@ErikBauer

And how much of this is actually done in the chipset and not by using the second thread of the CPU core... one can tell when (if) SAGA sources are available.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@Lou

If you have done some research why don't you share it? Because I've not seen you do anything other than claim something very strange.

Blitting is possible using a standard CPU but is slower and takes more space than doing it in hardware. A CPU can also draw sprites and generate/mix sound but again that would take more space and be slower than doing it in dedicated hardware. Doing these tasks with a second thread of a softcore processor is possible but would slow down the main thread even in a super-pipelined barrel processor design as either it will share caches or will have delay inducing cache bypass paths. Why would they knowingly slow down 68k execution even before SAGA was created?

Doing it with a custom processor design intended for this specific purpose would be possible and could perhaps even save some space but the main workload is still more suitable for dedicated DMA controlled logic, the processor in this case would control dumb logic instead of adding more logic.

Last edited by megol on 04-Oct-2019 at 02:22 PM.

@thread

To ramp up my interest:
-develop microATX board with PCIe slot (for RadeonHD) and a CPU slot (for PPC/ARM/A64/anotherFPGA).

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@kolla

I'll rely on your expertise to dissect and explicate the sources, then, as I'm a C/C++ guru but not well versed in Verilog or VHDL. My bad for yielding to a sense of the sarcastic.

I'd say it was a stroke of brilliance and an elegant solution to the problem. Kudos to the team.

Now if only I had a government job and could afford one ...

@Snorg

Quote:

Snorg wrote: @kolla I'll rely on your expertise to dissect and explicate the sources, then, as I'm a C/C++ guru but not well versed in Verilog or VHDL. My bad for yielding to a sense of the sarcastic.

Well, it should be as easy as implementing the sources for a different system that doesn't have AC68080 and see how that works. Right? Among the people who develop cores for MiST, MiSTer, FPGA-Arcade and others, there should be enough expertise to verify things.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@megol

Quote:

megol wrote: @Lou If you have done some research why don't you share it? Because I've not seen you do anything other than claim something very strange. Blitting is possible using a standard CPU but is slower and takes more space than doing it in hardware. A CPU can also draw sprites and generate/mix sound but again that would take more space and be slower than doing it in dedicated hardware. Doing these tasks with a second thread of a softcore processor is possible but would slow down the main thread even in a super-pipelined barrel processor design as either it will share caches or will have delay inducing cache bypass paths. Why would they knowingly slow down 68k execution even before SAGA was created? Doing it with a custom processor design intended for this specific purpose would be possible and could perhaps even save some space but the main workload is still more suitable for dedicated DMA controlled logic, the processor in this case would control dumb logic instead of adding more logic.

I have done my research. That was the answer I got when proposing enhancements to other chip for the purposes of parallel processing functionality to be able to do 3D at something "close" to a PS1. Mind you - all consoles of the mid 90's had slow cpus but a dedicated 3d chip that would do operations in parallel are why they could pull off 100,000 polygons/second.

When I essentially called that design hubris and pointed out that this is exactly why the CELL processor failed and that's why Sony added the NVIDIA gpu to the PS3, I was banned.

If you are gonna call me wrong - do some f'ing research. Don't call me wrong then expect to prove you wrong. Your statements are empty. No one from the Vampire team is disputing this. Gunner is building a monolithic cpu...which is exactly why Intel is failing and AMD is succeeding in the current market.

You're SAGA-dreams are far from reality. The funny thing is I agree with what you said but that is not Gunner's reality. Amiga OS doesn't know about the 2nd cpu thread. So rather than it going to waste - it is SAGA. As I said before, an 020 with FASTRAM can blit faster than Agnus...so guess what's doing the blitting...

Last edited by Lou on 04-Oct-2019 at 09:27 PM.

Ok you've convinced me.

For an amusing read of competency:
http://apollo-core.com/knowledge.php?b=4¬e=17188&x=1

On page 2 it is mentioned that the 2nd thread is used to emulate the blitter.

The best is when Steve Ferrels mentions AMD incorporating the gpu to the cpu package (aka APU) thinking that it's all 1 chip. "all the rage but its 2018 now" LMAO!

Here's an AMD "chip"...

Last edited by Lou on 04-Oct-2019 at 10:28 PM.
Last edited by Lou on 04-Oct-2019 at 10:26 PM.
Last edited by Lou on 04-Oct-2019 at 10:25 PM.

@Lou

Quote:

The best is when Steve Ferrels mentions AMD incorporating the gpu to the cpu package (aka APU) thinking that it's all 1 chip. "all the rage but its 2018 now" LMAO!

And the even bigger laugh was you getting banned completely from the forum shortly thereafter because you're a complete buffoon who kept demanding that Gunnar incorporate an Akiko chip into the Vampire because you believe it has 3D capabilities.

