It would be SO COOL to have a low-power MoBo that can run HD MythTV. The next-gen KRP should be built around something like this. Then basically, you buy the board with matching processor, add a hard drive and DVD, and you're good to go. If it's a board with the component video option, then you've got VGA, DVI, S-Video, and Component Video all on the board. No need for a converter, if your HDTV doesn't play well with DVI (like mine).

If any one is interested I have a Biostar TF7050-M2 with a AMD 3800+

Biostar TF7050-M2
AMD 3800+ 35watt ADD3800IAA5CU Socket AM2
GeForce 7050pv / NF630a onboard graphics, HDMI, S-video, VGA
1GB DDR2-800 Note: motherboard will not go lower than 1.9 volt
400 GB Seagate SATA ,10 watts, clunks a lot very irritating
One pcHD-3000 capture card
Power supply 500 watt Coolmaster 70%, I had this laying around still looking for the appropriate one.
Scythe Ninja 150 mm cooler
Two 120mm Artic Cooling PWM fans, one in the case the other in the power supply run off the PWM of the mother board, at 100%load fan speed is 821 RPM @43 deg C

Off........................... 5 watts
Idle........................ 67 watts
Watching 1080i .....80 watt no XvMC
99%...................... 93 watts, watching TV and transcoding, no stutter

I can knockoff another 3 to 4 watt in idle with frequency / voltage scaling. I have had it running and it works very good. I lost my data I collected here so I am going from memory.

Better power supply could knockoff another 3 to 4 watts, In a perfect world I need a 125 watt 80 plus power supply. Even the 300 watt ones I drop off the bottom 20% or the operational range then all bets are off on any efficiency. This is my number one worst problem for making a power efficient machine.

Western Digital Green 7200 / 5400 RPM hard drive could also knockoff 5 watts

So wishful thinking here after upgrades, with pcHD-3000 installed (6 watt)
Idle....... 56 watts
100%.... 84 watts

So far this has been a very good performer
Has had lots of bleeding edge problems with 2.6.23 but most are getting worked out.
I have more information to post but I am running out of time.

TVBox

_________________
TVBox
LinHES R8.6.1

TVBox - great system! I'm "green" with envy. Happy belated earthday to you....

TVbox, that's great info -- thanks. A few questions:

1. Are you able to deinterlace and play 1080i content using the bob2x deinterlacer without XVMC with that processor?

2. Is that a dual-core or single-core processor? I'm finding references to both dual-core and single-core AMD 3800+ processors. I checked newegg and couldn't find it there -- any pointers would be appreciated!

Thanks again!

Marc

Thornsoft:

It doesn't look like either the 780G or 8200 support component or S-video. I suppose some manufacturer might implement them, but I haven't see boards with either. Mind you I haven't looked at every board manufacturer. My gut says that would have to be a chipset feature, and I don't think it is there.

Both S-video and Component seem to be on the boards with the 690G chipset. But I don't know how the support for them is yet. And I don't believe the 690 has hardware decoding (maybe Mpeg2 but not VC-1 and H.264).

If you need S-video and Component out, you might have to go with a discreet video card.

While you said your TV doesn't play well with DVI, how does it do with computer generated HDMI? Both the chipsets we've been discussing (and the 690G) have on-board HDMI.

Marc

I am away from home right now so I can't check but I don't use XvMC. I like the colors in the menus. I am not sure what deinterlacer I am using at this time but it is one of them.
Watching 1080i .....80 watt no XvMC is about 50% cpu
When I had frequency / voltage scaling set up I could watch a 1080i with no problems @ 1000hz with 100% cpu load (Which is 1/2 of normal 2000hz)


AMD makes many Athlon 64 X2 3800+ versions, 89 W, 65 W, 62 W, 45 W and 35 W
I am using a Athlon 64 X2 3800+ ADD3800IAA5CU 35 watt. This processor is very hard to find. Mine was a pull that came out of a HP small form factor. Most 3800+ out there are 89 and 65 watt. I was told once by someone who should know that most of the 35 watts went into the aerospace industry.

I originally started this project over noise issues. My thoughts where the easiest and cheapest way too reduce noise was to reduce power consumption. Every watt you dissipate in a box equates to so many CFM of air movement you need to keep it cool. So more watts = more CFM = equals more fan RPM = more noise. Right now my biggest noise challenge is hard drive noise. The rest of the box is almost silent.

It pays in so many ways to reduce wattage.
less watts = less money = less CFM = less dust = less noise = less heat = less RPM = longer equipment life = less maintenance. I could go on and on.

TVBox

_________________
TVBox
LinHES R8.6.1

I've been running with automated shutdown in place now for 59 hours. Of those 59 hours, 30 hours have been with the system powered down, which yields a 59% reduction in power consumption...


