+- Kodi Community Forum (https://forum.kodi.tv)
+-- Forum: Discussions (https://forum.kodi.tv/forumdisplay.php?fid=222)
+--- Forum: Hardware (https://forum.kodi.tv/forumdisplay.php?fid=112)
+--- Thread: Best XBMC HTPC Hardware with HDMI / HD / 1080p + HDMI Audio (/showthread.php?tid=34235)
- sebj - 2008-12-10
@magho
Quote:And now that HDMI audio finally works with nvidia,
Really? If you can spare a couple seconds, how did you do it?Been trying to get that to work for a while
sebj
- spozen - 2008-12-10
ziro
I was thinking maybe we could start a benchmarking thread?
But it seems like the E8400 is still the best CPU for XBMC if you have to buy a copper heatsink for the celeron processor, heatsinks are expansive as hell.
But i could be wrong, we will see when BLKMGK get his system up and running.
BLKMGK, are you going to test the celeron in the case you bought from newegg? It seems like a standard case, this is interesting because if you are using stock cooler + a fairly normal case (not 4 120mm fans), it will be cheaper than buying a E8400 system.
- Matt Devo - 2008-12-10
honestly, the killa sampla is really overkill, as you're unlikely to find any video anywhere near as taxing on a system. I think we need some samples that are better representative of expected system load, and leave the killa sampla as a worst case scenario (that's unlikely to ever occur)
- spozen - 2008-12-10
Matt Devo Wrote:honestly, the killa sampla is really overkill, as you're unlikely to find any video anywhere near as taxing on a system. I think we need some samples that are better representative of expected system load, and leave the killa sampla as a worst case scenario (that's unlikely to ever occur)
In the future movies might be encoded like the killa sample.
- ziro - 2008-12-10
@spozen
We could make a thread but first we need to define which information is valuable and necessary and then provide a list of samples to be able to benchmark more acurately the configs.
However I don't think it's a good idea to base a benchmark only on the killa sample since it's not well encoded. Nevertheless i agree that some heavy weight samples might be interesting to test for the future yet since we can't forsee it and especially the time when 2160p will become a new reference, i don't feel it's mandatory. In any case i have at my disposal some 100mbps samples if i remember well so tests are possible.
Now from the data i gathered so far, it is true that the E8400 stands out but it's also true that cheaper CPU manage to deal with the killa sample and 1080p in general. I would gladly cut the costs on cpu/mobo to spend it on storage or nicer case for example.
All in all i believe it's necessary somehow to have a little database with this valuable information but first the model has to be drawn. And we can't either drop all the amd based systems since the idea, as i see it, is to provide a benchmark, and the wider its variety, the better. We can't focus on one single cpu/architecture.
- spozen - 2008-12-10
ziro Wrote:@spozen
We could make a thread but first we need to define which information is valuable and necessary and then provide a list of samples to be able to benchmark more acurately the configs.
However I don't think it's a good idea to base a benchmark only on the killa sample since it's not well encoded. Nevertheless i agree that some heavy weight samples might be interesting to test for the future yet since we can't forsee it and especially the time when 2160p will become a new reference, i don't feel it's mandatory. In any case i have at my disposal some 100mbps samples if i remember well so tests are possible.
Now from the data i gathered so far, it is true that the E8400 stands out but it's also true that cheaper CPU manage to deal with the killa sample and 1080p in general. I would gladly cut the costs on cpu/mobo to spend it on storage or nicer case for example.
All in all i believe it's necessary somehow to have a little database with this valuable information but first the model has to be drawn. And we can't either drop all the amd based systems since the idea, as i see it, is to provide a benchmark, and the wider its variety, the better. We can't focus on one single cpu/architecture.
This sounds like an excellent idea but the problem with benchmarking on forums is that it will become like this thread: a lot pages to scroll up and down to search for valuable information. If you create a thread maybe you could manage it to look like this for example:
Code:
User:
CPU:
Motherboard:
RAM:
Graphics Card:
etc...By manage i mean that you edit the first post and update it after every new post.
About the sample, i agree that we should test a few other samples like scene samples from diffrent movies and in diffrent resolutions(720p/1080p). We could also test some other codecs but WMV sucks i think we all can agree on that.
- Matt Devo - 2008-12-10
spozen Wrote:In the future movies might be encoded like the killa sample.
doubtful, as an h264/AVC video stream direct from a bluray disc (no recompression) isn't even as CPU-intensive to decode as the killa sampla.
- TaeOH - 2008-12-10
I have been recoding my Blu Ray an HD DVD's to 720p in most cases, I only have a 50 inch 1080p set and sit about 12-15 feet away. I rarely see a difference in detail between 1080p and 720p.
Can I assume something like an AMD 6000+ should be able to handle them? I have been looking at upgrading my Media Server so I could run XBMC, but really want to keep the cost down.
I figure if I ever get a bigger set, I could just watch the disk to get the full 1080p. I do have some 1080p content on my server now, basically AVC Blu Ray's that are 20gb or less. It was just quicker not to recode and I have plenty of drive space currently.
- spozen - 2008-12-10
Matt Devo Wrote:doubtful, as an h264/AVC video stream direct from a bluray disc (no recompression) isn't even as CPU-intensive to decode as the killa sampla.
