I doubt if you'll get 7 times your performance but it should improve it considerably. First thing I noticed was the 8 GB RAM, which may be a bit low for the performance you want. I'd suggest at least 16 MB. Also, other factors besides your graphics card are at work, for instance your hard drive. Here's the benchmark for the one your DXDIAG shows for your C Drive:
http://www.harddrivebenchmark.net/hdd.php?hdd=Seagate+ST1000DM003-1CH162

A useful tool to check your overall performance with is the Windows Experience Index, which it appears is not available in Windows 8.1. However, a search shows that there is a program available to replace it here:
http://www.intowindows.com/windows-experience-index-wei-for-windows-8-1/

I use an SSD for my C drive and a GTX 660 for my graphics card, which gives me a WEI of 7.2, and my preview is still choppy at High Preview Resolution, so I use the Render Preview function to check for video/audio sync before rendering. If you're not familiar with it a You Tube search should turn up a video. Dell Precision 7510 Laptop
Windows 7 Pro 64-bit | Intel(R) XEON(R) CPU E3-1505M v5 @2.80 GHz
RAM: 32 GB
Windows Experience Index 7.5

Thanks for your answers

First thing I noticed was the 8 GB RAM, which may be a bit low for the performance you want. I'd suggest at least 16 MB.

Point taken. FYI: During current rendering about 4GByte of 8GByte RAM are still unused. (my videos are rather short for now). So for now, RAM is likely not (yet) a limiting factor for me.

other factors besides your graphics card are at work, for instance your hard drive.

Good point. But looking at PD12 it often produces only 0.1MByte/second (during heavy effects section) . Taskmanager shows a disk utilization of <2%. So I don't think the disk is the limiting factor for now. Even at a "Fast Production Speed"(no effects) the hard drive is barely used at 3MByte/second.

so I use the Render Preview function to check for video/audio sync before rendering.

I use it too, but it is no faster than production rendering (on same range)

I doubt if you'll get 7 times your performance but it should improve it considerably.

I doubt it too, but even doubling speed would be good.

Taskmanager shows my CPU up at 80-100% during producing.
During HD preview CPU is around 70%

******Summary: In my case (I guess) the CPU is the limiting factor. Hard-drive and RAM are not yet a limiting factor.




.............so now, what graphics card is the best to get?


I guess the key is to get a graphics card which is good with OpenCL. Not sure if the GeForce760 is the best choice for that. Most of the graphic card scores are based around gaming which might be very different from OpenCL.

So here is a link to a site which is benchmarking more on the OpenCL http://clbenchmark.com/result.jsp
If you only select "CPU" I found some surprising OpenCL scores for the various Intel CPUs.


Here is the 4770CPU on-chip HD 4600 : http://clbenchmark.com/device-info.jsp?config=16296429&test=CLB20101

Here the GEForce760 I had in mind: http://clbenchmark.com/device-info.jsp?config=16336907&test=CLB20101

Different tests give different results:
The Radeon HD 7950, scores lower than GeFroce760 on the G3D passmark test http://www.videocardbenchmark.net/gpu.php?gpu=GeForce+GTX+760&id=2561

But the Radeon HD 7950 beats the GF760 by more than double on on this test (if that means anything) : http://clbenchmark.com/device-info.jsp?config=11993914&test=CLB20101


Of course the HD7950 is almost double the price. Some tests indicate that the "Gigabyte GeForce 760" is running rather quiet. So that's what I'll probably go with ........ cheap and quiet.

Based on this thread ( http://forum.cyberlink.com/forum/posts/list/30304.page ) a 4770 CPU encodes faster than a HD7850 graphics card. (see message towards end of thread)

The HD7850 is performance-wise like a GeForce760.
Therefore I wouldn't gain anything on production times. .... and likely I might not reduce choppy pre-view either.

This is in stark contrast to the OpenCL test benchmarks I posted in prior message. According to those a GeForce760 should beat the HD4600 (in the i7-4770 ) hands down.

This is getting confusing......

Any PD12 user out there using a GeForce760 with a motherboard running an i7-4770 CPU ?

I recently built a i7 4770k with two SSDs, 16GB 2400 RAM, and ATI HD 7870 video card. I started out with bare minimum in the machine for stability. Had all kinds of issues with USB3 but that another story. I did install PD12 and rendered several video with the HD4600 using Quicksync. It's very impressive with MP4 file format. I don't use any effects. I mainly use PD12 to edit commercials out of Windows Media files (WTV) so I can stream the files wireless later.

A week or so ago, I did the final case mod that I needed and installed the HD7870 video card and rendered another TV show as before. I saw no significant difference in rendering or HD preview.

