Processor for video editing

I know the discussion for "which processor for what need" has been visited many times in this forum (C2D vs Turion). And I've concluded that Intel is the processor of choice. However I wanted to get an idea if a particular setup will suit my need. The most involved application I use at the time is Pinnacle Studio 11. I do no gaming at all. I had my eye on a HP dv2845se that's selling for about $799 at Best Buy. The price seems reasonable and I have a couple of gift cards to use. It has a Turion TL-62, 3GB RAM, GeForce Go 7150M, running Vista Home Premium.

So what I what to know is, will this setup allow me to run my editing applications over vista with no fuss. Or should I not focus too much on the processor but rather the Graphic Card? In the future I also plan on doing some live sound recording with the laptop I choose.

I would really appreciate any feedback.

 

well video editing, will take EVERY thing you give it, the Turion is not a BAD processor, it is just slower at some things. TL-62 =~1.8 intel(T7100) or something like that.

 

The only difference between Turion and Intel CPU's is speed. But other than that there are the same.

FOr video editing you already have a good setup, so you should run it with no fuss at all.

 

Pinnacle Studio will run fine on that machine.
The go 7150 may be the only slowdown, since it is a weak integrated graphics card.
Just make sure that the copy of pinnacle you have can run on a vista machine.

K-TRON

 

Thanks for the quick responses. I do think my version of Pinnacle Studio is designed to run on Vista also. Hopefully the go 7150 won't effect thing too much be discouraging. Anyway, with your feedback I can now make a comfortable decision on whether to proceed with this particular notebook. Thanks again guys!

 

Also consider battery life, the Turion laptops are not only slower than Intel counterparts but also aren't as efficient with power - not sure if you work on the road a lot but if so this is a big consideration.

 

You might wanna go after a better gfx card. Video encoding is faster with a better gfx card. Im guessing same will apply for video rendering..

 

Do you think that the gfx card will have a significant affect the rendering process? I usually have the video playback set to a lower quality during the editing process. I'm more concerned with the fluidity of the application during use and the overall speed of the actual rendering process.

 

I only do encoding so im not sure if rendering uses the same processes. Upgrading from an integrated card to a 7200GS (desktop btw) dedicated made a very significant improvement to my encoding time

Im sure the GO 7150m can handle it fine, well its alot better than the integrated card i had anyway

 

Actually the video chip has absolutely nothing to do with with video editing, encoding or rendering. Video, after all, is mostly 2-D rendering and even a Commodore 64 can handle that. Now if you're working with some heavy 3-D apps a better graphics card will render previews quicker, but still won't influence the finished product in the least. What will affect the outcome, you ask? Processor power coupled with software that's optimized for a multi-processor environment. Pure and Simple.

 

2D clock speeds are faster on better cards, thus making encoding faster (to a extent). Im gonna try OC my 2d clock speeds to see if it goes any faster.

Ive already proven that its significantly faster with the desktop PC i use at work.

 

So basically your saying that PCs with Dedicated Graphics card encode faster than PCs wwith only integrated.

Will overclocking the Standard 2D in Rivatuner also help encoding and Video Quality speeds???

 

Well i had a crap integrated card on my work PC I got about 20-30fps which encoding took right about the same time as the length of the video

when i slapped a 256mb 7200GS AGP, it easily flew up to 65fps+. At first i though encoding depended on the program you were using but i was wrong. gpu played a part too

About the 2d clock thing, im not sure yet, im still to test it out. If i see my encoding fps is higher when i OC 2D clocks then its true

Also, Overclocking 2D(or was it low power 3D) improves Aero. Sounds true but never actually tested it before

 

I'll test ot out too.

Just overclocked my 2D bar and it showed improvement on Dock effects.

Will try encoding them. I'll underclock it first then overclcok.(So that I could really see the difference)

 

What software do you use to encode videos???

Looks like converting videos does nothing to the GPU.

 

I use Handbrake or SUPER. I use 2 other programs but they are not for free

 

Overclocking the 2d mode of your GPU will help speed up aero effects. It will help smooth out flip 3d etc as well, when you first engage it. After a few seconds, the gpu clocks up to low power 3d or full 3d mode and it will then be smooth by itself. But for smoother runnings all the time or at least initially when first running things like flip 3d, ramp up the 2d speeds of your GPU. I found this helps on my machine but I don't leave it clocked higher to reduce heat.

I'm surprised though that a different GPU affects encoding - I really thought this was entirely CPU dependent, unless you're using maybe an integrated solution that steals RAM or even is unable to offload certain GPU tasks from the CPU.