You need more processing power for compressed video than for uncompressed.
TaeOH, 6000+ will be enough for 720p and most 1080p (from what i've heard).
- Matt Devo - 2008-12-11
spozen Wrote:You need more processing power for compressed video than for uncompressed.
I consider a 45Mbps AVC bluray video stream to be compressed. Rephrase perhaps? =)
- TaeOH - 2008-12-11
spozen Wrote:You need more processing power for compressed video than for uncompressed.
TaeOH, 6000+ will be enough for 720p and most 1080p (from what i've heard).
That does seem to be the case from what I have been read. Since I have PS3's handling my HD content now, they could always be a backup if I got into trouble with a movie.
Thank you.
Right now I am just looking at replacing my 4200+ and Abit AT8 32X with a 6000+ and ASUS M3A78-EM, I have a spare 1gb stick of DDR2 1066 too add in. If this works out I can see myself moving this to a Silverstone HTPC case and putting the 4200+ back into service as just a server as it is now.
- BLKMGK - 2008-12-11
Wow, this grew!
For storing off a spreadsheet you might look at Google Apps, I have seen shared spreadsheets done there. I think databases might be possible too, am not positive.
Case
My intent is to test this in the normal(ish) case yes. I have a spiffy HTPC case already in use. It's been a PITA - the LCD requires patched drivers, the IRDA receiver requires patched drivers, the bottom of it had to be drilled to allow air to exit the P/S, and the side vent was noisy - not the fan but the air rushing through the silly vent! Oh and it was pricey but quite pretty, shame the LCD cannot be read from more thna 2foot away even if it worked for me. I never stick DVD in it except to load software and it could easily be shoved behind something with the USB IRDA receiver in view for all I care. It would work just as well so I'm saving like $100 on this next build by using a normal EASY to work in case that I can hide. Or I hope easy to work in, I aim to find out tomorrow when I pick it up and begin assembly. I hope it will be pretty quiet - we'll see.
Heatsinks
I intend to try the stock CPU cooler first. If that doesn't work I will have to ponder some, maybe try an OEM cooler from a C2D on it. I really tend to prefer the Zalman copper units and have a low profile "flower" looking one in my HTPC case. My E8400 has one of their BIG ones - love it! But for this next build cost is an issue. I will try stock cooler and stock grease, then maybe C2D cooler, then maybe silver grease. My goal is Killasample with no dropped frames.
Benchmark
As for that as a benchmark... I have said it before that on my own encodings I saw dropped frames during pan shots that had lots of detail like jungle foliage. King Kong, encoded by me not some downloaded thing, had issues noticeable to my eye. Until I hit 3Ghz and at that point roughly things stopped dropping frames and Killa' played well.
Now, if someone wants to benchmark sure I'm all for it but BEFORE you do that choose a standard to go against. I agree Killa' is over the top so try encoding that SAME scene from the BBC Earth series with other settings - watch as it beats the snot out of your system. Yeah, my encoding of that does it too!
You're down to encoding settings to argue about. I tried the AVS "recommended" settings for BD recently, the ones that will supposedly give me a profile that can be hardware accelerated someday. I ended up with a TINY file that looked so bad I couldn't stand it. I do not know WTF but the settings were awful so I must have done something wrong. I'll try again later but for now I continue to use my proven settings - Dark Knight will get muxxed in a few days when I get home!
I'm certainly willing to share my settings for meGUI but know that that they are a bit extreme so I don't want to hear grief - maybe constructive grief if it results in smaller files w\no quality loss. I see ZERO artifacts in my encodes, ever, and reduce size by about 40% usually, sometimes more. If I had an easy way to get an original copy of just that single scene that Killa' was made from to encode with my settings I'd be willing to share it too.
Bottom line - be careful you do not choose something too easy. While doing so would make many AMD owners happy it would mislead people - especially people who encode their own and don't download stuff off the 'net. Apple trailers for instance don't even make my machine breathe heavy and *I* wouldn't accept that as a reasonable benchmark. <shrug>
Oh and I'm not a video guru, I just want the video to look good and I can spot artifacts, tearing, and dropped frames. I'm critical of that and hate it. I'm only building my second XBMC machine because my aTV drops frames on downloaded 720P TV shows
- TaeOH - 2008-12-11
@BLKMGK
I would love to know how you encode. Currently I use the DXVA profiles and do constant quality 18. Most everything satisfies me with those settings and the PS3 can play DXVA profiles.
- magho - 2008-12-11
sebj Wrote:@magho
Been trying to get that to work for a while
sebj
Nothing fancy, really. Installed ubuntu intrepid (no desktop or wm, just X), compiled alsa-driver 1.0.18a and nvidia 180.11 and it just worked. What nvidia chipset did you try with?
@ziro. Tried killa sample last night and it didn't play.
- sebj - 2008-12-11
@magho
I have the integrated 7100 nvidia part on a gigabyte board.
I used a script from the ubuntu forums that downloads an compiles the latest alsa
I'm running the current driver 177.82 from nvidia(ie not the ubuntu one), is the hdmi functionality a new feature in the beta driver by any chance?
I think I'll try when i get home from work, you never know.
Thanks
sebj