System Verification Blog

Today, someone asked in the forum what restrictions are placed on the use of FMF models.

FMF models are as open as we can make them. They are distributed as source code and are licensed under the Gnu Public License (GPL).

There is only one restriction: If you choose to sell, distribute or publish one of our models (or a derived work) you must also make the source code available to anyone at no cost.

Beyond that you may do anything you want with the models with no further obligations. Your rights to the hardware you design are not encumbered. No royalties are ever due.

The last week was busy migrating the Free Model Foundry website, and several others, to a new server.  If everything went well, you should see no difference except for slightly improved performance.  However, the migration was necessary.  The old server was running obsolete versions of the operating system, database, and several supporting packages.

Have you been clinging to any outdated tools or methodologies?  We all grow comfortable with what we know and we all dislike change.  But sometimes, we have to make the leap to something new or risk going the way of the dinosaurs.

Don’t look now but there is a mammal sneaking up on you.  Good luck in the new year.

Free Model Foundry is an open source company. Like other open source companies our strength comes from our user community. Users tell us what models or features we need to add. They also tell us how they use our models and which IC vendors we should approach with modeling proposals. But I do not see a lot of user-to-user communication.

As a source of verification knowledge and methodology, FMF is just a middle man. It is time for more direct conversations between engineers. To facilitate this, we have added a new Forum section to the FMF website. Here users can ask questions and write about experiences and leave comments for each other. I will monitor and jump in whenever there is an FMF related question appropriate for me to answer.

Now, if only there was a way to provide beer…

I was contacted a few weeks ago by a large systems house that occasionally contracts with FMF to provide models of logic components. They got their memory models from another company that they said they were quite happy with - until recently. The other company provided compiled models that only ran in a proprietary environment. The systems house had very long product life cycles so some of the parts in their simulations were at or near end of life. The company supplying the memory models decided their own business interests were best served by discontinuing these models of obsolete parts.

This left the systems house in a bind. They had long term contracts to support and upgrade fielded hardware but their model supplier had just pulled the models out from under them. What to do? (more…)

The Flash Memory Summit, August 7-9, came just two weeks after MemCon. Given that at least half of the conversations at MemCon were about flash memory, I wondered if anyone would bother to attend a second conference on flash. Well, there were plenty of attendees and exhibitors. My completely unscientific estimate is that the Flash Memory Summit was bigger in all respects than MemCon.

Eli Harari, founder of SanDisk gave the first keynote address on Wednesday morning. SanDisk is betting its future on NAND flash and Eli had numbers to back up his enthusiasm for the technology. Since NAND became chimericaly available in the 1990s, its price per megabyte has declined by a factor of 5,000. By 2012, the cost should be about one-tenth that of DRAM. However, NAND scaling is beginning to outstrip progress in lithography and may reach its limit at the 20 nanometer process node. That point we will be getting 256Gb per chip which may be enough to hold us over for a few days anyway. (more…)