i7 or wait for new sockets?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

bigpow

Platinum Member
Dec 10, 2000
2,372
2
81
People seem to be misunderstanding the OP's problem here (at least to the point that I under stand it :))

This is not a question of smooth 1080p playback, it's a question of the EXTRA stuff he does with ffdshow post processing.

@ZimZum
See http://mod16.org/hurfdurf/?p=12 for (some) explanation on interlacing.

I understand interlacing very well, I also used ffdshow to double-up, deinterlace and pp before - 3-4yrs ago when I bought my 1st 1080p LCD and when full HD contents were rare.

My question is, why still do that today?
Only reason to do it:
1) QuadHD resolution monitors
2) 80" inch and above 1080p screen
3) try to be an elitist/haxor? :D
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
I understand interlacing very well, I also used ffdshow to double-up, deinterlace and pp before - 3-4yrs ago when I bought my 1st 1080p LCD and when full HD contents were rare.

My question is, why still do that today?
Only reason to do it:
1) QuadHD resolution monitors
2) 80" inch and above 1080p screen
3) try to be an elitist/haxor? :D

or you might just enjoy the quality improvements of using high end filters?

I personally don't care about them but it is important to some people.
 

punker

Member
Oct 24, 2007
121
0
0
(it just can't handle HD as well as I need it to at this point) and it is due to the CPU.

system specs:
E6600 @ 3.0GHz
Gigabyte X48-DQ6
Powercolor HD5750
2x750GB Seagate 7200.11
2x2TB WD Green
Hauppauge HD-PVR
.

What a funny joke

my old system from 06 handles 1080p fine


using a 9800GT
and cuda
 
Last edited:

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
27,287
16,123
136
What a funny joke

my old system from 06 handles 1080p fine


using a 9800GT

you must be high on drugs or something or you just want to upgrade and waste money

If you will read the entire thread it makes a lot more sense.

Please don't reply with such negativity until you have read the entire thread.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
"Intel's next-generation Sandy Bridge processors may not be due out for at least another six months"


6 Months is still a lot of time, let alone factor in the high prices at first as well as not a large number of boards either.

yap... 6 months is a while.
My prediction? 6 months from now the first sandy bridge processors to be released would be 300+$ and would have mobos that are ~300$.
Then, half a year later, cheaper mobos and a cheaper sandy bridge version will be released.
 

TheRyuu

Diamond Member
Dec 3, 2005
5,479
14
81
My view of things is that waiting for something is pointless. You want an upgrade or a new system asking whether to wait is also a pointless question, you do it now with what's out. Moreover, waiting is pointless since the next best this is always just over the horizon.

Could be a bad way to view it but that's just my personal opinion on the matter.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
My view of things is that waiting for something is pointless. You want an upgrade or a new system asking whether to wait is also a pointless question, you do it now with what's out. Moreover, waiting is pointless since the next best this is always just over the horizon.

Sometimes waiting is pointless, sometimes waiting is useful.

Prices only go down if there is a pricewar, or if a new product has obsoleted the old one.
If you are a month away from the release of a next gen piece of hardware that has been on the market for 14 months already, then waiting 1 more month will give you a lot more for your money, since the moment the next gen hardware is released your current hardware is devalued.

Hardware is constantly upgraded, with old hardware sold and new hardware bought to replace it.

If X_0 is the price of generation N hardware at release, and X_1 is the price of generation N hardware when generation N+1 is released, and T is the time until generation N+1 is released, then the effective cost to you is (X_0 - X_1) / T

So if you a hypothetical chip was released on 1-1-09 for 200$ and is reduced in price to 100$ on 1-1-10 when the next gen chip arrives, then the monthly cost to you is (200$-100$)/12 month = 8.3$/month if you buy it on release day, and (200$-100$)/1 month = 100$/month

There is an additional cost for the sale of older hardware and purchase of new hardware, a a transition cost if you would. But that further advantages buying as early as possible in the product cycle.

Generally speaking we do not know when the next gen arrives. So it is hard to predict, but you can make general estimates.
 

Fallen Kell

Diamond Member
Oct 9, 1999
6,221
540
126
[HTPC newb] What image quality benefits do you gain from these? [/HTPC newb]

Currently, the quality benefit is in doing a proper de-interlace. As good as most software out there is, a proper de-interlace algorithm is something that just isn't done properly, mainly because you need to buffer the frames. Most of the hardware de-interlace algorithms do not do this and instead simply do a "bob" or worse. This results in a lot of saw-tooth artifacts.

Now, I might be able to look into it more now that I no longer do much resizing with ffdshow (since I am now recording everything in 1080i or 720p (the latter of which I don't need to worry about the interlaced video issues)), but there are lots of issues if you resize before you deinterlace. Take a look at the following page for details (the first 3-4 photos will tell you all you need to know about resizing before deinterlacing): http://www.100fps.com/

That above site will give you some general overview of the issue (even though it is clearly written for PAL with 25 fps, the technique is the same for NTSC, simply at 30 fps).



The denoise is to try and remove some of the "fuzz" and "mosquito noise" :
http://i.cmpnet.com/videsignline/2006/02/algolith-fig2.jpg

In the past I also did edge sharpening using UnSharp Mask, since that is an area where deinterlacing tends to make a little blurred no matter the algorithm, and thus it benefited from getting some sharpening. For some basic info on it see the following:
http://www.luminous-landscape.com/tutorials/understanding-series/understanding-usm.shtml