|
|---|
Golgoth 27
France
| |
Posts 185
29 Mar 2012 15:25
| André Jernung wrote:
|
You would have to ask the person who made the page, Gunnar von Boehn. I would guess that some of the numbers are tested and some predicted.
The page certainly was not meant to be spread to the public in it's current state, that is for sure. But unfortunately Samuel got a bit confused and posted it.
If we are a bit more patient, we might hear some information from the people who did the majority of the work on the core, Jens Kuenzer (Deep Sub Micron) and Christoph Hoehne.
|
Thanks.
Things still goes on, it's the most important. Samuel made us know that the core is making progress, now it's time to wait for the official statement when things will be ready.
| |
Megol .
| |
Posts 680
29 Mar 2012 16:14
| There have been soft cores with >450MHz clock frequency on two generation old FPGAs.
Did some analysis earlier and ain't convinced superscalar execution would be the best fit for FPGAs. Had a partial design with a deep decode pipeline running at 2x the base frequency and support for 2x frequency execution of a limited subset of 68k instructions. Never did enough work on it to know if it would work in reality though. The prototype didn't really use 68k instructions, had no data cache nor data memory access paths, no branch prediction or execution resources for anything complex...
| |
André Jernung
Sweden
| | (MX-Board Owner)
Posts 988
29 Mar 2012 16:42
| Nixus Minimax wrote:
|
Do you happen to know the price of the FPGA used in the Natami?
|
See my first post in this thread:
CLICK HERE
| |
Rune Stensland
Norway
| | (MX-Board Owner)
Posts 871
29 Mar 2012 17:15
| Nixus Minimax wrote:
|
[ if the performance of the N68k is really quite a bit better than that of other commercially available softcores, this would mean that a good hard macro of the processor should do quite well performance-wise. Even when compared to (low-end) present day technology.
|
Remember that the apollo creators have many years of experience in creating commercial logic. Jens was working in the core team that created the cell cpu in playstation 3. If you put a group of elite engineers and let them create a cpu from the heart, then you would expect something good.
But working in the spare time is difficult, and fixing the bugs is more borring than tuning the peak numbers:)
| |
Nixus Minimax
Germany
| |
Posts 273
29 Mar 2012 17:43
| Rune Stensland wrote:
| | Remember that the apollo creators have many years of experience in creating commercial logic. Jens was working in the core team that created the cell cpu in playstation 3. |
How could I remember if I never knew? :o) It is amazing how many Amigans became electronics engineers. Amiga coding was the reason why I became interested in processors, microchips and electronics. Eventually I became a transistor-level microchip designer for both analogue and high-speed digital logic (up to 10 GHz in standard CMOS more than ten years ago). A group of elite engineers built the Amiga in the early 1980s. The downside of the general technological progress is that it is impossible to fund such a project in today's world. You would have to have millions of venture capital only to produce a single test chip.
| |
Adrian Browne
Ireland
| |
Posts 172
29 Mar 2012 17:45
| All i got was jens worked on the ps3 cell processor.There is seemingly no shortage of talented people working on the team natami.
| |
Marcel Verdaasdonk
Netherlands
| |
Posts 3979
29 Mar 2012 19:30
| Actually lowest price i found was $46,000.- for designing a chip and making a production run this does not include cost for the designer's time spend.(don't ask about the quality)
| |
Nixus Minimax
Germany
| |
Posts 273
29 Mar 2012 19:59
| Marcel Verdaasdonk wrote:
|
Actually lowest price i found was $46,000 |
What technology would that be? Ten years ago just making the masks for the chip already cost MUCH more than that (magnitudes) in what was contemporary CMOS technology then.
| |
Marcel Verdaasdonk
Netherlands
| |
Posts 3979
29 Mar 2012 20:36
| Nixus 90nm node, the trick is to share the mask cost with other companies.(thus also sharing wafer space)
| |
Nixus Minimax
Germany
| |
Posts 273
30 Mar 2012 08:54
| Marcel Verdaasdonk wrote:
|
Nixus 90nm node, the trick is to share the mask cost with other companies.(thus also sharing wafer space)
|
Yes, sharing the masks for many test chips is what they always do for test chips. I'm still surprised that you can get a 90nm test chip for as little as that!
