new HTPC build with HD support

[ShiningDragon]

Is that ION statement the original ION, the Next Gen ION (12 core) or the NG ION (16 core)? I've been looking at the ASUS AT5IONT-I which has the full 16 core NG ION. ION Comparison I was hoping to have this as my new frontend (with cardreader + DVD), and move the bulk of the HW to a backend and perhaps add another frontend for another TV in the house.

My statement belongs to the original ION. The "nextgen ION" is not faster then the previous one. In a lot of circumstances it's even more slower. It's even not possible to browse smooth through the internet with an ion / atom (330). A little bit flash advertisement can slow down the atom alot (even with an ion).
ATOM was never really thought as dtr or multimediaplatform. It's concepted as netbook or smartphone processor. My first enthuasiasmens of this platform went very fast down. You'll find better things: on amd and intel side.

[ShiningDragon]

Do you prefer Intel designed boards for any particular reason as you've gone for the DH55TC and DH57JG?

Yes i have. I tested and bought a lot of different chipsets/boards from different manufacturers over the years. Many manufacturers had good products or just bad products. Gigabyte, as an example, is well know for implementing own ways. This may be interesting for oc-kiddies or gamers, but the heller never for a reliable and stable usage. On gagabyte (yes, gagabyte, not gigabyte) platforms it can happen, that identical extensioncards (s-ata controller, tv tuner etc.) may prevent the system from startup, just because gagabytes own way to implement features of the bios functions.
ASRock and ASUS (one factory) can produce good things. While ASUS is a lot more expensive (for what? the support is an 08/15 callcenter without any knowledge.), ASRock is the low budget line (with better support); but the products have a lower quality standard. If you get a good board, all is fine.

With genuine intel mainboards i never had any quality problems. Small bugs, yes. But never a defect whil usage. Freeze, restart... i don't know this on genuine intel platforms (unless there's buggy software responsible ).

Instead of a genuine intel mainboard you can try foxconn too. They're manufacturing a lot of boards for intel and label them "intel".


On the amd side you could of course grab an athlon 245. But i wouldn't buy the "e" modell. The additional price is not necessary, because nearly every regular regor could be dropped to the "e" used voltage. You should just watch to use a "real" am3 board, not an am2+ piece of junk (just google for "hyper transport sync flood error" and you'll know what i am meaning).

And this is the point why i had left amd after the socket 939 era: the quality standards of amd gone after socket 939 BAD! If they bring out new products, then people shouldn't buy them before 1 year is over. Just because the really bad qualitystandards there are often a lot of bugs outside.

[whurlston]

You should just watch to use a "real" am3 board, not an am2+ piece of junk (just google for "hyper transport sync flood error" and you'll know what i am meaning).

And this is the point why i had left amd after the socket 939 era: the quality standards of amd gone after socket 939 BAD! If they bring out new products, then people shouldn't buy them before 1 year is over. Just because the really bad qualitystandards there are often a lot of bugs outside.

I have several (read as 100+) AM2+ Gigabyte motherboards that I've built for myself and others and never had any issues. I also fail to see how the hyper transport sync flood error is AMD's fault. The hyper transport sync flood error is caused by the BIOS applying the wrong voltage to the RAM and it only seems to affect overclocked systems for the most part. Setting the right voltage eliminates the issue, so blame the BIOS manufacturer on that one.

[ShiningDragon]

The hyper transport sync flood error is caused by the BIOS applying the wrong voltage to the RAM and it only seems to affect overclocked systems for the most part. Setting the right voltage eliminates the issue, so blame the BIOS manufacturer on that one.

This is wrong. Both of this. I had myself, after the regors appears to the world, an AMD Athlon II X2 250 on an AM2+ ASUS Mainboard. Of course quality ram and not this cheap crap from G.Skill or OCZ. The voltage adjusted as suggested by JEDEC. And i received a few times per days this "Hyper Transport Sync Flood Error". And not only this one... sometimes the whole screen of Windows suddenly shrinked to a total height of 2 pixels. Just imagine: A resolution of 1280 x 720 shrinks down to 1280 x 2 pixels...
Curiously this didn't happened on native AM3 systems! Even ASUS, MSI and other didn't had any idea why this happens. But if i read the errata of amds current cpus then i can see, that this (oh wonder! they have finally documented this bug!) isn't really fixed yet.
AMD is a low budget manufacturer and has no high quality standards. Intel tells the manufacturers how to built the platforms and tells the manufacturers the limit specifications. AMD on the other hand is "more open" or "not good enough" to set their necessary specifications.
Even look into the newest thing from amd: the native 6 core cpu. The first months the turbo mode didn't worked properly. Now there a lot of people who have troubles with their x6 cpus: The just freezes, stops or hang. Just after a few days of running (as example distributed computing). I regret it, that amd has lost it's track after socket 939. Because with the Manchester X2 4.200+ i was VERY happy.

