1. Home
  2. Field-Programmable Gate Arrays
  3. Raggedstone3 - Altera PCIe Development Board

Raggedstone3 - Altera PCIe Development Board

If you didn't see it already in our our newsletter we have a new PCIe devopment board based on an Altera Cyclone-IV GX. The new board keeps most of the mechanicals and features of our Raggedstone product range but extends the bandwidth capability of the product range. The Raggedstone3 is capabile of bandwidths exceeding 800 MBytes/s over it's X4 PCIe interface.

Initial details of this product

formatting link
Anyone at ESC this week can see the board there on our stand.

I am expecting this board to ship to customers in low numbers in June or July with a significant ramp in shipping numbers after that.

John Adair Enterpoint Ltd.

Reply to
John Adair
Loading thread data ...

formatting link

Hi John,

I have a question for you (and the group) about PCIe. Let's say I build a PCIe target in my Altera device, and let's say for the sake of argument that it's a Ethernet MAC. My Linux single board computer (SBC) has a software driver for this Ethernet device. As the Linux boots, it sees the PCIe device and loads the driver.

What happens if, without rebooting the SBC, I reconfigure the Altera part? How does the OS react to the PCIe device vanishing from the bus and then subsequently reappearing?

Thanks, Symon.

Reply to
Symon

formatting link

With the plain old PCI the PCI core needs to be configured first (memory addresses etc). It has been too long since I worked with PCI and FPGAs but IIRC the OS should probe the card before it can be used. The proper sequence is to stop the driver, reconfigure the FPGA, force probing for hardware changes and then reload the driver.

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------
Reply to
Nico Coesel

formatting link

Hi Nico, So, if the OS initiates the FPGA reconfiguration, it can also reboot the driver? Thanks, Symon

Reply to
Symon

Symon

This isn't a simple area and one of the places that the PCIe spec usually falls down. Usually if you reconfigure the device sitting the PCIe bus this is a problem because the bios and the OS don't that you are doing this. Consequently if an access is attempted during the configuration many motherboards will freeze because an answer doesn't come back. The second problem is that the enumeration and setting of the card parameters is lost. Sometimes there are work arounds to this later problem the best of which is an OS that be forced to re- enumerate the target. One way to get round the problem is to use hot plug signalling but that does depend on the motherboard and OS if that is supported.

Using a partially configuration technique can sometimes work as well but again it is not a simple path.

John Adair Enterpo> >

raggedstone3.html.

d
Reply to
John Adair

formatting link

Most modern OSses support run-time loading and unloading of device drivers. The only real problem is writing the PCI configuration space. The OS needs to support that but it is very similar to hotplugging which IIRC is supported by PCIe.

This e-mail gives some useful pointers:

formatting link 

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------
Reply to
Nico Coesel

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.