AMD and Intel have done substantial work incorporating GPU's onto the same dies as CPU's for good reasons which went completely over your head. And they've been doing it for years.

You need to read this and stop making an @ss of yourself: https://www.quora.com/Why-cant-we-combine-GPUs-and-CPUs-into-one-processing-unit

natami/vampire/apollo followers spread their usual propaganda.
The truth about saga is it does not contain any tiny tiny piece of chips made by Commodore, any sigle bit of it.
saga was made entirely from scrach.
It is not evolution of chips made by Commodore, saga has nothing in common with chips made by Commodore. It is only partialy compatible.
After many years of lies gvb and his puppets about performance and compatibility, future customers should not believe them without proof, and wait for independent tests made by users outside natami/vampire/apollo sect.
Because it was made from scratch it does not share history with Commodore.
It is replica not original, even if partialy compatible.
v4 is 68k NG not classic Amiga.

this whole case of second core used in vampire is very easy to explain.
natami/vampire/apollo team has not enough resources to made real blitter in FPGA.
So they add second core and run emulation of blitter on it.
It solve problem in bad way of Commodore tradition, but as 1.44 MB FDD in A4000 or 256 colors bitplanes, makes saga many times slower than it should be.
It is good idea to add external graphics to vampire, of course it may be another fpga not common graphics card on pcie.

@Lou

Quote:

Lou wrote: Ok you've convinced me. For an amusing read of competency: http://apollo-core.com/knowledge.php?b=4¬e=17188&x=1

For a second I thought you actually given some information but then realized that is a link to page 2 of a subsection of the forum. A link to a thread is in another format.

Do you understand that I've searched the forum repeatedly and can only find one person saying a second thread of the processor is used to emulate a blitter? Guess who that person is?

Quote:

On page 2 it is mentioned that the 2nd thread is used to emulate the blitter.

Great. So one of the threads on page two the second you posted the link contains the information, why can't you say which thread that would be?

Quote:

The best is when Steve Ferrels mentions AMD incorporating the gpu to the cpu package (aka APU) thinking that it's all 1 chip. "all the rage but its 2018 now" LMAO! Here's an AMD "chip"...

That's a Zen 2 chiplet design intended to share dies scaling from mainstream processors up to multi-socket server/supercomputer processors. But all AMD APUs, those processors that include a GPU, are a single die.

Last edited by megol on 05-Oct-2019 at 10:10 AM.

@ppcamiga1

Quote:

ppcamiga1 wrote: natami/vampire/apollo team has not enough resources to made real blitter in FPGA. So they add second core and run emulation of blitter on it. It solve problem in bad way of Commodore tradition, but as 1.44 MB FDD in A4000 or 256 colors bitplanes, makes saga many times slower than it should be. It is good idea to add external graphics to vampire, of course it may be another fpga not common graphics card on pcie.

It is a better 'classic' computer - 'emulation' which is much, much faster than the original (how fast, exactly ... well it would never be fast enough, I guess ...) which actually has been realized.

They took what they had available to them and made it work. Come on, people, be reasonable everyone - were they going to put a 3D card into the thing any time soon?

We are reasonable.
It has not original chips form Commodore so it is not classic, it is 68k NG.
Some thing may be done better and should be done better.
vampire should have faster graphics.

@kolla

Quote:

kolla wrote: @ErikBauer And how much of this is actually done in the chipset and not by using the second thread of the CPU core... one can tell when (if) SAGA sources are available.

Yes, because secondary 68k CPU cores are well known for handling Hardware Sprites, Coppering and behave like Paulax2. Of course all and every 68K machine had this functions natively

LOL

_________________
God created Paula so that Allister Brimble and Dave Whittaker could do music

Check my Amiga gameplays (ITA)!

@ppcamiga1


ppcamiga1 wrote:
natami/vampire/apollo followers spread their usual propaganda.
What about that propaganda of yours?

Quote:

The truth about saga is it does not contain any tiny tiny piece of chips made by Commodore, any sigle bit of it. saga was made entirely from scrach.

Of course it is like you say. SAGA has ben built from scratch

Quote:

It is not evolution of chips made by Commodore, saga has nothing in common with chips made by Commodore. It is only partialy compatible.

Define partially compatible: more and more 68K demos, games and programs are running upon SAGA. Thee harware registers are there, the blitter blits, the copper cops and audio chip plays sounds executing commands directed to Paula.
Where's the problem?

Quote:

After many years of lies gvb and his puppets about performance and compatibility, future customers should not believe them without proof, and wait for independent tests made by users outside natami/vampire/apollo sect.

What about the Anti vampire/apollo sect of yours?

Quote:

Because it was made from scratch it does not share history with Commodore. It is replica not original, even if partialy compatible. v4 is 68k NG not classic Amiga.

