"clean" dtivo easily 100% "faster" -- why so?

[gobsmack]

Gang,

I just got done recovering from a major hardware crash on my DTivo.

I had no mfstools backup, so lost everything and had to start with a completely new image. Also had to go to one drive from 2 (120 G is enough for me, given that I extract quickly and often).

So, here was the real surprise for me: this thing is now lightning fast. Like, "oh my god, did someone just slip a pentium 5 in my Tivo" fast.

Even when I run TiVoWeb, response is much faster than my wife's tivo that isn't running TiVoweb.

What bogs down a "mature" TiVo such that it is so much slower than this "virgin" TiVo? Is it all the SPs? Thumb ratings?

I did go to a 127M swap on this, which I don't think my Wife's tivo has -- could that make that much of a difference?

I haven't done a ps aux to see what is/isn't running on each, but that is a next step.

Thoughts on why such a dramatic speed diff? Thanks.

[captain_video]

My guess is that as you record and delete more and more recordings, the files on the hard drive start getting more and more fragmented. I don't know for sure if this is true in a Linux filesystem but I'm sure any of the Linux gurus here can confirm or deny this. On a PC, when your files start getting fragmented, file transfers start getting slower and slower because the drive has to seek out all of the fragments for the requested file(s) and link them all together. This would seem to be a plausible cause for a Tivo to slow down after a period of time and also why a "clean" drive would seem much faster.

[TheWickedPriest]

There's not much fragmentation on a typical Linux filesystem. However, Tivo doesn't actually use a Linux filesystem (except for / and /var) -- it uses MFS, which is their own weird user-space filesystem. Fragmentation there is limited by the use of enormous block sizes (like a meg each), but perhaps it's still a problem... I dunno.

[BubbleLamp]

I think it has more to do with SP's, Wishlists, and existing recordings than it does with fragging.

[gobsmack]

The difference is amazing, and clearly I'd love to have this speed *forever*, so:

- is it worth testing this in a systematic way to see what drives a "slowdown"?
(to me, yes; to others, weigh in)

- is there an objective way to evaluate tivo "speed"? File transfer speed is limited by a bunch of things off the TiVo, so it isn't perfect, but it is one test. Are there others? Is there some utility that could be set to a really low prioirty then start a stopwatch, count fingers/toes, and stop the stopwatch to report a figure.

Obviously, with some objective measures, you know how slow your TiVo is, and with the first (some hypothesis testing) you then have some control over speeding your TiVo up.

One last side note: i was amazed that it took a virgin dtivo nearly an hour to clean out all the thumbs up ratings while doing a setup -- and this was before I had even connected the satellites, so this unit had nothing else going on. If a slowdown comes from lots of thumb ratings, I'll never enter a thumb rating again!

[GREEK]

It has ALOT to do with suggestions and thumbs ratings, I dont use them and havent lost any speed from day one really. The only slow stuff is now playing if you have tons of stuff in there, would be nice to have folders huh? (SA 4.0 anybody) I think 8 mb cache drives are faster, even thought the tivo is ata33 it "seems" faster than other drives.......

[gobsmack]

Originally posted by GREEK
It has ALOT to do with suggestions and thumbs ratings, I dont use them and havent lost any speed from day one really. The only slow stuff is now playing if you have tons of stuff in there, would be nice to have folders huh? (SA 4.0 anybody) I think 8 mb cache drives are faster, even thought the tivo is ata33 it "seems" faster than other drives.......

Greek, I think you're totally right. I've also enjoyed that with no thumbs up ratings, it doesn't record tons of crap. Perhaps I'll just go one step further and turn off suggestions....

Any thoughts on Linux benchmarks for TiVo? I'm curious enough that I want to do some testing...

[Gill]

Is there any way to keep it from giving a thumbs-up rating to everything you record? I turned off suggestions after my last reload too since I noticed it was a lot faster......

Gill

[Generiq]

Originally posted by gobsmack
Any thoughts on Linux benchmarks for TiVo? I'm curious enough that I want to do some testing...

I saw an MRTG module for TivoWeb, but you would need to setup an MRTG server. Search the forum for the link in someone's sig "Tivo has been MRTG'ized"

[FredThompson]

Originally posted by GREEK
I think 8 mb cache drives are faster, even thought the tivo is ata33 it "seems" faster than other drives....... Sure, the drive keeps transferring data when the head is seeking. 7200 rpm also makes a difference. It's not that ATA33 is slow for a TiVo, it's all the little delays as the head has to move around during which data won't be transferred with a 2M 5400 rpm drive. Isn't MFS using 1M blocks?...