Comments

arrmyslowrdr wrote on 6/28/2012, 6:30 PM
Not in 670 either. I asked in the NVIDIA beta drivers forum even and have been ignored.

I saw in a professional video editing forum that I stumbled upon not long ago that the opinion of Sony GPU acceleration is VERY low, when it works. Some even claim that the speed degrades video quality. All say the processor is much more important. Since I read this I gave up because apparently we are not missing anything except for a process which degrades image quality.

Perhaps the card developers AND Sony realize this and thus the slowness to roll out this "feature", that is apparently not ready for quality applications, any further.
glpride wrote on 6/28/2012, 6:38 PM
I guess I'm being a bit naughty, and trying to nudge our hero's (THE VEGAS TECHS) to work on a solution to get their software talking to my new computer which just happens to have a GTX690 GPU !!!

To be fair - I'm doing the same on Nvidia's forums - but it's not a 'gaming' problem so trying to get heard isn't easy :-(

So if your having a problem with a Geforce card that Vegas is having problems with - let them know too...
glpride wrote on 6/28/2012, 6:47 PM
Sounds interesting - what forum @ armyslowdr ?

I've grown up with Sony for 35 years - TV's. camera's, computers etc .... guess I'm a bit of a Sony geek!!

I will say that their gear is up at the top in most things ..... I hope they can sort this GPU thing out because I've just bought a Sony NEX VG20EH camera!!
Guy S. wrote on 6/28/2012, 6:57 PM
I installed a 680 card; it worked fine, but the performance benefit isn't as great as would be expected - it was perhaps slightly better than the GTX 460 card it replaced - and this was true with Sony and Adobe apps.

If it makes you feel any better, our IT folks installed a $1k Quadro 4000 card in my system and it's not noticeably better than the GTX460 either, so they are putting the 460 back in my system tomorrow.

As far as I can tell from reviews the newest nVidia GPUs look great on paper but don't perform much better (and in some cases are slower) than previous generation cards.
john_dennis wrote on 6/28/2012, 7:03 PM
I read an editorial comment in the most recent maximumpc (hard copy) magazine that nvidia was to start making different versions of their chipset for gaming and workstation class cards perhaps beginning with the 6 series cards. It wouldn't surprise me.

My next hardware upgrade money will go to Intel for a processor upgrade while GPU rendering shakes out with the NLEs. I'll be keeping my GTS450 unless I get a killer deal on a 5 series card.

RE: Tom Halfhill, GPUs Split Personalities, MaximumPC, July 2012 - Page 6
Steve Mann wrote on 6/29/2012, 12:18 AM
nVidia has always had three "tiers" of products. One for the gamers, one for the workstation pros and one for the research organizations like NASA and DOD.

But.

GPU support is through OFX calls by the hosting software (Vegas in this case). There is a finite amount of overhead for the processor to prepare the data for the OFX call. If you have an i7 running 3+ GHz, then it could take longer for the processor to send the data to the GPU via OFX than it would to just do the process in CPU alone.
megabit wrote on 6/29/2012, 1:30 AM
" There is a finite amount of overhead for the processor to prepare the data for the OFX call. If you have an i7 running 3+ GHz, then it could take longer for the processor to send the data to the GPU via OFX than it would to just do the process in CPU alone."

- and yet with both my machines running on i7 @ 3+ / 4+ GHz respectively, my Quadro 4000m/ GTX 580 cards double the speed (both in preview and render)...

Piotr

AMD TR 2990WX CPU | MSI X399 CARBON AC | 64GB RAM@XMP2933  | 2x RTX 2080Ti GPU | 4x 3TB WD Black RAID0 media drive | 3x 1TB NVMe RAID0 cache drive | SSD SATA system drive | AX1600i PSU | Decklink 12G Extreme | Samsung UHD reference monitor (calibrated)

ritsmer wrote on 6/29/2012, 2:36 AM
@ Piotr: Nice.
What media do you use for input and what Motherboard have you got ?
megabit wrote on 6/29/2012, 3:07 AM
Media: 1080/25p (from EX1 and FS100 via nanoFlash), in the 35-100-220 Mbps MXF format.

The laptop is top-end Dell Precision M6600, while the self-build PC is based on the modest Asus P8P67 Pro mobo..


