From: Noel Chiappa
Sent: Tuesday, October 18, 2016 11:51 AM

>> From: Rich Alderson

>> Data was transferred via FTP over a 100baseT crossover cable connected
>> to a Slackware server; the Rabbit was able to keep up with 4 drives at
>> this speed

> Were the bits actually stored on the Slackware server, or was it just used
> to put bits on the 'drive' to start with? If the latter, what were the
> actual bits stored on? (I know, not that relevant, since this is the prior
> rev, but I'm curious.)

Stored on a big honking JBOD array (set up as RAID 5 in Linux), since an
RP06 stored as described is nigh on 900MB, and served up on that FTP link
from Slackware.

>> a Mesa 5i22 Anything I/O card (includes a Xilinx Spartan-III FPGA) that
>> plugs directly into the PCI bus in a server-class X86-64 box, and used
>> a revision of a separate driver/receiver card designed for MDE 1.0 to
>> connect to the Massbus

> Let me make sure I understand this; was there some sort of cable or
> somehow a connection from the Mesa 5i22 directly to the driver/receiver
> card, which was purely 'level conversion', with the Mesa doing the
> 'protocol' on the MASSBUS?  (I.e. they didn't communicate over the PCI
> bus?)

Yes, the d/r card is strictly level conversion, and the microcode in the
Xilinx does all the Massbus protocol.

>> a control program for the PC side which runs under Windows 2008/2012
>> Server.

> So the actual bits are stored on something (disk?) controlled by the PC?

Again using RAID 5 arrays on the PC servers, but PCI makes it a lot faster
than Ethernet.


Rich Alderson
Vintage Computing Sr. Systems Engineer
Living Computer Museum
2245 1st Avenue S
Seattle, WA 98134

Reply via email to