Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
Please visit our Apollo-Discord Server for support.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Performance and Benchmark Results!

Why Is TerribleFire 160 Times Slower Than Vampire?page  1 2 3 4 

Kyle Blake
(Needs Verification)
Posts 108/ 1
22 Jan 2020 21:30


Adam Whittaker wrote:

 
David Wright wrote:

  Careful Kyle, Adam has rage quit this place more than once.
   
 

 
  Cock sucker
 

 
Amazing how difficult it is to be banned here. You're at least more concise than Thierry Atheist was.


Kamelito Loveless

Posts 260
22 Jan 2020 21:39


igor majstorovic wrote:

@Kamelito Loveless
  Ok let's make this constructive, this was the one. Kickstarter rejected me two times with explanation that this is not innovative enough and that it is not for them.
 

In the end you succeeded, for KS at least you tried so no regrets.


Mark Mc Fadden

Posts 36
22 Jan 2020 23:08


Mr Niding wrote:

@Mark Mc Fadden
 
  I was thinking the same thing when I saw this thread.

Thanks man, glad im not the only one :)

Gunnar von Boehn wrote:

Mark Mc Fadden wrote:

  I dont like this attitude, this is disingenuous to Stephens work and progress,
 

 
 
  Hello Mark,
 
  We discussed technical facts.
 
  - Discussing problems allows you to understand them.
  - When you have the understanding then you can give advices.
  - If advices are followed this can lead to an improvement.
 
 
  The idea of an open source 68k accelerator is good.
  This idea is not new, already in the 80th AMIGA and ATARI Magazins did publish "open-source" the plans to several simple accelerators similar to the TF520 / TF530.
 
 
  If any discussion can leads to improvements on the Terrible Fire then this is good for all users of it.
 
 
  I think there are 3 areas which could be reviewed.
 
  a) Bus access.
  b) E-Clock
  c) Caps
 
  a = can improve performance
  b = can improve compatibility
  c = can improve stability of the card and is important for the live time of the chips
 
 
  I have to admit, that I'm very annoyed about the missing caps on the Terrible Fire cards.
 
  Adding enough caps is crucial for stability and reliability of the card - and its also crucial for the live time of the CPU.
 
  When I look at the Terrible Fire cards, then my first though is "NOT ENOUGH CAPS".
  Is the reason for this laziness or lack of "knowledge"?
  If its lack of knowledge then such discussions - can help to provide the knowledge.

Gunnar von Boehn wrote:

Mark,
 
  Lets make clear here:
 
  we are not doing unqualified bashing or ranting here.
 
  We talk about technical problems and we can offer advice.
 
  I think you know that our advice is highly qualified.
  Everyone in the AMIGA scene does know that our Memorycontroller and our Buscontroller are second to none and  outclass all other Amiga cards.
  Its obvious that we perfectly understand the topics and we can give expert advice. And we are willing to help.
 
  Regarding the CAPs topic I'm disappointed.
  Mainly because I raised this topic in good faith not long ago, and was told that there would be XX caps on the 060-card, but I could only count halve of them.
 

Gunner, i understand what you are saying and i agree completely that that version of the card can be subject to some interesting analysis( and honestly i enjoy such analysis ), however that is not how it is being portrayed here, technical analysis should be unbiased and non political.
The thread title is "Why Is TerribleFire 160 Times Slower Than Vampire?", not "Interesting analysis of an underperforming Accelerator", the OP also reiterates the same "In fact its only providing a 160th fraction of the performance of a Vampire....." in a sales pitch/one-upmanship.
A technical discussion does not need this and a biased technical discussion is not a technical discussion.

Igor's comments in this thread are a prime example of keeping a technical discussion technical and only technical and he does a fantastic job of it.

If it is to be a valid and unbiased technical discussion, change the title and remove performance comparison in the OP with the vampire( its pointless anyway a valid argument here would be to compare with various 14MHz 68020 cards for A500 or higher clocked 68020 cards.
Again i have no interest creating an account to a forum just to be complaining( or telling you what to do ), i'm simply giving my honest( and what i consider quite fair ) opinion and i feel  there's already enough non-sense in the Amiga community unfortunately that this sort of comparison with Vampire is not warranted and achieves nothing in the context of an open technical discussion on the under performance of the card.



M Rickan

Posts 177
23 Jan 2020 02:09


Gunnar von Boehn wrote:

You can today buy V500/V600/V1200 and V4SA.
  No other card is planned. 

So reading between the lines, the next generation of the Vampire will be ASIC?



Sean Sk

Posts 488
23 Jan 2020 05:42


Adam Whittaker wrote:

Cock sucker

Grow up! What are you, a child?



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
23 Jan 2020 06:41


m rickan wrote:

  So reading between the lines, the next generation of the Vampire will be ASIC?
 

   
No - I did not mean this.

ASIC production does cost a huge a mount of money
and we do not have this yet and we are also not there in our development cycle.
   
