Overview Features Instructions Performance Forum Downloads Products OrderV4 Reseller Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Performance and Benchmark Results!

V1200 Updatepage  1 2 

Gunnar von Boehn
(Apollo Team Member)
Posts 4848
27 May 2020 12:57



Igor did some nice work on his excellent V1200 CPU accelerator card.
The upcoming new Core release of the V1200 will improve the chip bus speed nicely:

Here some preview of the result:

Well done Igor!


Adam A

Posts 101
27 May 2020 13:13


Nice job Igor.

v1200 is the most powerful Amiga accelerator. indeed.


Amiga 4Life

Posts 80
27 May 2020 13:43


Nice, a1200s are coming back to the desktop...


Jean-Baptiste Bolcato

Posts 53
27 May 2020 14:06


Wow. Very nice!!

But bit unclear on how much faster it actually is?
- VS the previous core on V1200?
- VS the V500 2+ or V600?

...can someone run the same bustest and post here maybe? Thanks!

Also, down the line, can all V500/V600 Appollo cards benefit from that speed boost in chip bus speed? (same FPGA, right?)
JBB


Gunnar von Boehn
(Apollo Team Member)
Posts 4848
27 May 2020 14:15


Jean-Baptiste Bolcato wrote:

  Also, down the line, can all V500/V600 Appollo cards benefit from that speed boost in chip bus speed? (same FPGA, right?)
 

 
The A4000 and A1200 have 32bit bit chipmemory.
While the A1000,A500,A2000, and A600 have 16bit chip memory.
 
The maximum speed you can technically ever reach on the A1200 and A4000 with their 32bit memory is a little bit 7 MB/sec
The 16bit Amiga can at maximum reach a little bit over 3.5 MB/sec.
 
Not every CPU card model is able to reach those numbers.
Some cards reach noticeable less than the maximum.
You can check how good the bus interface of the card is, to see whether it reaches 7 MB/sec for READ on the A1200.

The improvement of this release is that the V1200 is now at the maximum possible on the V1200.
 
The V500 and V600 are already on the maximum possible for those machines. So there is no more improvement possible.
 
 


Andy Hearn

Posts 318
27 May 2020 15:02


awesome stuff! really looking forward to seeing this for myself in the flesh :)

i would do some comparison numbers just so we can all see, but my 500 and 600 are out of action at the moment... i really need to sort that out...


Jean-Baptiste Bolcato

Posts 53
27 May 2020 16:27


Thanks for clarification Gunnar!
Could be a very nice extra selling point for the V1200 for the ones hesitating between cards...


Michael AMike

Posts 134
27 May 2020 20:39


Impressive boost - never seen those numbers before.


Smartroad 78

Posts 69
28 May 2020 07:54


Stop making the wait for my V1200 purchase invite more painful!! Lol


Gunnar von Boehn
(Apollo Team Member)
Posts 4848
28 May 2020 16:26



 
We improved the bus access more.
No CPU card ever reached these results on any AMIGA.


Ronnie Beck
(Apollo Team Member)
Posts 182
28 May 2020 18:05


That is some seriously impressive stuff.  I guess we are at the limit of what the old chip ram can offer now.

Time to hit some old school demos and enjoy the new performance boost. :-)


Knight Stone

Posts 117
29 May 2020 01:30


Will this address any of the Vampires IDE issues? or will there be a version for those who want to use Vampire IDE ?

thanks


Gunnar von Boehn
(Apollo Team Member)
Posts 4848
29 May 2020 07:00


The next release update for the V1200 will be a "stable x12" core.
 
As teaser here how a x14 looks like:



Andy Hearn

Posts 318
29 May 2020 09:54


COR BLIMEY!
COME ON MR POSTMAN!!!

question. the V1200, is the system clock 14Mhz, so you're doing a  divide by 2 and then do the x10 x11 x12... x14 multiplier on that? or are you doing a 5x 5.5x 6x... 7x multiplier on the 14Mhz  standard base clock?


Gunnar von Boehn
(Apollo Team Member)
Posts 4848
29 May 2020 10:30


Knight Stone wrote:

or will there be a version for those who want to use Vampire IDE ?

This update will be an improvement to the last official core release.
The last official core release supports PCMCIA and IDE on mobo.

This update does include one big improvement.
It maximizes the Amiga bus controller.
Nothing more.

I'm a big fan of making 1 thing really good and then to make a release.
And then tackle in the next release the next big item.




Gunnar von Boehn
(Apollo Team Member)
Posts 4848
29 May 2020 10:33


Andy Hearn wrote:

COR BLIMEY!
  COME ON MR POSTMAN!!!
 
  question. the V1200, is the system clock 14Mhz, so you're doing a  divide by 2 and then do the x10 x11 x12... x14 multiplier on that? or are you doing a 5x 5.5x 6x... 7x multiplier on the 14Mhz  standard base clock?

The X-factor refers to the 7Mhz original clock of the AMIGA 1000/500.

This means the CPU of an "x12" core runs 85 MHz and the CPU of an "x14" core runs at 100 MHz.
An x12 core runs at the same clock on all cards: V500,V4,V1200 




Knight Stone

Posts 117
29 May 2020 12:12


Gunnar von Boehn wrote:

 
  I'm a big fan of making 1 thing really good and then to make a release.
  And then tackle in the next release the next big item.
 
 

I'm a big fan of having the basic hardware actually working first, before squeezing performance out of it.


Andy Hearn

Posts 318
29 May 2020 12:25


yup, i get the 7mhz equivalent :) but i was wondering what you're physically getting from the A1200 in clock signal, and then what you're doing to it to generate your "Xnn" *equivalent* clock speed?

anyway no worries. just idle thoughts :)

additional idle thoughts that i found interesting:-
i was looking over the sonnet PPC stuff as i have A4k+mediator etc. and i respect the time and energy these guys spend on this. But i've always found PPC stuff a monumental pain to deal with and get working. but anyway.
A Sonnet card loaded with 256 meg ram, a PPC 7410 CPU - G4 @450Mhz generates the following numbers in amiga SysSpeed
Mips:- 516.14
MFlops:- 400.89

With your Sysinfo numbers, a Vampire1200@100Mhz
Mips:- nearly 200
MFlops:- nearly 100

take that for what you will. but to my mind, the PPC doesn't really win here at all.


Igor Majstorovic
(Apollo Team Member)
Posts 343
29 May 2020 13:02


@Andy Hearn
Like BigGun explained we base our X factor of multiplication on 7MHz signal. In reality X12 is 7MHzX12=84MHz for A500/600. For A1200 we say X12 but in fact it is 14MHzX6=84MHz.
  On V1200 clock we are getting clock from Amiga motherboard is first polished by RC filter then connected to PLL input of FPGA. Using that PLL we create several clocks inside FPGA who drives core, memory and other devices on the card.


Andy Hearn

Posts 318
29 May 2020 16:01


perfect, many thanks guys :)

posts 32page  1 2