04-08-2013 09:40 PM
I still uses ISE 14.2 .
I found out that Device speed data version is : "PRELIMINARY 1.06 2012-07-14" at PAR report.
I think more latest version can be exist, because the version is preliminary and
ISE tool was updated after 2012-07-14.
Is there no updated device speed data for virtex7 vx485t ?
Thanks for your attention.
04-09-2013 08:34 AM
Often speed files remain with the original designation forever, as there is nothing new. Nothing wrong with it.
04-09-2013 08:59 AM
Hey Austin / skyseanu0,
I cracked open 14.4 and do see an updated Speedfile:
Version identification for speed file is: PRODUCTION 1.08 2012-12-17
There seem to be about 27 differences.
It may be worth updating your tools.
04-09-2013 09:13 AM
Interesting. If they update to a newer version of the tools, that may be disruptive to their project. As long as they have a working design, and they have invested time testing it, I wouldn't recommend changing anything. Unless we have advisories on why they should/must use a newer speed file, it may be more of a problem, than a solution.
The 27 differences, out of thousands of values, may be just 'fine tuning' and tweaking. The changes may be entirely unable to be seen by this custome on their design.
I know our standard line is "use the latest tools" but the real world doesn't work that way. Almost all companies freeze their development tools (place them under engineering change control). No one could, or should, change tools in the middle of a project, unless they have an overwhelming reason why they must change tools. Knowing how real company engineering development works, we may recommend a newer version of the tools, but unless the customer is just starting a project, that will not be a possibility. I know FAE's often visit customers to 'patch' their environments to provide needed specific fixes to their environments, but changing the tools wholesale after the project is in process is a huge expense (and pain the neck). Changing tools invalidates all previous testing, for example (you need to start over testing).
04-09-2013 09:19 AM
Yes it doesn't look like a major issue. From a quick glance the changes seem minor and should make timing easier to meet.
04-09-2013 09:30 AM
That is what I thought. A long time ago (~10-12 years) I had a 'discussion' about changing a speed file with a Xilinx Vice-President...
I was in IC design, and I was all excited about 'fixing' the speed file...
I was told (in no uncertain terms) that we can make the speed file better in subsequent releases, but to make it worse was totally unacceptable. I got it. It is all about making it easy for the customer, and not placing traps for them to stumble into.
So, in the verification and characterization, we may refine and upgrade the speed file, but is a gloomy day (for those responsible) if we have to go the "wrong way" with a timing number!
The speed file methodology has improved greatly over the years, and we intentionally start out safer than we could possibly imagine the numbers ever being in reality. Then we go back and tweak them after verification and characterization for subsequent releases.
04-09-2013 09:35 AM
For Production parts and Production timing closure designers must use the Production speed file that is documented in the datasheet for each family. For the 7VX485T device that the OP is using that is DS183, Table 14 which documents that speedfile version 1.09 is the one that must be used.
Since the OP is currently using speed file version 1.06 they must upgrade.
04-09-2013 09:40 AM
Thanks for stating the company line so clearly.
For all the reasons I have mentioned, do not be surprised if they ignore you (as most companies do).
I have been told that typical verification ofa system may run into the hundres of thousand of dollars in costs (or more), and we may say "upgrade" and they will say "pay me."
We come back and say "oh, never mind."
04-11-2013 04:08 AM
Thank you for your comment.
At this time, It look not to be updated.
Later, I'll update the ISE to the latest version.