Hi The MRC training data structure is not public. We will see if we can get some information regarding differences in the training data generated on two separate training cycles. Thanks Ravi From: Sbl-devel [mailto:sbl-devel-bounces@lists.01.org] On Behalf Of Kruno Peric Sent: Tuesday, April 30, 2019 8:12 AM To: sbl-devel(a)lists.01.org Subject: Re: [Sbl-devel] MRC training parameters Some of the data in the previous attachment had extraneous bytes in the "====decompressed====" line. I've reattached the cleaned data here. From: Kruno Peric Sent: Tuesday, April 30, 2019 10:02 AM To: &#39;sbl-devel(a)lists.01.org&#39; <sbl-devel@lists.01.org<mailto:sbl-devel@lists.01.org>> Subject: MRC training parameters We ran some tests on memory training at different temperatures and found some interesting results. We reflashed the board at -20 C, +50 C and room temperature a few times, read back the flash and decompressed the MRC parameter data that was stored. We were looking to see how temperature effects the MRC training parameters. We found that the MRC data always changes between flashes, even when at the same temperature. Is there any documentation that would help with interpreting MRC training data? We are interesting in having a gauge for how stable our training parameters are across different temperatures. We would like conservative training parameter settings since we will only train during the first boot, and our product will operate across a variety of temperatures. I have attached our compressed and decompressed experiment data which was collected.