Some of the effects now show the ATI icon so I assume those use hardware acceleration if used. If you use a lot of effects, then you might have different results in hardware rendering. Intel i7 4770k, 16GB, GTX1060 3GB, Two 240GB SSD, 4TB HD, Sony HDR-TD20V 3D camcorder, Sony SLT-A65VK for still images, Windows 10 Pro, 64bit
Gary Russell -- TN USA

A week or so ago, I did the final case mod that I needed and installed the HD7870 video card and rendered another TV show as before. I saw no significant difference in rendering or HD preview.

Some of the effects now show the ATI icon so I assume those use hardware acceleration if used. If you use a lot of effects, then you might have different results in hardware rendering.

So today I bought and installed an Asus GTX770.

I then created a short sample project:
- Put 2 clips of Nature.wmv (comes with PD12 samples) in track1
- Applied Colordirector "SciFi" preset to both clips

Preview was set to HD, and was very choppy (=unusable) once "SciFi" was applied.
Rendering was slow , with hardware acceleration (OpenCL) on and off!
When I looked at the GPU usage it was 0 during rendering , all the work was done by the i7 4770 CPU running at 50%

Am I missing something? Certainly looks like other people have the same experience (with other cards).
So what's the point of having a powerful GPU for PD12 if all the work is done by the CPU ?

Well. I got 4 weeks to test and return the GTX770 ..........stay tuned


(BTW: As I understand it PD12 will NOT use CUDA if OpenCL is available on your Graphic card. )

Hi,
The strain on the system comes in when you have a full length video in the timeline and effects and such added and begin the final process of producing and writing to disc and folder in the various formats available, also work in with your own camera footage.
Jim Intel i7-2600@3.4Gz Geforce 560ti-1GB Graphic accelerator, windows 7 Premium 12GB memory

Visit GranPapa64's channel for your YouTube experience of the day!

I've also got a new i7-4770k with a GTX 780Ti card, all 540MB/s SSDs and an ASUS Z87 mobo with Thunderbolt 2 (20Gbps) ports. I can watch the full screen preview at actual speed on a second monitor in Full HD - but ONLY if there are no effects or transitions. As soon as there's any extra rendering work, I get the choppy playback.

The whole point of spending almost $3k on a dedicated machine was to improve my editing experience, but so far I'm not happy
I started with a Quadro K4000, but that card was slower than the GTX 770 card I had in my i5 3570K rig despite being more than 2x the price. I returned the Quadro and got the 780Ti, and while the gaming benchmarks are through the roof I only see about a 10% improvement in rendering/producing times.

FWIW, I work with 1920x1080/60p/28Mbps .mts from a Canon HF G30 camcorder. I put together 5 of those clips with 4 simple fade crossover transitions. My test video has a play length of 3m56s. With my CPU overclocked to 4.3GHz and using the brand new nVidia 334.89 WHQL drivers, the time to produce H.264 AVC 1920x1080/60p/28Mbps .m2ts video is as follows:

Hardware Acceleration: Enable hardware decoding = ON, Enable Open CL = OFF; Fast Video Rendering: Using Hardware video encoder, NO preview: 2m22s or 1.66x actual speed
Same hardware settings but with Enable Open CL = ON, using Hardware video encoder:
2m38s or 1.49x actual speed (11% SLOWER than with CL turned OFF!)
Same hardware settings and Enable Open CL = OFF, using SVRT:
3m37sor 1.08x actual speed (53% SLOWER than GPU)
Same hardware settings and Enable Open CL = ON, using SVRT:
4m04s or 0.88x actual speed (88% SLOWER than GPU)

Basically, it looks like using Open CL slows every down every rendering combination and should NOT be checked! Maybe there are benefits on older or slower cards, but it's definitely a drag on my system.

BTW, Win 8/8.1 does have the Windows Experience built in, you just need to access it via Power Shell. My system is rated as follows: CPU = 8.5; Direct 3D = 8.8, SSDs = 8.1 (the weak link!); Graphics = 8.8; Memory = 8.5.

I realize that anything involving 60p and/or 2k or 4k resolution takes a ton of horsepower, and I was hoping to se a big performance increase moving from an Ivy Bridge to Haswell with AVX2 and hyper-threading. But if this setup can barely keep up with full HD previewing on bare clips only, how will any of us be able to do real editing in real time?

