Excitement about NatAmi

[Louis Dias]

If there's one thing I've learned when doing low level work with real hardware, there's the theoretical speed of which the hardware is capable and the actual speed that is attained in the real world with applications that test it. The interface between software and hardware is often a murky place.

User
|
Application
|
OS Graphics API
|
Driver
|
Hardware

Remember, to use the blitter in an OS friendly manner, you have to wait for it, own it, set it up to do your operation, disown it etc. etc. All of those steps take time, which is more or less independent of how fast the blitter itself can do the operation you've asked.

Will you see a 100x speed up in a real world application? It all depends. If the setup cost is small, maybe you'll experience more than that. OTOH, if there are many small blits going on frequently, then the setup latency may dominate and you'll observe a lot less.

Theorectical maximums are the only thing worth comparing because all the inefficiencies you mention also apply to the old hardware.  Just by adding buffers, yarube has made the TG68 core run like an '040...yet I'm not allowed to cite that as co-oberating evidence?  Natami bus is about 4x faster than MikeJ's board directly due to DDR2 vs. DDR1.  It's fpga is also superior.  These are facts.

One of the factors that helps make full '020's, '030's etc... faster is cache.  The '050 core is getting, iirc, 32k of data and instruction cache.  This is also what yarube did to the TG68 core.  It's not rocket science to people who know what they are doing.  Gunnar and Thomas have real world industry experience.  I see too many people here belittling their work.  It annoys me, especially when some of those people are clearly threatened by the product because they are a developer for an OS who's market could be undermined by it's success.  Other people, **cough**the_leander**cough**, just aren't too bright.

[the_leander]

Other people, **cough**the_leander**cough**, just aren't too bright.

Now watch everyone as Hattig overlooks this too.

Remember folks, if you disagree with Lou, you are by his definition, subnormal.

[Louis Dias]

And instead of saying anything like you're implying, he went off on one about it being pure maths to prove his case that what he said was in fact, a fact. One which if I didn't agree to, would mean I was subnormal.

All my math proved was that I was being consevative.  That fact clearly was lost on you as is much I gather.  Even Karlos knows what I said is quite possible...yet you still are in denial.

You had no math to counter.  No argument what so ever other that "numbers don't mean anything" and "mr gamecube".  You couldn't even tell me the bandwidth of an A500.  My quick math tells me it's 7.14MB/s (potentially).  Natami's is 3200MB/s (potentially).  It's simple math and my 100x claim is the bottom end of what is stated is the Natami Q&A page.  So I was only repeating someone else's claim.  You are simply being combative out of ignorance.  Go educate yourself then come back with a real counter-argument if you can find one...or continue to defend the position of your own ignorance.

[jj]

Theorectical maximums are the only thing worth comparing because all the inefficiencies you mention also apply to the old hardware. Just by adding buffers, yarube has made the TG68 core run like an '040...yet I'm not allowed to cite that as co-oberating evidence? Natami bus is about 4x faster than MikeJ's board directly due to DDR2 vs. DDR1. It's fpga is also superior. These are facts.
 
One of the factors that helps make full '020's, '030's etc... faster is cache. The '050 core is getting, iirc, 32k of data and instruction cache. This is also what yarube did to the TG68 core. It's not rocket science to people who know what they are doing. Gunnar and Thomas have real world industry experience. I see too many people here belittling their work. It annoys me, especially when some of those people are clearly threatened by the product because they are a developer for an OS who's market could be undermined by it's success. Other people, **cough**the_leander**cough**, just aren't too bright.

 
Which people and which OS.  AROS or MorphOS or AmigaOS4 would be undermined by Natami?

[Louis Dias]

Now watch everyone as Hattig overlooks this too.

Remember folks, if you disagree with Lou, you are by his definition, subnormal.

It's one thing to disagree.  It's another to call someone out without a shred of merit behind it.  You started the insults.  Since mods seem to allow it, I'm simply returning the favor.

[Karlos]

Just giving you all enough rope to hang yourselves.

You don't have to take it.

[jj]

Ahhh the good ol asphyo w**k

[the_leander]

It's another to call someone out without a shred of merit behind it.

Yes, you hate being asked to back up your claims, we get it.

You started the insults.  Since mods seem to allow it, I'm simply returning the favor.

Actually, I think you'll find that you started throwing them around first.

But hey, it's not like either you or Hattig bother with little things like facts get in the way of your proving yourselves right.

All my math proved was that I was being consevative.  That fact clearly was lost on you as is much I gather.  Even Karlos knows what I said is quite possible...yet you still are in denial.

Not arguing possibilities, all I asked for was for you to provide evidence to back up your specific claims. Claims you have stated are facts.

What's genuinely terrifying is that Karlos has had to spell out to you the difference between theory and fact. Given your responses, I'm not entirely sure that it's a lesson that's sunk in.

You had no math to counter.

I'm still waiting for you to provide the benchmarks showing that the blitter is actually able to do what you claim it can. See Karlos' post for why.

