Lattice Diamond and VHDL-2008

[rickman]

I don't have any trouble getting the simulation (Active HDL) or
synthesis (Synplify) tools to work with VHDL-2008, but the Lattice tool
itself doesn't seem to understand it. When Diamond analyzes the source
files it complains of syntax errors. The rest of the tool seems to work
just fine and this doesn't stop me from completing the project.

I did an Internet search and found a post at eevblog.com about this from
last year with no response. I put in a ticket to Lattice. I am using a
slightly old version of the tool, 3.3 while the latest is 3.7 I believe.
I hope that's not my problem. lol

--

Rick C

 

[rickman]

On 6/16/2016 11:23 PM, rickman wrote:

I don't have any trouble getting the simulation (Active HDL) or
synthesis (Synplify) tools to work with VHDL-2008, but the Lattice tool
itself doesn't seem to understand it. When Diamond analyzes the source
files it complains of syntax errors. The rest of the tool seems to work
just fine and this doesn't stop me from completing the project.

I did an Internet search and found a post at eevblog.com about this from
last year with no response. I put in a ticket to Lattice. I am using a
slightly old version of the tool, 3.3 while the latest is 3.7 I believe.
I hope that's not my problem. lol

Wow! This has been dragging out for a bit. They replied saying the LSE
tool can be set for VHDL-2008. I wrote back that I don't have a setting
for LSE. They sent me a screen shot showing the setting in a dialog
box. I sent back a screen shot showing the same dialog box in my tool
that doesn't have the LSE setting. I've downloaded the most recent
version of the tools and installed them before doing this.

Regardless, I'm not sure that LSE is the right setting. That's an
alternative synthesis tool, right, "Lattice Synthesis Engine"? Is that
what they use to parse the source when the design is "refreshed"?

Meanwhile, I also asked if the bitstream files were the same for the
LFXP3E-3TN100C and LFXP3C-3TN100C and they said I should re-generate the
bitstream. The only difference should be that one uses an external 1.2
volt core supply and the other has internal LDOs to work off 3.3 volts.
Would that really result in any difference in the bitstream? I don't
want to have to regenerate the bitstream since it would require
re-qualification. I'm tempted to try it. I don't have much confidence
in the support.

--

Rick C

 

[GaborSzakacs]

rickman wrote:

Meanwhile, I also asked if the bitstream files were the same for the
LFXP3E-3TN100C and LFXP3C-3TN100C and they said I should re-generate the
bitstream. The only difference should be that one uses an external 1.2
volt core supply and the other has internal LDOs to work off 3.3 volts.
Would that really result in any difference in the bitstream? I don't
want to have to regenerate the bitstream since it would require
re-qualification. I'm tempted to try it. I don't have much confidence
in the support.

It's certainly worth a try, but if Lattice is anything like Xilinx,
there would be at a minimum a different device ID and the programming
tools would detect the difference and refuse to program the part.

--
Gabor

 

[rickman]

On 7/8/2016 9:57 AM, GaborSzakacs wrote:

rickman wrote:

Meanwhile, I also asked if the bitstream files were the same for the
LFXP3E-3TN100C and LFXP3C-3TN100C and they said I should re-generate
the bitstream. The only difference should be that one uses an
external 1.2 volt core supply and the other has internal LDOs to work
off 3.3 volts. Would that really result in any difference in the
bitstream? I don't want to have to regenerate the bitstream since it
would require re-qualification. I'm tempted to try it. I don't have
much confidence in the support.


It's certainly worth a try, but if Lattice is anything like Xilinx,
there would be at a minimum a different device ID and the programming
tools would detect the difference and refuse to program the part.

I talked to a local FAE and he said he thought they used the same die
with different bond out. I can't get support to acknowledge they use
the same bit stream, but maybe I can get them to say it is the same die.
If so, the ID would have to be the same.

--

Rick C

 

[Brane2]

Dne petek, 08. julij 2016 20.53.43 UTC+2 je oseba rickman napisala:

On 7/8/2016 9:57 AM, GaborSzakacs wrote:
rickman wrote:

Meanwhile, I also asked if the bitstream files were the same for the
LFXP3E-3TN100C and LFXP3C-3TN100C and they said I should re-generate
the bitstream. The only difference should be that one uses an
external 1.2 volt core supply and the other has internal LDOs to work
off 3.3 volts. Would that really result in any difference in the
bitstream? I don't want to have to regenerate the bitstream since it
would require re-qualification. I'm tempted to try it. I don't have
much confidence in the support.