YouTube/optodata

DS365 | Win11 Pro | Ryzen 9 3950X | RTX 4070 Ti | 32GB RAM | 10TB SSDs | 5K+4K HDR monitors

Canon Vixia GX10 (4K 60p) | HF G30 (HD 60p) | Yi Action+ 4K | 360Fly 4K 360°

hellebauer, optodata;

You havn’t by any chance read my thread

http://forum.cyberlink.com/forum/posts/list/32409.page

I have been spending a lot of time trying to find out why PD12 and GTX 770/Win 8.1 are not
working together at all. Several members have tried to help out.
During all trials, I installed Win 8.1 over Win 8, and now PD 12 is worse than ever. Not only is
the display very odd when overlaying videos. Now, the whole program is just about useless;
various windows (PiP etc) takes forever to show up, scrubber is almost impossible to move by hand (have to click to help it moving), the over all lagging/chopping is unbelievable, preview strange, keyframes jumping,
rendering ......well, very slow- jumpy. This is when having just under a minute of mp4’s and an image on timeline.
I have a couple of other video programs on this laptop, they have no problem at all, neither have any
other program installed. As of now, I simply cannot use PD 12.
Installing the new driver yesterday did not make a difference.

Just something.
https://www.petitpoisvideo.com

1Nina, I had seen some of the earlier posts on your thread but haven't experienced the yellow line and hadn't been following it recently.

My PC is a week old and it was built with 8.1 Pro x64 from the start. I agree that the editor experience is not what I had expected with all this power under the hood, especially since the whole point of spending this much money was to be able to do things in PD12 smoothly and without having to wait.

One thing I've noticed is that it takes a VERY long time for some of the transition controls to open. I have some FX from NewBlue and Pixelan as well as Pixelan transitions, and today I timed how long it took to open up the control window for Pixelan's SpiceMaster transition. It ... took ... SIXTEEN (16!) ... seconds for the menu to appear!!!?!!!

The Pixelan FX panel is a little faster, but waiting 11 seconds each time I want to make one adjustment is still very frustrating. The NewBlue FX and the PD transition and FX control panels are very simple, and they open immediately. The other editor functions seem to operate normally and I don't have any issues using the scrubber or waiting for the preview to play.

My previous computer was an Ivy Bridge i5 3570K with a GTX 770 running Win 7 Home x64. I no longer have access to PD12 on it, but I remember the Pixelan wait times were more like 3-6 seconds and I don't know why they're so much slower here.

I hope you find the magic setting that makes everything work like it's supposed to and share it with everyone on the forum . There's a new Beta release of PD12 (2613) and NVidia (334.89 WHQL) out today. They haven't made much difference on my machine, but you may want to try each one to see if either makes a difference on your laptop.

One thing I found earlier was to uninstall the NVidia 3D Vision Driver and 3D Vision Controller Driver, unless you have 3D apps and glasses to watch them. You can uninstall those items from the Programs and Features window or run the nVidia installer again. Each time you install (or reinstall) an nVidia driver, remember to choose the Custom/Advanced option and uncheck those two boxes.

Let me know if that helps!

YouTube/optodata

DS365 | Win11 Pro | Ryzen 9 3950X | RTX 4070 Ti | 32GB RAM | 10TB SSDs | 5K+4K HDR monitors

Canon Vixia GX10 (4K 60p) | HF G30 (HD 60p) | Yi Action+ 4K | 360Fly 4K 360°

Optodata wrote: Hardware Acceleration: Enable hardware decoding = ON, Enable Open CL = OFF; Fast Video Rendering: Using Hardware video encoder, NO preview: 2m22s or 1.66x actual speed
Same hardware settings but with Enable Open CL = ON, using Hardware video encoder:
2m38s or 1.49x actual speed (11% SLOWER than with CL turned OFF!)
Same hardware settings and Enable Open CL = OFF, using SVRT:
3m37sor 1.08x actual speed (53% SLOWER than GPU)
Same hardware settings and Enable Open CL = ON, using SVRT:
4m04s or 0.88x actual speed (88% SLOWER than GPU)

What was your GPU usage AND your CPU usage in above trials?

Maybe we can start a standardized test using Nature.wmv and various PD12 effects ?
And open a new thread like: "Performance Test of GTX760 GTX770 GTX780 in PD12" ?

@1Nina:

I see you are also using "Grass Valley Edius" . How does its preview speed during editing sessions compare to PD12 ? Any noticeable difference in rendering speed ?

When looking at a GPU meter do you see your GTX770 being used in "Grass Valley Edius"?
What about the CPU usage?

When looking at a GPU meter do you see your GTX770 being used in PD12?
What about the CPU usage?


Similar footage and effects should be used for above comparison

I'm glad to see that, no matter how powerful the machine, there doesn't seem to be any way to make this program run smoothly, as it will save me considerable $$ that I may have spent upgrading, as a previous poster has done.

