGA chipset, for cache for the 3Dcore, and for cache for the 070.
Our current plan is to allocate ~ 36KB to the caches of the 070.

I think this would give a good performance mix to each unit.