Ad homs

Indeed.

--edit--

I'm done. Karl, you're right. I'll just leave anything with natami in the title in future.

[Karlos]

Theorectical maximums are the only thing worth comparing because all the inefficiencies you mention also apply to the old hardware..

I disagree. Most people buy new hardware on the basis of the real-world benefits they get, not what it's theoretical performance is. You don't see the latest CPU and graphics card round-ups quoting the theoretical performance of the hardware except as an annotation in the write up. All the comparisons are done with real-world applications, since that's what the target audience are most interested in.

By persistently hyping up the as-yet unreleased hardware based on theoretical performance projections of just the silicon, you are creating expectations that the end product might not fully live up to.

When the actual benchmarks of the NatAmi hardware appear that can be compared to real and emulated 68K systems running the same tests, then we can start drawing some meaningful conclusions.

[Louis Dias]

Yes, you hate being asked to back up your claims, we get it.


Actually, I think you'll find that you started throwing them around first.

But hey, it's not like either you or Hattig bother with little things like facts get in the way of your proving yourselves right.


Not arguing possibilities, all I asked for was for you to provide evidence to back up your specific claims. Claims you have stated are facts.

What's genuinely terrifying is that Karlos has had to spell out to you the difference between theory and fact. Given your responses, I'm not entirely sure that it's a lesson that's sunk in.

I'm still waiting for you to provide the benchmarks showing that the blitter is actually able to do what you claim it can. See Karlos' post for why.


Indeed.

--edit--

I'm done. Karl, you're right. I'll just leave anything with natami in the title in future.

Why do you continue the path of ignorance?

So published DD2 specs are lies?
So published Amiga bandwidth specs are lies?

The world is a lie.

[jj]

I love how definite everyhting is in your world.
 
All people are asking is for you to stop spouting these figures as facts.  They are not.  They very well be, but at the moment you have no proof or first hand experience to backup thse claims.
 
All you have to say is, the blitter could be 100x faster and poeple will be happy and except that.

[Louis Dias]

I disagree. Most people buy new hardware on the basis of the real-world benefits they get, not what it's theoretical performance is. You don't see the latest CPU and graphics card round-ups quoting the theoretical performance of the hardware except as an annotation in the write up. All the comparisons are done with real-world applications, since that's what the target audience are most interested in.

By persistently hyping up the as-yet unreleased hardware based on theoretical performance projections of just the silicon, you are creating expectations that the end product might not fully live up to.

When the actual benchmarks of the NatAmi hardware appear that can be compared to real and emulated 68K systems running the same tests, then we can start drawing some meaningful conclusions.

You can disagree but it doesn't make you right.
The bandwidth of the memory is the bandwidth of the memory.  These are facts.

The only thing that makes things into "real world" specs is the code driving the application.
To execute a memory move may take a couple of clocks.  So if you are moving 8 bytes, it may cost you 4 clocks in total because it may take 3 to set up the move.  In that case, it will look inefficient.  However the memory did move at full speed once it was initiated.  Now move 1MB.  It still only took 3 clocks to set it up, but once started it ran at full speed.  Now it looks closer to the theoretical max.  Real world applications don't change the facts that the memory copy actually happened at full speed of the hardware...they just throw setup time and expose natural application inefficiencies.

[jj]

There is really no hope is there.....................

[bloodline]

Ahhh the good ol asphyo w**k

Did you know I am indirectly responsible for Michael Hutchence takin his own life... True story!

Even weirder is that it is indirectly Amiga related!

[Hattig]

And if you look carefully, you'd see that there was a link explaining the meaning behind the "You are X and I claim my £5" from wikipedia.

Yeah, it's a light hearted comment.

Context, will you bother to supply any?

Bored of you writing this.

Maybe it is considered "aggressive phrasing" amongst the chattering classes, but down here on the front lines it's every day language used as a polite replacement for "Fuck All", the verbal equivalent of putting stars * in place of letters. In short, you're talking nonsense.

It wasn't the FA that I had the issue with. It was the entire phrase, including 'chum', which actually doesn't mean 'friend', 'buddy', 'mate', but is used to make a phrase more unfriendly and aggressive.

And instead of saying anything like you're implying, he went off on one about it being pure maths to prove his case that what he said was in fact, a fact. One which if I didn't agree to, would mean I was subnormal.

He tried to explain why it was likely to be the case, due to the more modern technology on NatAmi compared to the original Amiga technology, and relating it to FPGAArcade's attainment level in the same regard. I never saw anything from him stating that is was a solid fact set in concrete.

He would have done better to link to the NatAmi FAQ where it says 100x to 200x faster blitter directly, and then let the NatAmi developers back up their claims.

As for bugs in the chipset that were mentioned a lot, surely the NatAmi LX video from earlier this year should alleviate many such concerns?

[Hattig]

Other people, **cough**the_leander**cough**, just aren't too bright.

No need to reduce yourself to making insults too.