   
What I tried to say is that designing, testing and producing the Vampire card needs a huge amount of time.
I think this should be clear.
We have today no free time available to develop more models.
This means for the next 2 years
I do expect that the current models are it.
We might maybe make a small change on the V1200 so that it physically also fits CD32 but there are few resources to develop more models. Maybe there might come a modified card for e.g. the A3000/4000 but I can not promise any dates yet.
Our main effort will go into continue to produce the existing models
to continue to support these models and to develop
core improvements like 3D core or GOLD 3 release,
help to test and improve AROS and ATARI EMUTOS.

My feeling is that its in the best interest for all to
produce the amount of cards that the people want
and to support them best possible.

I see very little sense in "splitting" the community
and splitting our team support capacity by creating unnecessary number of sub models. E.g card with less memory, or cards with no video.




Kyle Blake
(Needs Verification)
Posts 108/ 1
23 Jan 2020 14:41


For all our sakes don't even daydream about new models when there's work still to do on the existing ones.

When V4 becomes feature complete and as reasonably possible bugfixed, then it makes sense, not a moment later.

I have my own daydreams about a bigger vampire standalone, but I'm not going to go begging or bothering people for it.


Chris Sanz

Posts 25
23 Jan 2020 19:52


Gunnar von Boehn wrote:

m rickan wrote:

  So reading between the lines, the next generation of the Vampire will be ASIC?
 

   
  No - I did not mean this.
 
  ASIC production does cost a huge a mount of money
  and we do not have this yet and we are also not there in our development cycle.
   
   
  What I tried to say is that designing, testing and producing the Vampire card needs a huge amount of time.
  I think this should be clear.
  We have today no free time available to develop more models.
  This means for the next 2 years
  I do expect that the current models are it.
  We might maybe make a small change on the V1200 so that it physically also fits CD32 but there are few resources to develop more models. Maybe there might come a modified card for e.g. the A3000/4000 but I can not promise any dates yet.
  Our main effort will go into continue to produce the existing models
  to continue to support these models and to develop
  core improvements like 3D core or GOLD 3 release,
  help to test and improve AROS and ATARI EMUTOS.
 
 
  My feeling is that its in the best interest for all to
  produce the amount of cards that the people want
  and to support them best possible.
 
  I see very little sense in "splitting" the community
  and splitting our team support capacity by creating unnecessary number of sub models. E.g card with less memory, or cards with no video.
 
 

So I guess this means there will be no V4 for A500? I'm having a bad day now.



Markus B

Posts 209
23 Jan 2020 21:24


@Chris

This was stated quite some time ago.
I think only Vojin refused to accept that fact.


Markus B

Posts 209
23 Jan 2020 21:34


Gunnar von Boehn wrote:

You can today buy V500/V600/V1200 and V4SA.
  No other card is planned.

I think it makes some sense to publish this important information more prominently.

There are still people around who have hopes for all kind of mentioned possible products in the future. Somehting for the CD32, A3000/4000 and V4 versions as accelerators.
Vojin even makes new models up as he goes. If I remember correctly, his ASIC version is the Vampire V6. ;-)

Absolutely no offense intended, as I can totally understand the huge amount of effort needed to support and develop the current line of products. But I observe that this is not clear to everyone out there.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
23 Jan 2020 21:45


Markus B wrote:

Gunnar von Boehn wrote:

  You can today buy V500/V600/V1200 and V4SA.
  No other card is planned.
 

 
  I think it makes some sense to publish this important information more prominently.
 
  There are still people around who have hopes for all kind of mentioned possible products in the future. Somehting for the CD32, A3000/4000 and V4 versions as accelerators.

You are right.
Stuff like a CD32 version is very likely to come sooner or later
- as CD32 and A1200 is the same and only some "wire" change is needed.
But today we know hundreds of people wait for the Vampire.
We should support them now and focus on production instead spend month on different model design.


Kyle Blake
(Needs Verification)
Posts 108/ 1
23 Jan 2020 23:24


Would it be possible to just make passive adapter for CD32?

Come to think of it, i'm sure there's already some extended i/o board for cd32 with serial port, floppy connector etc that also gives a1200 cpu slot. But I might remember that wrong.


John Heritage

Posts 111
25 Jan 2020 09:39


Gunnar - TF520 does much better on Atari ST :

EXTERNAL LINK 
What bus differences would explain this?  Is STs main bus accessed every two cycles instead of four?


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
25 Jan 2020 10:38


John Heritage wrote:

Gunnar - TF520 does much better on Atari ST :

 
Thats a very good question.
The bus cycle on AMIGA and ATARI is the same.
Its always 4 Cycles.
 
The TF520 is a very simple card.
It basically only has the CPU and nothing else on it.
Therefore the memory of the mainboard is used.
The bus to the main board becomes now very important.
If the access for example is not done in 4 cycle but in 5/6 cycle,
then the performance will be severely affected.