Piotr

AMD TR 2990WX CPU | MSI X399 CARBON AC | 64GB RAM@XMP2933  | 2x RTX 2080Ti GPU | 4x 3TB WD Black RAID0 media drive | 3x 1TB NVMe RAID0 cache drive | SSD SATA system drive | AX1600i PSU | Decklink 12G Extreme | Samsung UHD reference monitor (calibrated)

wilvan wrote on 6/29/2012, 9:08 AM
Dell T7500 workstation double Nehalem W5580 Xeon processors and dual quadro 4000 and 48 Gigs of RAM..
Very little preview performance , very much slower in rendering when GPU turned ON.
16 cores rendering 1440x1080 50i is way faster. Very much faster.

Turned GPU permanently OFF .

Petty only Vegas Pro not at all using ( fast ) RAM unless when preview .

Sony  PXW-FS7K and 2 x Sony PXW-Z280  ( optimised as per Doug Jensen Master Classes and Alister Chapman advices ) Sony A7 IV
2 x HP Z840 workstations , each as follows : WIN10 pro x 64 , 2 x 10 core Xeon E5-2687W V3 at 3.5 GHz , 256 GB reg ECC RAM , HP nvidia quadro RTX A5000 ( 24GB ), 3 x samsung 970 EVO Plus 1TB M.2 2280 PCIe 3.0 x4  , 3 x SSD 1TB samsung 860 pro , 3 x 3TB WD3003FZEX.
SONY Vegas Pro 13 build 453  ( user since version 4 ) , SONY DVDarch , SONY SoundForge(s) , SONY Acid Pro(s) , SONY Cinescore ( each year buying upgrades for all of them since vegas pro 4 )
(MAGIX) Vegas pro 14 ( bought it as a kind of support but never installed it )
SONY CATALYST browse 
Adobe Photoshop  CC 2023
Adobe After Effects CC 2023 & Adobe Media Encoder CC 2023
Avid Media Composer 2022.xx ( started with the FREE Avid Media Composer First in 2019 )
Dedicated solely editing systems , fully optimized , windows 10 pro x 64 
( win10 pro operating systems , all most silly garbage and kid's stuff of microsoft entirely removed , never update win 10 unless required for editing purposes or ( maybe ) after a while when updates have proven to be reliable and no needless microsoft kid's stuff is added in the updates )

AtomicGreymon wrote on 6/29/2012, 9:28 AM
Well of course, lol. But not everyone has such a system. That's 8 cores, though; 16 threads. But GPU acceleration is more of a valid option for those of us with lower-grade CPUs. At least when it works. It may be a while before the GTX 600 series cards are fully supported for acceleration. Adobe only recently updated After Effects CS6's ray tracing GPU acceleration to support 600 series cards; so it may not just be an issue with the nVidia driver in the case of Sony.

What I'd really like to see is the performance from a dual Xeon E7-8870 configuration. 20 cores... it's hard to imagine, but I should think it would be a thing of beauty.
Zeitgeist wrote on 10/16/2012, 11:29 PM
GTX550 with gpu on makes the preview slower than with gpu off in Vegas 12 but not in Vegas 11. However, the preview speed increase is better in Vegas 12 with gpu off than in Vegas 11 with gpu on. So in the end Vegas 12's preview is faster than Vegas 11.
john_dennis wrote on 11/26/2013, 6:39 PM
Updated my previous comment with link to MaximumPC article that is now online for all to see.
wilvan wrote on 11/27/2013, 8:19 AM
@atomicgreymon :

Indeed :

My HP Z820's ( 32 threads 64GB RAM and newest one 40 treads 64GB RAM ) perform well :-)

( GPU = OFF for as well vegas as adobe AE CS6 , CC )

I , however , have to go into internal prefs with vegas in order to allow more than
16 threads and to allow to use more than 32 GB RAM.
( no idea why they retrict this ....... )

The difference between vegas and AfterEffects however is that AfterEffects notices automatically all cores and RAM and pushes all available cores to 100 % when previewing and rendering in order to get the max out of your hardware investment .


From version vegas12 , the RAM is now also used well but the cores are not ( only 16 threads , each thread a few up to 25 % , even after having increased amount of allowed threads in internal prefs ) .

