Reply by Guru ●November 1, 20062006-11-01
Using Virtex4FX12, laptop P4M 1.7GHz and ISE 8.1 SP3 and EDK 8.2 SP1.
Strange, because I rebuld one MB project in 8.2 for S3ESK and it worked
fine. I think this are some bugs that emerge from time to time. Lets
pray that EDK SP2 will solve the problem.

Cheers,

Guru


Zara wrote:

> On 29 Oct 2006 09:43:10 -0800, "Guru" <ales.gorkic@email.si> wrote:
>
> >I am having about the same experience.
> >I upgraded my computer to EDK/ISE 8.2
> >
> >1. I upgraded my GSRD2 (PPC, MPMC2, TEMAC) design to v8.2. There were
> >actually no problems detected until I used the flashwriter.tcl script.
> >I does not work at all. OK, minor thing to remedy.
> >
> >2. I upgraded my PPC CoreConnect project to 8.2. First the XMD connect
> >failure (wrong processor version 0x00000000). Then I connected
> >secesfully using "connect ppc hw", but the design does not work at all
> >- probably some hw error. The sw crashes when setting interrupt
> >controller (working nice in 8.1).
> >
> >Fu.. that 8.2 and hope for the SP2  to remedy the problems.
> >
> >Cheers,
> >
> >Guru
>
> I am jus wondering if ther problem lies:
> 1. Either in the target selected (some problem with Syn/Xlat/Map/Par):
> I am using an Spartan 3: 3S400
> 2. Or in the Computer: I am using an AthlonX2, both in the dual
> 8.1/8.2 installation, and in a clean 8.2 installation on a bran new
> laptop, that also fails.
> 
> Antti? Guru? Do you see any recurrence here?
> 
> Zara
Reply by Zara ●October 30, 20062006-10-30
On 29 Oct 2006 09:43:10 -0800, "Guru" <ales.gorkic@email.si> wrote:


>I am having about the same experience.
>I upgraded my computer to EDK/ISE 8.2
>
>1. I upgraded my GSRD2 (PPC, MPMC2, TEMAC) design to v8.2. There were
>actually no problems detected until I used the flashwriter.tcl script.
>I does not work at all. OK, minor thing to remedy.
>
>2. I upgraded my PPC CoreConnect project to 8.2. First the XMD connect
>failure (wrong processor version 0x00000000). Then I connected
>secesfully using "connect ppc hw", but the design does not work at all
>- probably some hw error. The sw crashes when setting interrupt
>controller (working nice in 8.1).
>
>Fu.. that 8.2 and hope for the SP2  to remedy the problems.
>
>Cheers,
>
>Guru


I am jus wondering if ther problem lies:
1. Either in the target selected (some problem with Syn/Xlat/Map/Par):
I am using an Spartan 3: 3S400
2. Or in the Computer: I am using an AthlonX2, both in the dual
8.1/8.2 installation, and in a clean 8.2 installation on a bran new
laptop, that also fails.

Antti? Guru? Do you see any recurrence here?

Zara
Reply by Guru ●October 29, 20062006-10-29
I am having about the same experience.
I upgraded my computer to EDK/ISE 8.2

1. I upgraded my GSRD2 (PPC, MPMC2, TEMAC) design to v8.2. There were
actually no problems detected until I used the flashwriter.tcl script.
I does not work at all. OK, minor thing to remedy.

2. I upgraded my PPC CoreConnect project to 8.2. First the XMD connect
failure (wrong processor version 0x00000000). Then I connected
secesfully using "connect ppc hw", but the design does not work at all
- probably some hw error. The sw crashes when setting interrupt
controller (working nice in 8.1).

Fu.. that 8.2 and hope for the SP2  to remedy the problems.

Cheers,

Guru


Zara wrote:

> On 27 Oct 2006 00:18:43 -0700, "Antti" <Antti.Lukats@xilant.com>
> wrote:
>
> >Zara schrieb:
> >
> >> I quit.
> >>
> >> A brand new project, with only a dcm, a debug module, 16K ram, an uart
> >> and  intc will not work (problems writing BRAM).
> >>
> >> Will continue using 8.1 for a long time, I suspect
> >
> >quit? your life?
> >
> >EDK 8.2 does work
> >I have done maybe over 50 EDK systems with 8.2
> >- all of them work
> >
> >if your system doesnt there is something wrong
> >and you should figure out what it is, not quitting.
> >
> >Antti
>
>
> No, it was a figure of speech. Whta I really mean is that I will stop
> investigation while I am completing the current phase of my project.
> As soon as some 40something boarsd are delivered; I will resume
> trying.
>
> As a sapnish adage goes "Last to be lost is Hope" ;-)
> 
> Regards,
> 
> Zara
Reply by Tim ●October 27, 20062006-10-27
Zara wrote


> As a sapnish adage goes "Last to be lost is Hope" ;-)


It's Pandora. When she opened the box all the misfortunes of mankind flew 
out, leaving only Hope. Of course, you can argue that hope (9.1?) is the 
worst of all evils ;-)
Reply by Zara ●October 27, 20062006-10-27
On 27 Oct 2006 00:18:43 -0700, "Antti" <Antti.Lukats@xilant.com>
wrote:


>Zara schrieb:
>
>> I quit.
>>
>> A brand new project, with only a dcm, a debug module, 16K ram, an uart
>> and  intc will not work (problems writing BRAM).
>>
>> Will continue using 8.1 for a long time, I suspect
>
>quit? your life?
>
>EDK 8.2 does work
>I have done maybe over 50 EDK systems with 8.2
>- all of them work
>
>if your system doesnt there is something wrong
>and you should figure out what it is, not quitting.
>
>Antti



No, it was a figure of speech. Whta I really mean is that I will stop
investigation while I am completing the current phase of my project.
As soon as some 40something boarsd are delivered; I will resume
trying.