EDIT: More information about this "top secret" error can be found on Erratum #327 "HyperTransport Link RTT Specification Violation" on amd documents. This problem appeared with the first 45nm cpu and isn't fixed. -.-

[whurlston]

Wow, I've never experienced a single one of those issues on any machine I've built: about 25 with the AMD Athlon II X2 250 (and even more 240/245s) on an AM2+ configuration but using Gigabyte mainboards, not ASUS. And I sometimes do use the cheap crap from G.Skill. I always use it in my own machines.

Just about all reports I've seen of the HTSF issue was resolved by voltage change or a BIOS update. I suspect that the affected BIOS is setting the sync flood state too aggressively (see http://www.hypertransport.org/docs/twgdo...6-0028.pdf )

Even that Errata proposes a BIOS change as a fix:

BIOS should set the Link Phy Impedance Register[RttCtl] (F4x1[9C, 94, 8C, 84]_x[D0, C0][31:29])
to 010b and Link Phy Impedance Register[RttIndex] (F4x1[9C, 94, 8C, 84]_x[D0, C0][20:16]) to
00100b.

[Tower]

How are you using the tuners on the PVR500 anymore? I gave up on using mine about 2-3 months ago when comcast up here in the cities retired their analog cable signals.

Charter in Rochester still has analog signal. There are a lot of people here who still get just basic or expanded basic with no digital service (aside from the "free" OTA digital on the cable"

[Tower]

I do use both tuners, though they are rarely both recording simultaneously anymore, as most of the recording comes via the HDHomerun tuners. I can't say I've noticed an issue with that scenario specifically, but I can give it a try tonight.

I've also ended up using yadif in ffdshow-tryouts for deinterlacing, since I gave up trying to get good (vector/motion) SD deinterlacing with the Intel HD Graphics.

Seemed to work fine. I have had some problems once in a while viewing HD and SD where the sound keeps up, but the video hits a patch of molasses. If I skip back 10 seconds everything is fine again.

It has never been anything I've been able to recreate... it just sometimes happens. One of these times I'll have to keep a performance monitor running and if it happens see if there is anything obvious going on there.

[ShiningDragon]

Wow, I've never experienced a single one of those issues on any machine I've built(...)

You seem to be a very lucky guy. ^^ I had no problems with amd between socket a (barton generation) until socket 939 (manchester); even nforce 2 ran absolutely nicely by me (all my terminals in my internetcafé were based on barton/nforce2). But after this only problems appeared, like the via generation with their kt133-e (dma bug, data loss on harddisks) and an athlon thunderbird c 1.4 ghz (ayhja). And now i am not longer interested to spent only one cent to this products.
But you can see yourself, that this problem exists. And it can't be part of the bios programmer to fix the shit amd is unable, or better uninterested, to fix. One cpu generation with such a bug... ok. But not all 45nm cpus, that's not a good sign for amd.

With this experience, reports from the internet and even amd documents itself, i am not willing to recommend anyone a amd specific system. I just can't guarantee, that any recommended configuration would be 24h/7d stable.

[whurlston]

I understand your position based on your experience. I just have to take the opposite one since I have had no issues myself with using/building/selling AMD products and they have been extremely stable for me even when run 24/7.

[johnsonx42]

put me in whurlston's camp on this one... I have hundreds of AMD workstations and servers in the field, and have run into this ht sync flood error on only one system. the eventual conclusion was that the error message was essentially a red herring - the box was crashing for another reason related to the raid controller, and the htsf error was merely a side-effect or after-effect of the crash.

overall I've had more odd headaches over the years with Intel stuff than with AMD, and that's from a far smaller sample.

the notion that AMD cpu's are budget, low-quality, etc. is pure sillyness. if they're good enough for multi-million dollar supercomputers, they're good enough for me.