The Xiling Zynq 7020 has a dual-core ARM Cortex-A9 and an Artix-7 FPGA in it.

What is the equivalent part number of these chips independently? (i.e. what individual part numbers have the same fabrication method / process size so it has the same electrical properties as these two subcomponents?) Unfortunately, their documentation does not specify this information, nor was I able to procure this information from calls to them.

My company has a database of extensive testing done on many chips. I'm currently evaluating the 7020 and would like to know "if I look at the ARM part ABC123 and the Artix-7 model DEF456, the test results of those should be analogous to the sub-components of the Zynq chip."

  • \$\begingroup\$ If Xilinx won't tell you how do you expect to get an authoritative answer here? And you aren't just asking for functional equivalence but the same fab process as well? I think you need to find another approach to your problem. \$\endgroup\$ – Elliot Alderson May 8 '19 at 22:11
  • \$\begingroup\$ @ElliotAlderson I'm evaluating space-based applications and radiation testing is expensive. If I've already tested an equivalent ARM core and an equivalent Artix-7, I can avoid doing a characterization test. \$\endgroup\$ – iAdjunct May 9 '19 at 13:56
  • 1
    \$\begingroup\$ No, that is not true. Just because two devices are functionally equivalent and built on the same fab line does not mean that their response to radiation effects is the same. Subtle differences in layout that have no significant electrical or functional effect can dramatically change radiation performance. Just changing the layout of the guard rings or well contacts will do it. Starting with a wafer with a different epi thickness or doping will do it. \$\endgroup\$ – Elliot Alderson May 9 '19 at 20:01
  • 1
    \$\begingroup\$ Furthermore, the manufacturer can change any of those things and not tell you. If you really care about radiation effects you will buy an entire lot of devices and characterize that specific lot. If you see significant wafer-to-wafer variability you must characterize each wafer. There is no cheap way out. \$\endgroup\$ – Elliot Alderson May 9 '19 at 20:02

The answer is no, you cannot in general make that analogy. Mainly because the Zynq is a single chip, and the things you're trying to compare it to would be multi-chip assemblies. The integration (communication) between the ARM cores and the FPGA fabric would necessarily be very different.

The best you could do would be to compare it to an Artix-7 that has slightly more logic elements than the Zynq — that would mean that you're looking at a comparable-sized die with 90% of the same technology on it. The ARM cores and other hard-coded logic on the Zynq don't really make that much difference.

| improve this answer | |
  • \$\begingroup\$ The specific thing that I need is for the transistors to have the same electrical characteristics (capacitance, current, switching voltage, doping, etc), for the LUTs and routing matrix to behave the same way, etc. Same is true for the ARM cores: same electrical characteristics down to the silicon layer. \$\endgroup\$ – iAdjunct May 9 '19 at 13:54

Not the answer you're looking for? Browse other questions tagged or ask your own question.