Bit of a problem - and need to catch up

[dragnil]

Thanks for that, Trev, perhaps I'll last that long and then it will all be SSDs ...

 

[Black Hole]

SSDs are not suited to PVR use unless you disable the instant replay facility. During a trial, I wore one out in 17 months continuous use (although there is a possibility that some other problem killed it).

 

[Trev]

I don't think that the OP was advocating the use of an SSD in his current device, but suggesting that in 8 years time all storage will be of SSD type rather than spinning bits of glass.
The only advantages that I can possibly see in using an SSD in a PVR is that they are totally silent and run cooler. But as they are not the only source of heat in a PVR, there is always the noise from the cooling fan to take into consideration if you want a silent PVR.

 

[Black Hole]

I don't think that the OP was advocating the use of an SSD in his current device, but suggesting that in 8 years time all storage will be of SSD type rather than spinning bits of glass.

Indeed, but they will have to be based on a different technology than floating gates and electron tunnelling (with an inherently limited number of write cycles), such as FeRAM.

 

[Luke]

I don't think that the OP was advocating the use of an SSD in his current device, but suggesting that in 8 years time all storage will be of SSD type rather than spinning bits of glass.

That is the way I read it, but magnetic tape still has its applications as well.

 

[prpr]

The only advantages that I can possibly see in using an SSD in a PVR is that they are totally silent and run cooler. But as they are not the only source of heat in a PVR, there is always the noise from the cooling fan to take into consideration if you want a silent PVR.

If you haven't got a hard disk then you probably don't need a cooling fan. The HD-FOX doesn't have one and it gets rid of its heat through the holes in the top. It is silent.

 

[MartinLiddle]

SSDs are not suited to PVR use unless you disable the instant replay facility. During a trial, I wore one out in 17 months continuous use (although there is a possibility that some other problem killed it).

If you buy the correct SSD it should be OK; what make and model of device did you use in your trial?

 

[Black Hole]

Only a cheapie, but what evidence do you have for that? By definition, despite wear levelling mechanisms etc etc, any Flash-based SSD has a finite number of write cycles it can sustain, and the wear levelling can only move the live buffer around in unoccupied areas, so the fuller the SSD is with existing recordings the faster it will clock up its write limit.

To make existing SSD technology properly viable in this application, the live buffer needs moving to RAM.

https://hummy.tv/forum/threads/hd-fox-cf-host-upd.4511/page-6#post-90906

 

[MartinLiddle]

Only a cheapie, but what evidence do you have for that? By definition, despite wear levelling mechanisms etc etc, any Flash-based SSD has a finite number of write cycles it can sustain,

I have recently purchased a Samsung Evo SSD (for a non PVR application) and the specs for the 128GB/256GB drives life time are 150TBW (Terrabytes written) see http://www.samsung.com/global/busin...te/SSD/M2M/html/ssd850pro/specifications.html
So if we assume a five year life that is about 80GB a day. If we assume the average PVR is on for 8 hours a day that is 10GB/hour. Assuming an HD bit rate of 10Mb/s then when recording an HD time shift buffer and two HD channels the data rate would be 13.5GB/hour. If we assume that only half the time two channels are being recorded then you get a five year life. With the 500GB/1TB models the lifetime is 300TBW so that would give plenty of margin.

and the wear levelling can only move the live buffer around in unoccupied areas, so the fuller the SSD is with existing recordings the faster it will clock up its write limit.

No that is not correct. What you are describing is called "Dynamic wear Levelling" but many modern SSD controllers also implement "Static wear levelling" that allows blocks that are in use to be periodically reallocated. See https://en.wikipedia.org/wiki/Wear_leveling

 

[Trev]

Hmm.

 

[Black Hole]

So if we assume a five year life that is about 80GB a day...

Without being nit-picking over the details of your calculations, that would be a good reason for not running such a unit 24/7 - but the one that died on me was run 24/7 (deliberately to test the life cycle). 17 months x 24/8 = 4 years 3 months, not far different. The life cycle for my HDD units is much greater.

 

[MartinLiddle]

Without being nit-picking over the details of your calculations, that would be a good reason for not running such a unit 24/7 - but the one that died on me was run 24/7 (deliberately to test the life cycle). 17 months x 24/8 = 4 years 3 months, not far different. The life cycle for my HDD units is much greater.

To really compare the calculations you need to give an estimate of how many hours HD recording it was doing per day.

 

[Black Hole]

Not a lot, 'tis true - just ticking over on the TSR mainly. Nonetheless, I still say an HDD will knock the spots off a Flash SSD for longevity in a PVR until specific design optimisations are made. There would be little need for wear levelling if the dynamic data areas were moved into a more appropriate technology.

 

[MartinLiddle]

Not a lot, 'tis true - just ticking over on the TSR mainly.

So if tuned to an HD channel that would be a life time of about 5 years for a small capacity Samsung Evo or 10 years for the larger capacity drives.

Nonetheless, I still say an HDD will knock the spots off a Flash SSD for longevity in a PVR until specific design optimisations are made.

I agree.

I am not advocating the use of SSDs in PVRs as they don't offer a performance advantage but it is feasible with a decent quality SSD.