Which GPU you have for video rendering doesn't matter as long as you have one

[crystal6tak]

1 hour ago, DocSwag said:

There's a few things with this that you have to take into account 

-This was done in premier. If you used something else (FCP, Sony Vegas, etc.) you could potentially see different results

-This was ONLY an encoding test. Jay didn't test ANYTHING else. So we have no idea what it could look like for other things.

Vegas hasn't updated GPU acceleration support for GTX cards for a LONG time now. Anything 600 series and up aren't fully supported from what I understand.

[DocSwag]

4 minutes ago, crystal6tak said:

Vegas hasn't updated GPU acceleration support for GTX cards for a LONG time now. Anything 600 series and up aren't fully supported from what I understand.

There's is da Vinci resolve or whatever it's called tho  

[goodtofufriday]

Misleading. Depends on the software being used. And just doing encoding is also misleading. Thats one task out of many video rendering task, even time line scrubbing. 

[PlayStation 2]

Really would be wise to test different software.

[Misanthrope]

6 hours ago, huilun02 said:

And when you're not rendering in Adobe?

This is an important point: From what I've seen there's a wide range of inconsistencies among different editing software. This basically translates to replace "Video Editing Workstation" with "Premiere Workstation", "Vegas Workstation" "Final Cut Workstation" "Hitfilm Workstation" etc.

 

Basically you can optimize the build for the specific software you'll be using. Sounds like a chore to me but when you start looking into professional software in general this is usually the way it goes: there's similar decisions to be made when building servers for example that are task specific, software specific even software version specific.

[MeDownYou]

5 hours ago, raphidy said:

IIRC, only some effects from Adobe software are utilize by the GPU computing. If you use Sony Vegas before version 13? It can be CUDA accelerated with keplers cards and newer versions can only use opencl (for processing only and not rendering) and AMD would have an edge in Vegas nowadays.

Vega dominates in Vegas ba dun....

 

I'll see myself out...

 

It's funny to me that he was like wth and never tested other software, yeah Hollywood uses these they also use programs like avid for editing

[Sauron]

I think we should specify this applies to premiere - this doesn't mean all editing programs work in the same way. On top of that, premiere may behave differently depending on what effects you use and other things. From the looks of it, it seems all the gpus he tested finish their work way before the CPU does, so you get the same result regardless. Perhaps with more (or faster) cores the cpu would finish its workload first and you'd actually end up seeing a difference between the gpus.

[DragonTamer1]

40 minutes ago, crystal6tak said:

Vegas hasn't updated GPU acceleration support for GTX cards for a LONG time now. Anything 600 series and up aren't fully supported from what I understand.

Vegas will still leverage GPU for certain effects, but for the most part it is all on the CPU.

 

If I remember correctly, premiere uses the GPU in a weird way where you don't see a difference in the render times but rather in scrubing, preview windows and special effects.

[Doobeedoo]

While I thought to see more difference like with RX Vega still surprising. Also, definitely more tests, cause bottleneck is there. Like Adobe software ey. 

[NumLock21]

No point reading topic, when title contains the spoiler/conclusion...

[alphaproject]

Yea... Adobe and Vegas supposedly stopped supporting after Keplar. The Cuda cores were vastly different. I used to want the 780 just because of that but then... I also play games. After FX used to also use Cuda acceleration but people were complaining their brand new GPU's weren't doing anything and it was a big mess. This was right when people wanted to use the 980 or 980Ti or even 970 but the Cuda's weren't showing up as usable devices. Oh well... I know it barely does anything in Vegas 13 and now Sony has sold it to another company. Sony Vegas as far as we know it is dead. Now... it's just Vegas.... What a shame.

[SpaceGhostC2C]

You have start by asking yourself why a GPU helps with this task in the first place. That will help you think about whether there is an upper bound to how much a faster GPU can help, and whether this bound could depend on the software used. Or the particular job.

[Master Disaster]

48 minutes ago, NumLock21 said:

No point reading topic, when title contains the spoiler/conclusion...

That's generally how headlines work, no point reading the paper when the front page says everything and no point watching the news when all the articles are outlined within the first 2 minutes of the show, right?

[LePawel]

23 hours ago, Jurrunio said:

So you suspect PCIe bandwidth to be a limitation on cards using PCIe 3.0 x16 and you suggest using PCIe 3.0 x4 to make this possible bottleneck even more serious? Seriously how do you do research with logic like that.

If they all bottleneck evenly it's pcie bandwidth, sounds pretty simple to me.. You can't get any faster than a pcie 3.0 x16, but you can get slower. 

[Jurrunio]

3 hours ago, LePawel said:

If they all bottleneck evenly it's pcie bandwidth, sounds pretty simple to me.. You can't get any faster than a pcie 3.0 x16, but you can get slower. 

I dont get your logic. How can you prove something to be the limiting factor by increasing its effect?

[SpaceGhostC2C]

3 hours ago, LePawel said:

If they all bottleneck evenly it's pcie bandwidth, sounds pretty simple to me.. You can't get any faster than a pcie 3.0 x16, but you can get slower. 

What you are saying is like "they are bottlenecked by the CPU. Let's test wirh a slower CPU!". That won't tell you anything. 

[Guest]

Woah, that is quite interesting indeed.

 

Well, its good to know if i ever find myself seeking to develop an adventurous movie project :l.

[LePawel]

8 hours ago, SpaceGhostC2C said:

What you are saying is like "they are bottlenecked by the CPU. Let's test wirh a slower CPU!". That won't tell you anything. 

what I'm saying is "We don't know what the bottleneck is, let's carry out a test for x to confirm or eliminate a factor". This really isn't hard to understand, it's basic research. 

 

 

Edited September 2, 2017 by wkdpaul
Cleaned up

[LePawel]

4 hours ago, Jurrunio said:

I dont get your logic. How can you prove something to be the limiting factor by increasing its effect?

Because there's more than 1 variable? Research 101?

[Jurrunio]

4 hours ago, LePawel said:

Because there's more than 1 variable? Research 101?

I only see pcie bandwidth as the independent variable. What's the other one?

 

4 hours ago, LePawel said:

what I'm saying is "We don't know what the bottleneck is, let's carry out a test for x to confirm or eliminate a factor". This really isn't hard to understand, it's basic research. 

 

 

Your "research" can only show whether PCIe 3.0 x4 is enough, not whether PCIe 3.0 x16 is enough

Edited September 2, 2017 by wkdpaul
Cleaned up

[LukaH]

lets just say the test was lacking any actual depth and investigation. Appreciate the effort but the conclusion "its all the same any gpu will do it" was very anticlimactic. I wish gamers nexus does a follow up with some actual data and answers.

[ARikozuM]

45 minutes ago, Jurrunio said:

I only see pcie bandwidth as the independent variable. What's the other one?

Your "research" can only show whether PCIe 3.0 x4 is enough, not whether PCIe 3.0 x16 is enough

If we do lower the PCIe to x4 and the numbers show us that performance has declined, it could be an indication that the PCIe is a limiting factor. 

 

I still think it's the memory or CPU. 

[dorin]

The Jay's test is absolutely incorrect: Premiere does not GPU accelerate the export as much if you have a simple clip in timeline with no effects or adjustments to it. Heavy projects with multicam clips, color correction, blending modes, transform parameters modified etc, will bring up the GPU load and there there will be differences. The test is flowed and the conclusion is wrong.