| |
Jacek Rafal Tatko
Espania
| |
Posts 607
31 Mar 2012 21:26
| Great info Marcel , I hope You are doing fine ... , this figure puts things into perspective , for good and better as it has become cheaper to do ...
| |
Marcel Verdaasdonk
Netherlands
| |
Posts 3979
31 Mar 2012 21:36
| Nixus Minimax wrote:
|
Marcel Verdaasdonk wrote:
|
Nixus 90nm node, the trick is to share the mask cost with other companies.(thus also sharing wafer space)
|
Yes, sharing the masks for many test chips is what they always do for test chips. I'm still surprised that you can get a 90nm test chip for as little as that!
|
I am cheating a little since I know who to ask when. ;)
Although my figures would be incorrect for the current economic climate.(I last checked before the current economic recession)
| |
SID Hervé
France
| |
Posts 663
04 Apr 2012 19:11
| Samuel D Crow wrote:
|
EXTERNAL LINK has some information about the CPU core.
|
HelloGreat work and interesting read! Does a unofficial simulation, an estimate, an idea or something that looks like was done with a combination (core; sub-​​core) or is it too early for that ? Thanks
| |
Samuel D Crow
USA
| | (Natami Team)
Posts 1295
04 Apr 2012 19:24
| Samuel D Crow wrote:
|
Samuel D Crow wrote:
|
EXTERNAL LINK has some information about the CPU core.
|
Correction: This is the future home of where some real numbers will be when they exist. These are currently placeholders. (Oops.)
|
@SID Hervé
It's too soon to say. The core isn't finished yet. The website linked is not finalized and contains fantasy numbers for now.
| |
SID Hervé
France
| |
Posts 663
04 Apr 2012 21:55
| No luck.
Anyway, it seems that the choices made have seduced. This should increase awareness of your work, which is good news.
But I regret that the designation of the processor have changed.
I think it may complicate somewhat the future communication of the project. Apollo is already known as an accelerator card in the Amiga world.
Amiga->Natami and M0x0->N0x0 was more coherent.
Nice work.
| |
Erik Bauer
Italy
| |
Posts 301
05 Apr 2012 14:46
| SID Hervé wrote:
|
No luck.
Anyway, it seems that the choices made have seduced. This should increase awareness of your work, which is good news.
But I regret that the designation of the processor have changed.
I think it may complicate somewhat the future communication of the project. Apollo is already known as an accelerator card in the Amiga world.
Amiga->Natami and M0x0->N0x0 was more coherent.
Nice work.
|
I too find Apollo a bit confusing... but whatever, a name is not that important when we have such (teorethical) performances.
| |
Børge Nøst
Norway
| |
Posts 53
05 Apr 2012 17:59
| Apollo was both a line of Amiga accellerators and a workstation/server brand in the 80s (_the_ brand for digital logic work AFAIK).
| |
Louis Dias
USA
| |
Posts 217
05 Apr 2012 18:34
| Børge Nøst wrote:
|
Apollo was both a line of Amiga accellerators and a workstation/server brand in the 80s (_the_ brand for digital logic work AFAIK).
|
EXTERNAL LINK
But perhaps this cpu should be call the Apollo 11
EXTERNAL LINK :)
| |
Deep Sub Micron
Germany
| | (MX-Board Owner)
Posts 567
06 Apr 2012 07:03
| Hopefully not like the Apollo1 space mission.
I don't use the codename Apollo and stick with n050 or n68050. The Apollo name is just to hide its 68k roots. It is to catch the attention of people who think 68k is dead. These people might stop reading when they know it is 68k (known as evil CISC).
| |
Erik Bauer
Italy
| |
Posts 301
06 Apr 2012 07:30
| LOL "Evil cisc"
| |
|
Welcome to the Natami lounge.