 

Im suprised myself, first i though it was software dependant. I can only guess the CPU/GPU/RAM and maybe even HDD speed all play a part in the encoding process

The only thing i added was a 7200GS 256mb AGP and a 512mb DDR stick of ram that was lying around.

My notebook on my sig does 100+fps encoding which is crazy fast compared to my work PC

 

To my knowledge the video card is doing nothing in the rendering process, at least nothing more than having to display the picture on your screen, so it should have nearly no effect in the rendering process.

 

Display speed has nothing to do with encoding. And just who have you proved this to? One possible explanation is that by upgrading to a card that didn't share memory and

Well duh...Doncha think adding RAM might have had something to do with it??? I do.

 

Just wondering.. why do hi-def videos require a decent gpu for smooth playback? They are 2d aswell, and they can also use the same video codecs as the encoding i do.

Im not saying the GPU did it all... but it did help. Im quite new with the video encoding scene. I dont think the the RAM could have done that much. Id take it out right now and test it but im busy converting audio files and cant afford to shut it down

 

This depends on your gpu and codec. Nvidia and ATI both have put codecs in there new drivers/gpus that let the gpu share decoding duty with the cpu, however without that codec in play or using an older gpu that doesnt support it it would be 100% cpu duty minus just having to put the image on screen. HD stuff is VERY hard on the cpu so most people wont have smooth playback unless the video card helps out.

 

Yup its Nvidia PureVideo and AViVO for ATi cards. The GPU takes some of the CPU's work

So why cant this technology be possibly used for faster encoding aswell?

 

SOME NLE Video editing CAN be helped by graphics cards, but the number of application is limited, and Handbrake is not one of them, and Pinnacle's consumer stuff only in real-time effects. And the major area of assist is in the transition and preview portion, render output is only assisted in a very small number of programs, and they are very specific in what they accelerate. As examples, Avid's Liquid and XpressProHD as well as Apple's Motion are GPU assisted in both preview and output, but Pinnacle is limited to preview/real-time effects not in helping with final output render times. Compositing is actually far better helped with GPU power and there are a bunch of apps out there that benefit from GPU-acceleration. But that's beyond your setup or an 'upgrade' to a weak discrete GPU. Some apps like Avid & Premier Pro have plug-ins (like Red) but you have to see which GPU they prefer (Ati, Matrox, nVidia, etc.) and offer and even then it's still mostly OpenGL effect support. For the level of support they offer, you would want something more modern and more powerful to get any benefit from them. Best results are for the ATi X1K, nVidia GF8 and ATi HD 2/3K series which can really make use of GPGPU acceleration with many more processors. But alot of this stuff is still either very early in development or else still very exclusive to the professional end of these companies' products and plug-ins.

For this situation CPU/RAM/HDD are most important, and in that order. You want a fast multi-core CPU above all else, then lots of and fast RAM, and then a good fast hard drive is good.

For Pinnacle, gpus help in preview & real-time effects, not final render.

 

So was it the RAM and not the GPU that increased my encoding fps? The system itself did not feel any faster when i added 512mb ram but the encoding times did significantly improve when i added both RAM and GPU. I would have thought it was the GPU's doing..

So Hardware GPU-acceleration is non existant for normal encoding?

 

Because it's dedicated to that task, and really it's not AVIVO and PureVideo those are the superset of the software packages and hardware components combined, the hardware part is UVD for ATi and VP/BSP units for nV.

It can be but it's very limited in it's current application.

Even when we thought that it was being implemented in Avivo encoder, it was really just a Device ID check not actually using the processing power advantages of the ALUs in the X1K.

 

Depends on the task, if you're watching effects, it may have done that faster (because it will preview transitions and such faster if it supports GPUs), but not handbrake, that would've benefited most from the added memory to encode more chunks and keep more work in flight.