As a sapnish adage goes "Last to be lost is Hope" ;-)

Regards,

Zara
Reply by Antti ●October 27, 20062006-10-27
Zara schrieb:


> I quit.
>
> A brand new project, with only a dcm, a debug module, 16K ram, an uart
> and  intc will not work (problems writing BRAM).
>
> Will continue using 8.1 for a long time, I suspect


quit? your life?

EDK 8.2 does work
I have done maybe over 50 EDK systems with 8.2
- all of them work

if your system doesnt there is something wrong
and you should figure out what it is, not quitting.

Antti
Reply by Zara ●October 27, 20062006-10-27
I quit.

A brand new project, with only a dcm, a debug module, 16K ram, an uart
and  intc will not work (problems writing BRAM).

Will continue using 8.1 for a long time, I suspect
Reply by Zara ●October 26, 20062006-10-26
On Fri, 20 Oct 2006 16:20:04 +0200, Zara <me_zara@dea.spamcon.org>
wrote:


>On Thu, 19 Oct 2006 10:46:24 +0200, Zara <me_zara@dea.spamcon.org>
>wrote:
>
>>On Thu, 19 Oct 2006 10:33:07 +0200, Zara <me_zara@dea.spamcon.org>
>>wrote:
>>
>>>On Thu, 19 Oct 2006 09:50:03 +0200, Zara <me_zara@dea.spamcon.org>
>>>wrote:
>>>
>>>>This is not strictly a call for help, it is more of a warning to
>>>>everyone.
>>><...>
>>>
>
>Continuing with the tale (please forgive me if I strip all the
>contents from previous messages):
>
>Now I do have a double Xilinx installation (8.1 and 8.2), with 8.1
>working perfectly.
>
>If I communicate with 8.2 XMD to a core created with 8.2, everything
>is fine, so there seesm to be no problem with XMD itself (al leats
>with ParallelCable III, tests with Platfrom USB are delyed until I
>stabilize the design).
>
>When I upgrade the  design (a copy of it, of course) to 8.2, the
>following modules are automatically upgraded: OPB, OPB_MDM, LMB,
>BRAM_BLOCK, DCM, OPB_INTC. The following are not upgraded: MICROBLAZE,
>LMB_BRAM_IF_CNTRL
>

<...>
But the versions remain the same! There must be something really
strange happening, because the design in 8.1 uses 83% slices, and only
80% on 8.2

The only apparent changes on the design are the nominal upgrading os
the above mentionaed modules (without changing the version  number!)
and changin the inputs to DCM from SIGIG= DCMCLK to SIGIS=CLK.Revesing
such change gives me no benefit at all.

I will be trying some more possibilities. Should there be any
improvement, I will tell you!
Reply by Zara ●October 20, 20062006-10-20
On Thu, 19 Oct 2006 10:46:24 +0200, Zara <me_zara@dea.spamcon.org>
wrote:


>On Thu, 19 Oct 2006 10:33:07 +0200, Zara <me_zara@dea.spamcon.org>
>wrote:
>
>>On Thu, 19 Oct 2006 09:50:03 +0200, Zara <me_zara@dea.spamcon.org>
>>wrote:
>>
>>>This is not strictly a call for help, it is more of a warning to
>>>everyone.
>><...>
>>


Continuing with the tale (please forgive me if I strip all teh
contents from previous messages):

Now I do have a double Xilinx installation (8.1 and 8.2), with 8.1
working perfectly.

If I communicate with 8.2 XMD to a core created with 8.2, everything
is fine, so there seesm to be no problem with XMD itself (al leats
with ParallelCable III, tests with Platfrom USB are delyed until I
stabilize the design).

When I upgrade the  design (a copy of it, of course) to 8.2, the
following modules are automatically upgraded: OPB, OPB_MDM, LMB,
BRAM_BLOCK, DCM, OPB_INTC. The following are not upgraded: MICROBLAZE,
LMB_BRAM_IF_CNTRL

And the design stops working, with MDM blocking before ending the
inital protocol with the opb_mdm. Only once I could connect to it, and
there I saw that some randomly distributed bits of an external ram
failed to obey orders. As the connection to RAM is doen through my own
designed IP, I suppose there has been some change with OPB bus that
forces tihis failure. In fact, it seems that something may be running
a little slower and timing is failing, without warning in spite of my
putting constraints over all extern clock feedbacks.

Next chapter: Unupgrading the upgraded modules, will it be possible?
Stay connected!
Reply by Antti ●October 19, 20062006-10-19
Zara schrieb:


> On Thu, 19 Oct 2006 09:50:03 +0200, Zara <me_zara@dea.spamcon.org>
> wrote:
>
> >This is not strictly a call for help, it is more of a warning to
> >everyone.
> <...>
>
>
> UPS cable will not work with EDK 8.1. Although impact correctly uses
> it as USB21, XMD tries to use it as USB22 and finds tha JTAG chain
> devices, but not the microblaze within a FPGS. Will file a WebCase
> later, if I don=B4t find some solution. meanwhile, I will install 8,.2
> as alternative to 8.1.
>=20
> Followups will certainly be coming soon.
>=20
> Zara