It is often part of a hardware validation test suite to initiate multiple PCIe bus retrains, looking for hardware design issues, or LTSSM RTL bugs in the device under test. These test suites take a very long time to run. Is there a way to speed them up?
Testing memories on Intel platforms can be a challenge. The normal options in production are structural tests using an ICT and/or functional tests leveraging the BIOS and POST codes. The problem with the ICT approach is that it may work for soldered-down memories, but not for DIMMs since those sockets are not populated at the ICT test stage. Another weakness of ICT is that it does not test the training sequence of the memory bus. This is a more functional-type test that must be successful before you can test the memories. A problem with using the BIOS is that it’s function is to boot at all costs, so some memories might be disabled through that process, and the POST code diagnostics are very limited.
Functional test used in manufacturing production has always been something of a challenge. Often it boils down to a make-versus-buy decision. If the tests are built in-house, there is always the pressure to get it done quicker, often due to the development schedule squeezing the production test development schedule into a shorter time window. Then development time becomes a major concern. If the test is bought, these are often generic tools that need environmental tweaking or limiting of the test robustness that was originally desired. There are simply too many tradeoffs.
ASSET joined the Open Compute Project (OCP) earlier this year, and we attended our first Engineering Workshop in Dallas last week. The theme of this session was the OCP Telco Project. What did we learn?
With the rapid pace of product development today, we are all tempted to take shortcuts to get the product to market sooner than our competition. One hopes that the shortcuts taken during manufacturing test do not result in a two-million-unit recall. Some of the shortcuts don’t seem obvious or deliberate.