It's certainly worth a try, but if Lattice is anything like Xilinx,
there would be at a minimum a different device ID and the programming
tools would detect the difference and refuse to program the part.

I talked to a local FAE and he said he thought they used the same die
with different bond out. I can't get support to acknowledge they use
the same bit stream, but maybe I can get them to say it is the same die.
If so, the ID would have to be the same.

--

Rick C

FWIW, I have finally solved my Breakout Board problem and so now I can check the difference between XO2-7000HC/HE, which is of the same nature:

XO2-7000HC: ID= 0x012BD043
XO2-7000HE: ID= 0x012B5043

Interestingly, I had that old basic blinking leds demo for Breakout Board, that was done for HE chip and Programmer utility refused to burn it into HC chip.

But then I made new design with same Verilog file ( just regenerated power controller with IPExpress) for HC chip to test new board and that one I managed to burn into HC as well as HE chip.
Perhaps because I have had generated jedec as well as bitstream file this time and programmer managed to use jedec...

 

[GaborSzakacs]

rickman wrote:

On 7/8/2016 9:57 AM, GaborSzakacs wrote:
rickman wrote:

Meanwhile, I also asked if the bitstream files were the same for the
LFXP3E-3TN100C and LFXP3C-3TN100C and they said I should re-generate
the bitstream. The only difference should be that one uses an
external 1.2 volt core supply and the other has internal LDOs to work
off 3.3 volts. Would that really result in any difference in the
bitstream? I don't want to have to regenerate the bitstream since it
would require re-qualification. I'm tempted to try it. I don't have
much confidence in the support.


It's certainly worth a try, but if Lattice is anything like Xilinx,
there would be at a minimum a different device ID and the programming
tools would detect the difference and refuse to program the part.

I talked to a local FAE and he said he thought they used the same die
with different bond out. I can't get support to acknowledge they use
the same bit stream, but maybe I can get them to say it is the same die.
If so, the ID would have to be the same.

The same die can have a different ID if there are bonding options that
affect the ID. Xilinx goes even farther with the Artix-7 series,
unabashedly selling the same part/package with a smaller number
of usable fabric elements. Their device ID might be affected by
some internal bonding or by OTP programming. In any case it's only
their tools that prevent you from using the "smaller" part as the
larger one. The die is identical, and the bitstreams are identical
other than the ID. If you open a design for the smaller part in
the layout view, you can see the full complement of internal
elements of the larger part. You can also place your components
anywhere in the die. The only restriction is the number of each
element type you are allowed to map into the part.

--
Gabor

 

[rickman]

On 7/11/2016 5:28 PM, GaborSzakacs wrote:

rickman wrote:
On 7/8/2016 9:57 AM, GaborSzakacs wrote:
rickman wrote:

Meanwhile, I also asked if the bitstream files were the same for the
LFXP3E-3TN100C and LFXP3C-3TN100C and they said I should re-generate
the bitstream. The only difference should be that one uses an
external 1.2 volt core supply and the other has internal LDOs to work
off 3.3 volts. Would that really result in any difference in the
bitstream? I don't want to have to regenerate the bitstream since it
would require re-qualification. I'm tempted to try it. I don't have
much confidence in the support.


It's certainly worth a try, but if Lattice is anything like Xilinx,
there would be at a minimum a different device ID and the programming
tools would detect the difference and refuse to program the part.

I talked to a local FAE and he said he thought they used the same die
with different bond out. I can't get support to acknowledge they use
the same bit stream, but maybe I can get them to say it is the same
die. If so, the ID would have to be the same.


The same die can have a different ID if there are bonding options that
affect the ID. Xilinx goes even farther with the Artix-7 series,
unabashedly selling the same part/package with a smaller number
of usable fabric elements. Their device ID might be affected by
some internal bonding or by OTP programming. In any case it's only
their tools that prevent you from using the "smaller" part as the
larger one. The die is identical, and the bitstreams are identical
other than the ID. If you open a design for the smaller part in
the layout view, you can see the full complement of internal
elements of the larger part. You can also place your components
anywhere in the die. The only restriction is the number of each
element type you are allowed to map into the part.

So much for the claim that they sell you the routing and give you the
logic or free.

--

Rick C