IDE - usb through a Spider - a good alternative to powerflyer?

[Karlos]

tjaoz wrote:

If busspeedtest in your A1200/060 reads 9MB/s, it means that your turbo card cannot access this area faster. It is the hardware limitation of your turbo card -- not limitation of Mediator.

More than once I got the kind folks here to run all kinds of custom benchmarking tools I'd written to assess the real speed at which the CPU can access VRAM on their RTG cards (many mediators there) etc.

8-10MB/s is quite normal for VRAM write speeds just about every mediator system tested, regardless of CPU / mediator system.

[Karlos]

tjaoz wrote:

Even in this thread you link to your website which encourages pirating Spider drivers, where there are notes how to crack Spider drivers. Moreover, you also offer software tools needed for pirating these drivers.

Whereas deliberately embedding malware into drivers that could trash someone's hard disk *even if they are not trying to decrypt the driver* (no memory protection in amigaos so no guarentee the trashing code cant be invoked due to an accident) is perfectly legal and moral eh?

[Karlos]

tjaoz wrote:

The $200000 - $9FFFFF area is not used by A1200 motherboard at all. So what speed turbo card accesses this area depends on the hardware design of the turbo card only.

All accesses from the turbo card to this area are directed only to the external hardware (like Zorro II, Zorro IV or Mediators) connected to the CPU slot. If external hardware acknowledges cycles immediately (the case of the Mediator WAIT jumper opened), there is obvious that the speed at which this hardware is accessed is limited by the turbo card interface hardware only.  :-D

So things like glue logic on the card, phsyical connections etc simply cease to exist eh?

[Karlos]

@tjaoz

I am curious. You seem determined that motherboard DMA is working for mediator4000. Could you provide us with a list of which products compatible with the mediator4000 use DMA to access the motherboard resources and some known examples where the feature gives a clear performance advantage over other solutions?

Presumably, DMA'ing to the motherboard memory is only useful for systems that have it. How does the feature help on systems that have local memory on accelerator cards?

Even if PCI -> motherboard memory DMA works, if you have an accelerator card in your system, the chances are you will end up having to use the CPU to shunt data between the local memory and the motherboard area accessible to the PCI.