Bitcoin Forum

Other => CPU/GPU Bitcoin mining hardware => Topic started by: intron on May 22, 2013, 06:17:39 AM



Title: - Avalon datasheet -
Post by: intron on May 22, 2013, 06:17:39 AM
Hi,

Read the datasheet of the Avalon 'A3256-Q48 SHA256 Processor'
and to me there are few things not really clear about this
REPORT_x pins. They seem open drain outputs, there are
pull-ups required, so it looks like the REPORT_x lines of all
Avalon chips can be tied togeter to form a party-line. This
makes sense as it saves a lot of pins on the microcontroller.

But if so, how are contentions resolved when two or more
chips have found a golden nonce at the same time?

Any ideas anyone?

intron


Title: Re: - Avalon datasheet -
Post by: mgio on May 22, 2013, 06:37:33 AM
Hi,

Read the datasheet of the Avalon 'A3256-Q48 SHA256 Processor'
and to me there are few things not really clear about this
REPORT_x pins. They seem open drain outputs, there are
pull-ups required, so it looks like the REPORT_x lines of all
Avalon chips can be tied togeter to form a party-line. This
makes sense as it saves a lot of pins on the microcontroller.

But if so, how are contentions resolved when two or more
chips have found a golden nonce at the same time?

Any ideas anyone?

intron



I haven't looked at the datasheet myself yet, but maybe they figure that won't be a problem since the chances of that happening are extremely slim.


Title: Re: - Avalon datasheet -
Post by: intron on May 22, 2013, 06:41:42 AM
Hi,

Read the datasheet of the Avalon 'A3256-Q48 SHA256 Processor'
and to me there are few things not really clear about this
REPORT_x pins. They seem open drain outputs, there are
pull-ups required, so it looks like the REPORT_x lines of all
Avalon chips can be tied togeter to form a party-line. This
makes sense as it saves a lot of pins on the microcontroller.

But if so, how are contentions resolved when two or more
chips have found a golden nonce at the same time?

Any ideas anyone?

intron



I haven't looked at the datasheet myself yet, but maybe they figure that won't be a problem since the chances of that happening are extremely slim.

Could very well be that way. I'll try to calculate the changes
of a collision and thus lost data.

intron