The timing on the AMIGA bus is "very short" in some areas.
The timing on the ATAIR is slightly different.
It could be that a minimally delay on a signal like AS
will create a 5/6 access on the Amiga but might
still be OK for a 4 cycle access on the ATARI.

Someone would need to scope the bus to verify this.

There were other cards on AMIGA begin similar and suffering from the very same problem.
Who can name one?
 


Christer Sigfrids

Posts 11
15 Sep 2020 17:54


I am surprised gunnar didn't know how to google it. the TF520 was a proof of concept card with no fastmem on it. just to see if it worked

Simple as that. Card talks to chipmem and maybe some 512kb slowmem.

530, is way much faster with fastmem. gayle emulation is fast too. 536 has if wanted buffered ide port too.

Terriblefire is aimed to be working stable cheap ass as possible accellerators.

Look at sysinfo speeds from aca620 or furia 020 for 020 performance with similiar memoryspeeds as 520 would have had if it have had fastmem on-board.020 more or less matches a 68000 cpu without fastmem.


Samuel Crow

Posts 424
15 Sep 2020 18:32


Christer sigfrids wrote:

I am surprised gunnar didn't know how to google it. the TF520 was a proof of concept card with no fastmem on it. just to see if it worked
 

Check the post dates on the thread you just necro-ed.  It's from January.  A lot has happened since then.


Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
15 Sep 2020 19:12


Markus B wrote:
 
  This was stated quite some time ago.
  I think only Vojin refused to accept that fact.

I just have never seen it such clearly stated (as Gunnar stated here).
And liked idea of "V4 for all" as it was in original announcement.

Fully accepted. Plans do change, as in personal life, more in business.

Also I feel sorry everything gets cloned nowadays. So bad to see another Amiga card out of production (Terrible Fire if I get that right).




Christer Sigfrids

Posts 11
15 Sep 2020 21:36



 
 
Samuel Crow wrote:

 
Christer sigfrids wrote:

  I am surprised gunnar didn't know how to google it. the TF520 was a proof of concept card with no fastmem on it. just to see if it worked
   

  Check the post dates on the thread you just necro-ed.  It's from January.  A lot has happened since then.
 

 
 
  You sad what ?
 
  Specificuly, the tf520 was made years ago.
  Only cpu a cpld for handle the bus. And a few caps to make it stable enough.
 
  From that day it was made only that way. There is no tf520 revided with memory or ide port.
 
  Soo Stephen steped up made tf530 rev 3 f.ex. is 030 50mhz + fpu/mmu. Gave it an ide port and 2 mb of fastmem. With kickstart in fastmem. Gayle emulation is fast. 8-10 mb/sec. 030 make around 10 mips sysinfo mips as long as its has fastmem. Give this tf520 z2 mem and it will go slightly faster.
 
 
 
 
  First of all tf520 is a board designed as proof of concept. I dont know why he steped up so fast to 030 but he did. There is also a tf540. Never released. Maybe ide was developed on that board.
 
  If you want to compare vampire with tf. Do it with atleast tf530. Tf534 or tf 536. Becouse those has fastmem. Tf534 and 536 is fully 32bit fastmem.


Christer Sigfrids

Posts 11
16 Sep 2020 00:38


Samuel crow. I dont know when tf520 was built. But i began to build my own tf530 @50 mhz around mid 2017. It makes 10.x mips and around 6.5 mb ide speed. Fpu 68882 is present too.
 
  I still got the pictures when i showed some friends.
 
  Soo tf520 is before that 2016 2015 ?
  Im pretty sure tf530 beats bluzzard 1230 in sysinfo mips. As an example. Iirc you get just above sysinfo 9 mips on a b1230. I guess speed is gain in the faster 2mb sram or i actualy dont know what type of ram it is. But 5 voltage not old dram.
 
  You sad alot since jan 2020 has happend.yep I would say alot has happend since 5 years ago.
 
  Its harder to look at the facts now. Becouse Stephen recently deleted his files and YouTube Chanel. But i beleave talks about tf5xx etc is documented on eab thanks to the administration who refuses to delete a whole thread.
 
 


Christer Sigfrids

Posts 11
16 Sep 2020 00:57


Just some last words for tonight, Since you all didnt have a clue until recent atleast. Tf520 is only a full 68020 cpu. Which means full 32 bit. Unlike ec020 which is 24bit adress. But that does not matters that much. Tf520 is talking to slowest mem on amiga. Best case scenario add a sidecar hdd with some z2 fastmem to speed it a little bit.

As you can see sysinfo reported the wrong cpu the reason for this is becouse cpu cant be detected. Soo sysinfo will have to take a guess. I dont know how. Without fastmem it might report it as EC020. 060 can be detected like even tell you what rev it is. Thnx. I only try to explain why tf520 is that slow. I gave you all the answer. I dont know if frequences above is accurate even. 34mhz sounds like a weird frequence. But maybe its 33 or 25 mhz osc.

posts 72page  1 2 3 4