MIG for Linux?

So I thought I would try out this MIG thing I see mentioned 
occasionally, But according to
http://www.xilinx.com/xlnx/xil_ans_display.jsp?BV_UseBVCookie=yes&getPagePath=25406

Software Requirements

- ISE 9.2.01i
- Windows XP (32 bit)

So is MIG really windows only?

I currently have ISE 8.2. That same page says:
- MIG is no longer provided as a separate download, but is now 
incorporated into IP Updates. MIG v1.73 is available through 9.2i IP 
Update 1.

So does anyone know if older versions of MIG are available somewhere (I 
have not been able to find them)? Does it run under Wine maybe? Is it 
worth bothering with (I am specifically interested in a plain DDR 
interface)?

hei duane,

i had the same problem a couple of weeks before, but in the end i
booted my winXP image in VMware generated the core and the used the
generated design under linux.

as far as i know you have to have the latest IP upgrade installed,
especially for the DDR controller , but MIG only works under winXP!

i would not recommend to try it with wine, cause if its running under
wine, you still have to install the hole webpack 9,2, and it needs
lots of diskspace.

but anyway i did not get the ddr controller core working, cause the [O|
I]DDR entity's used for data capture produced Xs in the calibration
process and those Xs propagated throw the hole core!
well it was the first time i used to try working with DDR, so it might
be an error or some wrong constraints by me..
i found a thread with people having the same problem, but in german:
http://www.mikrocontroller.net/topic/77320
and here my problem report:
http://groups.google.com/group/comp.arch.fpga/browse_thread/thread/d3c251b6a93f24d0/b52bcc700cf58537?lnk=gst&q=ddr#b52bcc700cf58537

if you get it to work please let me know!

mfg
paul


Duane Clark wrote:
> So I thought I would try out this MIG thing I see mentioned
> occasionally, But according to
> http://www.xilinx.com/xlnx/xil_ans_display.jsp?BV_UseBVCookie=yes&getPagePath=25406
>
> Software Requirements
>
> - ISE 9.2.01i
> - Windows XP (32 bit)
>
> So is MIG really windows only?
>
> I currently have ISE 8.2. That same page says:
> - MIG is no longer provided as a separate download, but is now
> incorporated into IP Updates. MIG v1.73 is available through 9.2i IP
> Update 1.
>
> So does anyone know if older versions of MIG are available somewhere (I
> have not been able to find them)? Does it run under Wine maybe? Is it
> worth bothering with (I am specifically interested in a plain DDR
> interface)?

pwie42 wrote:
> hei duane,
> 
> i had the same problem a couple of weeks before, but in the end i
> booted my winXP image in VMware generated the core and the used the
> generated design under linux.
> 
> as far as i know you have to have the latest IP upgrade installed,
> especially for the DDR controller , but MIG only works under winXP!

Well... I don't have WinXP (or Vista for that matter) so that is not an 
option for me. Okay, technically I guess I paid for them, but I never 
booted or used them. I am not going to bother with VMware/Windows. I do 
have Win2000 around that I can boot if absolutely necessary.

> i would not recommend to try it with wine, cause if its running under
> wine, you still have to install the hole webpack 9,2, and it needs
> lots of diskspace.

Diskspace is cheap and I have plenty of that, so I might give it a try 
sometime. I found a tool from Xilin called mig007_rel6, which is used 
for DDR designs on V2P (yes I am using V2P devices, and I know the 
current "real" MIG does not support them). I guess this is an early 
incarnation of MIG, though the exact relationship is a bit cryptic. It 
partially works under Wine, but doesn't generate the output files, and 
is prone to hangs. But it seems to work under Win2000, so I guess I'll 
use that for now.

> but anyway i did not get the ddr controller core working, cause the [O|
> I]DDR entity's used for data capture produced Xs in the calibration
> process and those Xs propagated throw the hole core!
> well it was the first time i used to try working with DDR, so it might
> be an error or some wrong constraints by me..
> i found a thread with people having the same problem, but in german:
> http://www.mikrocontroller.net/topic/77320
> and here my problem report:
> http://groups.google.com/group/comp.arch.fpga/browse_thread/thread/d3c251b6a93f24d0/b52bcc700cf58537?lnk=gst&q=ddr#b52bcc700cf58537

Well, it has been a long time since I bothered with simulating a back 
annotated design. I am not sure it is worth the trouble.

I have found the documentation for the design generated by this version 
of mig to be a bit cryptic, and the generated design/testbench doesn't 
seem to work as is, at least not in a functional simulation. I think the 
problem is that without actual LUT/routing delays for the DQS delay 
lines, the simulation won't work, though I am still checking into that.

Duane Clark wrote:
> ...
> I have found the documentation for the design generated by this version 
> of mig to be a bit cryptic, and the generated design/testbench doesn't 
> seem to work as is, at least not in a functional simulation. I think the 
> problem is that without actual LUT/routing delays for the DQS delay 
> lines, the simulation won't work, though I am still checking into that.

And indeed, adding a small bit of delay to the dqs_delay entity does fix 
the simulation. Since synthesis tools will ignore delays, it is a bit of 
a mystery why this was not already included, and how the simulation 
could possibly have worked without it.

The newest version of MIG will support 32-bit Linux Red Hat Enterprise
4.0. The newest version (2.0) will be in IP Update 2 for ISE 9.2i and
will be released on 10/17/2007.

What device are you targeting?

jacobusn@xilinx.com wrote:
> The newest version of MIG will support 32-bit Linux Red Hat Enterprise
> 4.0. The newest version (2.0) will be in IP Update 2 for ISE 9.2i and
> will be released on 10/17/2007.

That is good to hear.

> What device are you targeting?

My immediate use was for a V2Pro device, so I probably will have to 
stick to the old tool for that. But I was also looking ahead toward 
using the V5, and hoping the design doesn't need to change much in the 
migration.

Duane Clark wrote:
> jacobusn@xilinx.com wrote:
>> The newest version of MIG will support 32-bit Linux Red Hat Enterprise
>> 4.0. The newest version (2.0) will be in IP Update 2 for ISE 9.2i and
>> will be released on 10/17/2007.
> 
> That is good to hear.
> 
>> What device are you targeting?
> 
> My immediate use was for a V2Pro device, so I probably will have to 
> stick to the old tool for that...

Just for the sake of completeness, I discovered that in the user 
interface generated by that tool, some signals are clocked on the system 
clock, some on a 90 degree phase shifted clock, and some on a 180 degree 
phase shifted clock. Yikes! So much for an "easy" interface. So I guess 
I won't be using that.

On Oct 15, 2:09 pm, "jacob...@xilinx.com" <naude.j...@gmail.com>
wrote:
> The newest version ofMIGwill support 32-bitLinuxRed Hat Enterprise
> 4.0. The newest version (2.0) will be in IP Update 2 for ISE 9.2i and
> will be released on 10/17/2007.
>
> What device are you targeting?


That's very nice!

currently XC3S1600E for me...

Regards
Sandro