NVMe SSD designed for audiophiles

[yamamoto2002]

IIRC Luke (the tall man on the right) is audiophile, he has a decent audio on his house.

 

[yamamoto2002]

11 hours ago, AudioDoctor said:

 

Oh, so the only thing that is glass is the fiber optic network connection, not the SSD itself? That makes a lot more sense.

 

IIRC I/O port of current gen NVMEoF Ethernet SSD is copper based: a protocol converter on the SSD speaks RDMA UDP protocol over its 25Gbe copper ethernet interconnect and Copper to Optical Media converter is needed to convert it to fibre optic. In the near future, I hope it will be replaced with direct silicon photonics optical protocol converter (for lower power consumption).

 Source: https://www.servethehome.com/marvell-88ss5000-nvmeof-ssd-controller-shown-with-toshiba-bics/

 

In the photo, there are two SFP28 ports. most certainly it is for redundancy like SAS drives

[yamamoto2002]

One concern is power consumption and need for high CFM fan ... Behold the counter rotating fan of Ingrasys enclosure, its sound is like a jet engine. I have a enterprise SAS12G SSD and it quickly becomes 70 ℃ and thermal shutdown with no airflow.

[yamamoto2002]

I read somewhere on the internet, in early 2000s, HDD manufacturer actually listened the requests some requirements from audio visual industries, more specifically, from manufacturers of video recorder, and produced purpose built firmware for them! This special firmware do following things:

 

When data read/write CRC error happens, conventional HDD firmware retry several times to read/write the same sector until data is read/write successfully. But for video recorder use-case, user experience is better to read/write wrong data on one sector (one frame of image is corrupted) and continue to record/play, than retrying for 10 seconds to r/w one sector of data and 10 seconds of video is lost/stopped

 

and this kind of special firmware is used until today, as special HDD model for surveillance video recorders

[yamamoto2002]

I searched a bit, enterprise HDD firmware has “error recovery control” feature called TLER or CCTL to give up r/w retry early and let RAID controller to handle the error. consumer HDD do not have this feature and array is degraded when retry takes 10 seconds. I guess video recorder firmware is developed from this knowledge? on video recorder typical use case scenario, when recording video A, simultaneously, several hours ago recorded video B is played, and if read B retry takes 10 seconds, recording A is failed. video HDD should prioritize write A operation somehow

 

from https://en.wikipedia.org/wiki/Error_recovery_control

Quote

Modern hard drives feature an ability to recover from some read/write errors by internally remapping sectors and performing other forms of self-test and recovery. The process for this can sometimes take several seconds or (under heavy usage) minutes, during which time the drive is unresponsive.

 

Also found interview article with HGST engineers, they do not develop special firmware for audio industries at all, on circa 2013