There some very good additional performance ( preview and render ) should be possible . ( am only interested in real-time preview in order to make real-time editing more pleasant )








Sony  PXW-FS7K and 2 x Sony PXW-Z280  ( optimised as per Doug Jensen Master Classes and Alister Chapman advices ) Sony A7 IV
2 x HP Z840 workstations , each as follows : WIN10 pro x 64 , 2 x 10 core Xeon E5-2687W V3 at 3.5 GHz , 256 GB reg ECC RAM , HP nvidia quadro RTX A5000 ( 24GB ), 3 x samsung 970 EVO Plus 1TB M.2 2280 PCIe 3.0 x4  , 3 x SSD 1TB samsung 860 pro , 3 x 3TB WD3003FZEX.
SONY Vegas Pro 13 build 453  ( user since version 4 ) , SONY DVDarch , SONY SoundForge(s) , SONY Acid Pro(s) , SONY Cinescore ( each year buying upgrades for all of them since vegas pro 4 )
(MAGIX) Vegas pro 14 ( bought it as a kind of support but never installed it )
SONY CATALYST browse 
Adobe Photoshop  CC 2023
Adobe After Effects CC 2023 & Adobe Media Encoder CC 2023
Avid Media Composer 2022.xx ( started with the FREE Avid Media Composer First in 2019 )
Dedicated solely editing systems , fully optimized , windows 10 pro x 64 
( win10 pro operating systems , all most silly garbage and kid's stuff of microsoft entirely removed , never update win 10 unless required for editing purposes or ( maybe ) after a while when updates have proven to be reliable and no needless microsoft kid's stuff is added in the updates )

skeeter123 wrote on 11/27/2013, 11:34 PM
woelf:

What setting did you change to use more RAM? Did it help at all?

thanks!

..
Lovelight wrote on 11/28/2013, 1:04 AM
Here's one opinion:

http://wilbloodworth.com/video-encoding-sony-vegas-pro-12/


I have gpu off because the latest Vegas updates killed cuda. Wish I could have it back on for the faster renders & timelines.
OldSmoke wrote on 11/28/2013, 7:35 AM
Again someone build a system without reading the SCS GPU acceleration website.

Proud owner of Sony Vegas Pro 7, 8, 9, 10, 11, 12 & 13 and now Magix VP15&16.

System Spec.:
Motherboard: ASUS X299 Prime-A

Ram: G.Skill 4x8GB DDR4 2666 XMP

CPU: i7-9800x @ 4.6GHz (custom water cooling system)
GPU: 1x AMD Vega Pro Frontier Edition (water cooled)
Hard drives: System Samsung 970Pro NVME, AV-Projects 1TB (4x Intel P7600 512GB VROC), 4x 2.5" Hotswap bays, 1x 3.5" Hotswap Bay, 1x LG BluRay Burner

PSU: Corsair 1200W
Monitor: 2x Dell Ultrasharp U2713HM (2560x1440)

Stringer wrote on 11/28/2013, 9:42 AM
Reply by: Lovelight



I have to wonder what the blogger was reading..They obviously did not read anything at Sony or visit these forums ..

Hulk wrote on 11/28/2013, 12:25 PM
A couple of observations from my benchmarking "project."

Keep in mind that Vegas uses OpenCL for the assembly or compositing of the timeline and that the Sony engineers have full control over this implementation. The better your GPU is at compute the better your preview will be and the faster your timeline will be sent to the render engine of your choice.

IMO Veges multithreading is not as fine grained as it could be and that's why we're not seeing all cores floored at 100%. But Sony has come a long way in this regard and I'm sure they are working on optimizing it now.

I don't know how much control Sony has over the GPU acceleration of the render codecs,which are licensed. Only two render codecs are GPU accelerated, the Sony AVC, and the MC AVC.

Sony obviously has control over the Sony AVC and no one is having problems with GPU acceleration with that codec. It seems to work with all nVidia and AMD cards quite well I think?

But the MC codec was most likely written by MC and uses OpenCL for AMD GPUs and Cuda for nVidia CPUs. I would also assert that it was coded for the cards "of the day" when VP12 was under development. Which would explain why OpenCL render is broken for AMD 7xxx and new cards, and older nVidia cards work better with this render format.

I would think that Sony would be thinking about the following for the next release of Vegas.
1. Continue to work on improving stability and error handling.
2. Code more of the video engine for OpenCL GPU support.
3. Finer grained threading so that more CPU and GPU resources can be fully utilized.
4. Urge render "partner" codecs to implement GPU acceleration and keep them up to date with the latest GPUs and drivers. Or perhaps it might be better to move the GPU rendering engines fully to a more generalized version of OpenCL, akin to the video render engine, which seems to not suffer performance swings among generations of GPUs and drivers as much as the proprietary MC codec using both OpenCL and Cuda? Perhaps that last bit of performance isn't as necessary as it would be better if we could actually use all of the (or most of) the compute of the newer GPUs during rendering as well as compositing the timeline.