Sata performances on Skylake

[VVeri1]

We have experienced lower Sata performances on Skylake platforms (B150, H170) including Intel CRBs . Basically comparing performances with a B75 chipset with Core i5 3550 using Crystal Disk report benchmark , we have major differences especially on writes.

I have attached the two benchmark as a reference . I have already issued a support request to Intel but at present no solution

[CarlosAM_INTEL]

Hello vinver,

Thank you for contacting the Intel Embedded Community.

The information that may help you can be found at the http://www.intel.com/content/www/us/en/benchmarks/intel-product-performance.html Intel® Product Performance and http://www.intel.com/content/www/us/en/library/benchmarks.html Performance Benchmark Library web sites.

By the way, could you please clarify us if this situation happens with third party implementations or your designs? In case that they are third party, could you please give us all the information related to them?

Also, could you please tell us if this performance situation happens with an specific SATA unit or various? If it only happens with a SATA device, could you please try to reproduce this condition with different ones, made by different companies; with diverse characteristics, and inform us the results?

Thanks in advance for your cooperation.

Best Regards,

Carlos_A.

[VVeri1]

We tested boards from Asrock with B150 chipset , MSI with H170 and Intel CRB with H170. Similar results with the same HDD . Depending on the HDD drive (Toshiba or WD Green 1TB Sata drives) performances may vary slightly due to different behaviour in sequential or random writes, but with the same HDD , CPU and memory size results were similar and always much lower than B75 platform with Core i5 3550 cpu. Eve using DDR4 memory on MSI H170 vs. DDR3 on Asrock B150 didn't have any appreciable impact on Sata performances.

With the actual customer application, (rendering of a large image acquired from an MRI body scanner, and save on the HDD) time differences on the elaboration of a single file instance could show differences up to 1 minute, which is a lot. Even the boot time of the OS shows visible differences.

[CarlosAM_INTEL]

Hello vinver,

Thanks for your update.

In order to better understand your consultation, could you please give us all the information related to the Intel CRB? Please include pictures if it is possible.

Thanks again for your collaboration.

Best Regards,

Carlos_A.

[CarlosAM_INTEL]

Hello vinver,

Thanks for your update.

By the way, could you please let us know the procedure and tool to determine the mentioned condition?

Could you please try to reproduce the problem with more SATA devices from different manufacturers and let us know the outcomes?

Thanks again for your collaboration.

Best Regards,

Carlos_A

[VVeri1]

We tested Toshiba Sata HDD MK1001TRKB , WD 10EZRX with similar performances , one was slightly better on sequential read/write , the second on random read/writes , but that has to do with the drive carachteristics. We also tested a WD 1003FYBZ Raid Edition that showed an improvement given it's Enterprise carachteristics, however with this drive we were unable to reach the same performances we had with the Toshiba and WD Green drives on the DB75 platform.

In my opinion the behaviour is quite clear and has nothing to do with the drive themselves , but rather with the chipset or the BIOS of skylake based platforms. I've opened a case with Intel but after several weeks the only answer I was able to get is that there's no difference in the Sata interface between DB75 and skylake chipsets , therefore it has to be a BIOS issue. We all know that all board makers tend to use standard versions of the AMI Bios so the products we tested from different makers including the Intel CRB were pretty much similar. I don't know what else I can tell you rather than try to make a comparison for yourself if possible and you will likely come to the same conclusions

[CarlosAM_INTEL]

Hello vinver,

Thanks for your reply.

Based on your previous communication, we would like to address the following questions:

Could tell us details of the SATA GEN that your are using? By the way, how are you doing this connection? Are you using any adapter?

Could you tell us the version of driver that you are using for the Ivy Bridge Platform and for Skylake platform? Also, could you confirm that you are using the same OS in both systems, and in which mode you are using the SATA ports (HCI or RAID)?

Thanks again for your cooperation to better understand this condition.

Best Regards,

Carlos_A.