New eFlasher available

[mboehmer_e3b]

Ah. Finally. Let's train some rats...

Only 2.8MB/s with Algor?

Well-designed Zorro II cards (e.g. Picasso IV) can easily reach transfers much over 3MB/s. Not to mention Zorro II cards supporting Fast Zorro II transfers in ZIV boards.

Lection I: synchronous busses. Zorro II is a purely synchronous bus. This means, each bus transaction is based on a fixed clock (in Zorro II PAL machines, this is 7.14MHz). Each bus access is 4 clocks long. You cannot get faster that the clock given by the master. Picasso IV is not working as bus master in Zorro II modus.
So, enlighten me...

Lection II:
Zorro IV modus. It is not "Zorro". It's some patchwork done by Elbox (by doubling the bus clock, AFAIK). Unfortunately not even Elbox itself is supporting this patch anymore.

Lection III:
Speed. Reading 1MByte (or 8MBit as Elbox is proud to say) in Zorro II mode takes 1/3 second, in Zorro III mode is claimed 1/16s. Wuha. Add the mandatory reset time needed by the eFlash, and compare again. Or leave it out - can you distinguish between both times without a test program ?

It follows clearly from the text that the point is about Flash Zorro II cards, that is that the description refers to KickFlash cards. I saw such a card. Its producer went so far in savings in

Did you just "see" it or backward engineer it ?

production costs that both Flash memory modules in the card are connected to the same eight data lines of the Zorro II connector. Thus he saved some 0.50 EUR on PCB costs... and at the same time cut Zorro II reading speed by half.

Wow. I'm impressed. You don't know a thing about synchronous busses and are able to draw such conclusions ?

I also have only positive impressions from using Elbox products, which I have many.

It is normal for someone working at Elbox to have their products...

[mboehmer_e3b]

Platon, the Delfina Flipper uses a trick to do 5,2 MB/sec. So saying that full Zorro II speed is 2.8 MB sec is slightly incorrect.

The trick you mention is to use address lines as additional write lines (hey, people did this on the Atari ROM slot once a while . This works in only one direction, which is writing. For reading, there is no way of doing such a hack, as address lines are always set by the busmaster (in this case, the CPU).
For Delfina this is a very nice way of enhancing the write throughput for playing back sound.

So Platon is not wrong here, talking about read speed.

A remark on Zorro II bus speed: you have 7.14MHz clock, with four clocks building up a standard bus cycle. There is no block transfer mode in Zorro II, so add one clock to divide the bus accesses. You end up with five clock cycles of 140ns each => you can transfer a 16bit word in 700ns (assuming that nobody else wants access to the bus).

Here we go: (5x 140ns)^(-1) x 2bytes = 2.72 MB/s

(for rats: ignore it, may contain too many facts)

[mboehmer_e3b]

You speak exactly like Mr. Bush a year ago.

Sigh. Wasn't Poland in the "Coalition of the Willing" ? Didn't Poland get its own sector in Iraq ? Come on, guy, Mr. Bush is your big friend.

Anyhow, this is a little bit offtopic, so take it as a comment, nothing else.

[mboehmer_e3b]

A well designed Zorro II card achieves over 3MB/s in such backplanes.

Did you benchmark the ALGOR in such a non-Commodore backplane ? No, I'd guess.
You know, a well designed card doesn't care about such things, it just works at the speed given by the bus system. Like the ALGOR does.

I do not work in Elbox. I do not know if anyone from Elbox uses Amiga at home.

Fore sure they do. They all are working on Shark PPCs with OS4  :lol:
 
Always a please to read your comments here. Gives me a laugh or two  :-P

Michael

[mboehmer_e3b]

Ah... Rat's back...

You may wrongly understood what your logic analyzer showed or you do not understand what we are talking about here.

Funny you are. Ever worked with a logic analyzer? Ever built any hardware yourself?
No, I'd guess. And now you start pissing at somebody's foot who is in hardware business for years now.

BTW, if talking about bandwith... busses usually are shared, so you won't be able to do bus hogging for too long, which means that you normally get about 80% of bus bandwidth in ideal cases.

Mediator 1200 must use banking due to the design of turbo cards for A1200.

Even if there were 32 address lines, you would have to do banking, as PCI address space itself is 4GB, and 32 address lines are exactly 4GB.
The Amiga itself needs some of this address space.
AFAIK, the Mediator is mapped into the system as Zorro card, and the size of Zorro cards is limited again (Zorro II: 8MB, Zorro III: 1GB).

Michael

[mboehmer_e3b]

Ah... Rat again...

Algor FlashROM is available ONLY in the processor data space?

Another proof of your knowledge... how do you access memory without address lines ? Good god...

If any company tried influence me, well, Michael Boehmer from E3B tried to frighten me in his private email to make me change my words on Algor. I do not want to listen to him. I always try to base my statements on verifiable facts.

Well... nice try. But unfortunately you always fail with your attempts to give any proofs for your statements.

What you refer about was simple (you may look in the corresponding mail): if you think you found a mistake on a web page, contact the webmaster of the site. You didn't. You started spreading public accusations on mailing lists.

Are you planning an attack? Bombing them or hitting a plane in their building?

I don't think such dramatic action is foreseen. And I don't think it is necessary  :-) .

What I could think here is that with Poland joining the European Community same laws apply to all of us. Laws regulating unfair competition, dealing with advertisements and warranty issues.
Some cheated customers might just be enough.

Are you from Al-Qaida?

Are you from Al-Buggs  :lol: ?

Michael