Marc

APCI Wakeup is great. Even at my place the WAF was high. That 5 watts during power off is the best bang for the buck you are going to get. I setup my bios to restart on power failure so if I ever get a power loss (which is extremely rare) it will boot up and continue recording.

TVBox

_________________
TVBox
LinHES R8.6.1

The best way to deal with that one is by the Hard Drive Suspension technique. Basically put the 3.5" drive in a 5.25" bay, suspended with one of several different products. Some people get really carried away with this, and make quite the Rube Goldberg type setups. But the simplest methods are "good enough" and work well. There is also a trend to use notebook drives, as they are cooler and quieter, but maybe not big enough for our application.

I think the broad underwear elastic won't stand the test of time. I've just bought some of the "Stretch Magic" that is the current fad. We'll see when I actually try it out, but my gut says it has a twang like an elastic band. I think most Bungee Cord is too heavy and hard to work with.

I'm leaning towards medium duty shock cord (ie not the thinnest - next gauge up). That the stuff you'll see in most newer tent poles. It's made to be under stress all the time (as in the tent polls), is a smaller diameter than Bungee thus more easily strung and knotted. And the stuff in my tent poles has stayed stretchy for years.

There is much discussion of this over a Silent PC Review. The have a whole section of their forums dedicate towards Silent Storage. Heck, the New York Times recently linked to their original article on hard drive suspension. Take note of the April 2008 postscript at the bottom of the page. You can see some of the overly elaborate techniques in this forum thread.

Thought I'd provide another update. I've been running with "shutdown" mode enabled now for 17 days. Over that 17 day period the system was off 11 days or 63% saving 33 KWH.

Marc

I just noticed that newegg has the Intel Core 2 Duo E7200 for $135. Specs are 2.53ghz, 3MB cache, 45NM, 65 watt TPD. I think this is the way to go. While this processor is not covered in the Tom’s hardware benchmark , a sensitivity analysis of the results there suggest that this process will do a bit better than the E6600 on xvid transcoding. I’ve outlined the basis for this conclusion below.

In terms of power consumption – it’s not clear to me that the 45watt AMD 4850 will consume materially less power than the 65watt intel processor when idle. (Customer review dated 5/6/08 at Newegg indicates that his setup draws the same power with both a 45watt AMD x2 4050e and a 65watt AMD X2 3800+.) It is clear that the transcode will go way faster on the intel processor, resulting in the system drawing higher levels of power for a much shorter period of time.

I would be interesting in anyone’s thoughts on the following 3 questions:

1. Do you agree with my conclusion that the E7200 will outperform the E6600?

2. Are the E6600 and E7200 sufficiently powerful to do hidef playback without the use of XVMC? My guess is that the answer is “yes”, but it would be good to have an explicit confirmation.

3. Any insights into the power consumption difference between the intel E7200 and the AMD 4850? Does my conclusion seem reasonable?

Thanks!

Marc

Basis for concluding that E7200 will do xvid transcoding a bit faster than the E6600.

1. Compare E6320 to E6300. Both have clock of 1.866ghz; change is in cache size going from 2MB to 4MB. Result: Growing cache saves 4 seconds (171-167).

2. Compare E6420 to E6400. Both have clock of 2.133ghz; change is in cache size going from 2MB to 4MB. Result: Growing cache saves 3 seconds (153-150).

3. Compare E6300 to E6400. Both have 2MB cache; change is in clock going from 1.866ghz to 2.133ghz. Result: Increased clock speed saves 18 seconds (171-153)

4. Compare E6320 to E6420. Both have 4MB cache; change is in clock going from 1.866ghz to 2.133ghz. Result: Increased clock speed saves 17 seconds (167-150)

Conclusion#1: Growing the cache beyond 2MB has a minimal impact on performance. Increasing the clock speed is much more important than increasing cache size to improve transcode performance.

Conclusion#2:E7200 will outperform the E6600 on xvid transcode. The E6600 is 2.4ghz, 4MB cache, The E7200 is 2.53BHZ, 3MB cache. The increased clock speed of the E7200 will likely have a more positive impact on transcode performance than the potential loss from the cache size going down by 1MB.

Conclusion#3: If you compare the results of the E6700, E6750 and E8200, you see minimal performance change between the E6700 & E6750, but a significant improvement gong from E6750 to E8200. Looking across all 3 results, my guess is that when the FSB goes from 1066 to 1333, increasing cache size beyond 4MB does help. Unfortunately, I don’t have enough data points to be sure this is the a valid conclusion, as there are too many variable at play. It could also be some other aspect of the Wolfdale vs. Conroe core architecture.