PD 12 provides me with much better videos than the previous program I used, and perhaps the features that enable those results are the same ones that slow things down so much.

In any case I can still perform the tasks I want to without too much difficulty by using render preview and other tactics, so for now I'm not considering returning to "Brand N." Dell Precision 7510 Laptop
Windows 7 Pro 64-bit | Intel(R) XEON(R) CPU E3-1505M v5 @2.80 GHz
RAM: 32 GB
Windows Experience Index 7.5

hellebauer;

the program you mention is on my desktop / Win 7 x 64/GTX 470. The version I have is not compatible with Win8/Win 8.1.

GPU-Z shows the card is working, I believe? I played about 30 sec in preview, and you can see reaction.
Hope it is what you asked for, not sure, really.

optodata;

thanks for giving pointers.
At this time, I will try almost anything to solve some of the troubles.
I found the NVidia 3D Vision Driver and uninstalled it. Couldn't find something called 3D Vision Controller Driver. Where?
I will certainly untick those the next time a new driver is offered.
(BTW; is anyone using 3D vision editing in PD these days?)

I havn't yet had time to really try out PD (without the NVidia 3D Vision Driver). I have just started to see if there are
differences. I will get back with possibly findings.
I do believe that for the time being, PD struggles with the newest Nvidia cards/drivers/Win 8/8.1/x64.



Just something.
https://www.petitpoisvideo.com

Quote:

Optodata wrote: Hardware Acceleration: Enable hardware decoding = ON, Enable Open CL = OFF; Fast Video Rendering: Using Hardware video encoder, NO preview: 2m22s or 1.66x actual speed
Same hardware settings but with Enable Open CL = ON, using Hardware video encoder:
2m38s or 1.49x actual speed (11% SLOWER than with CL turned OFF!)
Same hardware settings and Enable Open CL = OFF, using SVRT:
3m37sor 1.08x actual speed (53% SLOWER than GPU)
Same hardware settings and Enable Open CL = ON, using SVRT:
4m04s or 0.88x actual speed (88% SLOWER than GPU)

What was your GPU usage AND your CPU usage in above trials?

Maybe we can start a standardized test using Nature.wmv and various PD12 effects ?
And open a new thread like: "Performance Test of GTX760 GTX770 GTX780 in PD12" ?

I like the idea of a standardized test, but I don't know if the sample Nature etc. clips will be very taxing on high end systems. I can experiment when I have more time.

I did run GPU-Z (at 10 sec refresh rate) along with Task Manager (% utilization over 4 minutes) showing the graphs for both GPU and CPU under three conditions: HW encode with OpenCL ON and with it OFF, along with SVRT with OpenCL OFF. As you can see in the attached graphs, the utilization profiles are very different but the GPU is barely even used! See attached PNGs

Hardware encoding with OpenCL OFF shows the highest GPU usage and lowest CPU usage, as you'd expect, but while GPU-Z shows the "Video Engine Load" as high as 99%, the "GPU Load" barely budged. Does anyone know the difference?

SVRT was also not surprising, as it shows all 8 CPU virtual cores at maximum almost the entire time. I was surprised to see that the GPU Load and Video Engine Load graphs were also very active during SVRT, with the GPU Load even higher than during hardware encoding.

The Hardware encode with OpenCL ON was very interesting, because there was 0 GPU activity and the CPU cores were operating at about 50%. If the GPU wasn't doing anything and the CPU wasn't being worked very hard, than how did the processing get done nearly as quickly as when the GPU was doing all the work?

YouTube/optodata

DS365 | Win11 Pro | Ryzen 9 3950X | RTX 4070 Ti | 32GB RAM | 10TB SSDs | 5K+4K HDR monitors

Canon Vixia GX10 (4K 60p) | HF G30 (HD 60p) | Yi Action+ 4K | 360Fly 4K 360°

Optodata wrote:

The Hardware encode with OpenCL ON was very interesting, because there was 0 GPU activity and the CPU cores were operating at about 50%. If the GPU wasn't doing anything and the CPU wasn't being worked very hard, than how did the processing get done nearly as quickly as when the GPU was doing all the work?