No it's out there, but it's out there for very specific applications, like the ones I listed which are geared towards exploiting the GPGPU side of the equation. And even then they still focus mainly on the areas that are harder for a CPU to do, which is things like compositing, and image noise reduction, and such. Straight forward DV->MPEG or MPEG->DIVX are more difficult to adapt right now, and they are a little more complex than the simple stream processors can do, and don't benefit as much from small parallelism, whereas the difference computations in noise reduction and such are perfect for that type of thing. I think the main disadvantage right now, is even as co-processors, it's early for them to apply these benefits more universally (currently still pretty much stuck inside CTM and CUDA), most of the current GPU assist is through OpenGL and DX pathways, mainly it's the transitions area that can be assisted, not the overall output render, just parts if they can be properly flagged in advance then they are done. And even with the limited application in CTM and CUDA how long this will take to get to consumer level versions of these products was unknown, but it looks like they're saying by the fall, so somewhat ahead of what I expected;

http://www.bit-tech.net/hardware/2008/04/14/nvidia_analyst_day_-_biting_back_at_intel/4

 

Ram helps with the video editing and stuff because more can be stored in the ram. The specifics of what is stored there depends on the program and what your doing as well as your settings. Me personally I have been working with adobe after effects alot lately and having more ram speeds up encoding significantly because it can store an effect in the rams buffer and the raw video feed with that effect rather than having to render it each and every time you view it or render it. Something along those lines.

I saw a video tutorial on it somewhere that got into the specifics of rendering with after effects and when they were going over the advanced settings thats when they got into the ram preview and ram cache for rendering.

Video editing and 3d rendering and things are one of the few applications that can make good use of a raid 0 setup since the files are so large.

Shame I had 4gb of ram for my laptop but it wont boot with 4gb, so downgraded to 3gb and it only runs on A/C power with all 3gigs in. To get it to run on battery I have to use 2gb of ram. Id be interested in seeing how much that extra gig or two would help me out in editing since I only tried it in games and it didnt make much a difference.

Still with 2gig of ram and a c2d @ ~2.8ghz things go pretty fast except full quality preview is pretty slow.

 

So I guess you need more CPU power to increase FPS in Video Encoding(learned something new)

Better buy more powerful CPU next time

 

I guess this explains my notebooks ultra fast encoding speeds. 4gb x64

For 3D rendering/CAD programs a good GPU is required correct?

 

Yes and no, for most 3D rendering and CAD a GPU is good, but it's still mainly just for viewport acceleration (ie the preview) final render is still CPU/RAM/HDD dependant. A better GPU lets you see more faces, polygons, layers, etc at once, and lets you move the object or around the object quicker when manipulating it.

Now if you used Gelato, it will use the GPGPU power of a GF7 & especially GF8 GPU to assist in the final render output acting like an efficient co-processor. But for 3DSMax, Maya, Rhino, etc, it's still mainly just (viewport) preview that is affected by GPU power.

 

FWIW there are Hardware encoders, from the simple MPEG 2 encoder in my Hauppauge USB2 PVR to arrays and servers like these: http://www.digital-rapids.com/ProductWizardResults.aspx?answer=Post Production. I don't believe I've seen accelerators in any kind of LapTop friendly form factor, however. Doesn't mean they don't exist, just haven't seen them yet.

 

We aren't talking about dedicated hardware encoders there are many options out there for dedicated solutions that do the task, we're talking about using generalized GPUs to accelerate tasks like encoding, not even something like Tesla, but just an average GPU potentially even an IGP acting as a co-processor.

Something that was promised for later this year and announced earlier today by nV is the CUDA application RapidHD;
http://www.bit-tech.net/hardware/2008/04/14/nvidia_analyst_day_-_biting_back_at_intel/4

It's an example of the limited application that doesn't have general consumer level support yet, but will trickle down eventually.

 

umm

i got lazy for reading all this 4 pages, just want to come in and deliver my $0.02

i have a desktop with E6400 for about a year and half now with 2GB ram and x1950 256MB.

i normally convert RM/RMVB files to AVI/DVD files for dvd burning, initially i OC'd my E6400 from 2.16GHz to 3.2GHz and a normal 45 minutes/200ish MB file will take about 30 minutes to convert, while if i leave my CPU at stock speed and convert the media files, they would take 1 hour and 20 minutes or so. I think for video encoding and editing, it would be the same. higher clock speed will perform better for these things, and multiple cores wont help.

that's why nowadays its still better to use E6800 (2 core, 2.93GHz) instead of Q6600 (4 core, 2.4GHz) to do video rendering. (i think)

 

I realize the OP was asking about using a GPU for accelerating encodes. I was merely pointing out that specialized processors do exist for encoding.

 

Because most people use innefficent codecs and players, making them need the gpu for offloading part of the rendering from the CPU.

With the right Codec/filter, you can watch highdef video on old CPU's. 1,7GHz P-m are enough for 720p video, for instance, using the CoreAVC filter. Even the h.264 decoding part of the FFdshow filter is more efficient than most regular decoders.