Of course it is replica, nowhere a sane person could believe it is "Commodore original"
And of course V4 is 68K NG

What is the point here?

_________________
God created Paula so that Allister Brimble and Dave Whittaker could do music

Check my Amiga gameplays (ITA)!

@ErikBauer

Quote:

What is the point here?

Indeed. No idea what your "LOL" was about, I am just saying that all this speculation is pointless - Gunnar har been quite explicit about a few things, like the MMU and possibilities to use second CPU thread to implement DMA and other "behind the curtain" work, exactly how much of this has been implemented and is actually already in use... only a handful few know. Regarding the blitter and other stuff I don't care squat about, it will be evident when the SAGA sources are made available for use with other FPGA systems - I don't grasp what's so controversial about this.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@megol

http://apollo-core.com/knowledge.php?b=4&note=17188&order=&x=2

Last edited by kolla on 05-Oct-2019 at 01:16 PM.
Last edited by kolla on 05-Oct-2019 at 01:15 PM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@ferrels

Quote:

ferrels wrote: @Lou Quote: The best is when Steve Ferrels mentions AMD incorporating the gpu to the cpu package (aka APU) thinking that it's all 1 chip. "all the rage but its 2018 now" LMAO! And the even bigger laugh was you getting banned completely from the forum shortly thereafter because you're a complete buffoon who kept demanding that Gunnar incorporate an Akiko chip into the Vampire because you believe it has 3D capabilities. AMD and Intel have done substantial work incorporating GPU's onto the same dies as CPU's for good reasons which went completely over your head. And they've been doing it for years. You need to read this and stop making an @ss of yourself: https://www.quora.com/Why-cant-we-combine-GPUs-and-CPUs-into-one-processing-unit

LOL! I take my ban as a badge of honor.
Using Intel to prove my point is a bad example, by the way. Intel is losing market share because they aren't using a chiplet design and makes those monolithic dies (that still use separate gpu and cpu cores).

Do enlighten us with some more hilarious non-insightful technology comments though!

@megol

Quote:

megol wrote: @Lou Quote: Lou wrote: Ok you've convinced me. For an amusing read of competency: http://apollo-core.com/knowledge.php?b=4¬e=17188&x=1 For a second I thought you actually given some information but then realized that is a link to page 2 of a subsection of the forum. A link to a thread is in another format. Do you understand that I've searched the forum repeatedly and can only find one person saying a second thread of the processor is used to emulate a blitter? Guess who that person is? Quote: On page 2 it is mentioned that the 2nd thread is used to emulate the blitter. Great. So one of the threads on page two the second you posted the link contains the information, why can't you say which thread that would be? Quote: The best is when Steve Ferrels mentions AMD incorporating the gpu to the cpu package (aka APU) thinking that it's all 1 chip. "all the rage but its 2018 now" LMAO! Here's an AMD "chip"... That's a Zen 2 chiplet design intended to share dies scaling from mainstream processors up to multi-socket server/supercomputer processors. But all AMD APUs, those processors that include a GPU, are a single die.

https://en.wikipedia.org/wiki/Die_(integrated_circuit)
Do the gpu instuctions exist in a CPU core? No. They are still separate cores.
Gunnar's solution to everthing is to add more cpu instructions creating more incompatibilities...then wondering why there is no GCC support. This is a CISC-to-the-extreme when long a go everyone else realized that a cpu is good a cpu things and other chips are better at specific things and can work independently speeding up the system as a whole. The Amiga was about custom chips working in unison with the cpu not about the cpu doing everything.

To beat a dead horse - look at the cpu power of the PS1, Sega Saturn, N64. Other than the N64 which had a 90ish Mhz cpu, all those systems ran at around 30Mhz but had a dedicated parallel-processing "3D chip"...and will always run circles around the Apollo core for 3D. Even adding multiple cores to Apollo you eventually flood the memory bus where all that's coming to the cpu is 3D instructions and there is no room to run actual game/app code. That's why you let a cpu be a cpu and a gpu be a gpu.

Last edited by Lou on 05-Oct-2019 at 04:42 PM.
Last edited by Lou on 05-Oct-2019 at 04:37 PM.
Last edited by Lou on 05-Oct-2019 at 04:30 PM.
Last edited by Lou on 05-Oct-2019 at 04:30 PM.
Last edited by Lou on 05-Oct-2019 at 04:22 PM.
Last edited by Lou on 05-Oct-2019 at 04:16 PM.

@kolla

Thanks. Still that doesn't make any sense.

@Lou

Yes it makes more sense to use dedicated hardware. It reduces overhead, increases processor performance, and increases graphics performance with everything else being equal.

Last edited by megol on 05-Oct-2019 at 11:34 PM.