It may be worth it starting a 2nd instance of GPU-Z to also record the load on the HD4600. At the bottom GPU-Z you can select which GPU to record. I guess this is were the processing was done in above quoted case.
(I think that taskmanager.exe is only showing the load on the CPU but not showing the load on the HD4600 GPU (which is internal to the i7 CPU). )

I should be able to contribute some reproducible data some time this coming weekend ..... we also need to be a tad more scientific, otherwise the data will be very confusing. It was not entirely clear which of the 3 graph sets you posted was executing the fastest. (BTW: I'm sure pre-view was off during rendering)

I somehow have a hunch that GTX7xxx cards (or any graphic card) don't buy anyone much in PD12 (assuming an i& CPU) since the work is mostly done in the CPU's HD4600 in OpenCL. Since GTX7xxx also supports OpenCL this is what is used by PD12 (instead of CUDA). OpenCL allows Cyberlink to market their software to wider audience, whereas CUDA optimized software would only cater to NVIDIA owners. As for the performance difference between CUDA and OpenCL, if any, I have no idea .... but different hardware architectures in graphics cards (NVIDIA, ATI, etc) may demand many different OpenCL coding strategies to achieve max. speed. That be a major PIA for Cyperlink. So they might just streamline codebase for the most common element, which is the Intel GPU. .....eh, video editing for the masses !

optodata, the only thing really surprising in your timed results is the SVRT results you posted, they should be seconds, not minutes if your timeline is 5 clips and 4 simple crossfades between them, nothing else. Go to Preferences > Produce and see if by chance you don't have "Allow SVRT on single...." unticked. If so, tick it and run your SVRT test again. If that does not correct the poor performance attach a short clip from the camera so others could have a chance of replicating the poor SVRT performance.

Jeff

Quote: Optodata wrote:It may be worth it starting a 2nd instance of GPU-Z to also record the load on the HD4600.

I have never gotten both GPUs working at the same time. Is there a trick to it? Intel i7 4770k, 16GB, GTX1060 3GB, Two 240GB SSD, 4TB HD, Sony HDR-TD20V 3D camcorder, Sony SLT-A65VK for still images, Windows 10 Pro, 64bit
Gary Russell -- TN USA

Quote: optodata, the only thing really surprising in your timed results is the SVRT results you posted, they should be seconds, not minutes if your timeline is 5 clips and 4 simple crossfades between them, nothing else. Go to Preferences > Produce and see if by chance you don't have "Allow SVRT on single...." unticked. If so, tick it and run your SVRT test again. If that does not correct the poor performance attach a short clip from the camera so others could have a chance of replicating the poor SVRT performance.

Jeff

Holy Mackerel! I've been using PD since version 9 and have NEVER checked that box because of the warning that comes up about not being standards-compliant (see attached png). At your suggestion, I enabled it and the SVRT produce took 60 seconds! That's more than twice as fast as the GPU can do it and almost 3.4x faster than SVRT with the box unchecked. Wow.

Unfortunately, I played the output file using VLC and it has at least two pauses near the end. The first one happens at the 3:40 mark, just after the 3rd transition. The video freezes for 2 seconds but the audio plays normally. The second pause is at 3:48 and lasts for 1 second. The video plays normally in Win8.1's built-in media player, so that seems to point to an encoding or decoding issue - maybe what the checkbox warning is for?

I'll let you guys take a look. The finished file is in this SkyDrive folder. I have MultiAVCHD and it's related utilities and codes installed, but I don't know if that might make a difference in the production or playback of this video.

The project and source files are in the "Project Files" folder on SkyDrive, but it'll take a while for the files to all be uploaded. Feel free to download and produce the project on your end and see what the production times are. I'd be really interested to see if this runs noticeably faster on Win 7, as that might help us figure out where to look next.

YouTube/optodata

DS365 | Win11 Pro | Ryzen 9 3950X | RTX 4070 Ti | 32GB RAM | 10TB SSDs | 5K+4K HDR monitors

Canon Vixia GX10 (4K 60p) | HF G30 (HD 60p) | Yi Action+ 4K | 360Fly 4K 360°

I Wrote:

I somehow have a hunch that GTX7xxx cards (or any graphic card) don't buy anyone much in PD12 (assuming an i& CPU) since the work is mostly done in the CPU's HD4600 in OpenCL

That statement is wrong. I did a quick test and found that some (NVIDIA supported) effects will make the GTX770 GPU work pretty good: I applied "Pencil Sketch 2" to the Nature.wmv clip (drop on clip) and then added "Color Painting" in the effects track. And voila, the GTX770 GPU-load is around 70%. Full HD real time pre-view was a tad choppy. Went to HD preview....smooth! GPU down to 30%. :


Then tried rendering with hardware and OpenCl on(using only one monitor):
1m08 with GTX770
Around 3min with HD4600 (had GTX770 disabled)


Lesson learned tonight: The benefit you get from your GPU depends very much on your type of effect usage of PD12.

More to come .......