@fordie

Looks as if your overclock is working for your CPU-only measurement. GTX560ti doesn't suck, either.

3930k@4.3GHz
Win 7-64bit
VP 12, 486
16GB RAM
All SSDs; Sytem, Render & Projects
Storage is on a RAID1
GTX570 O.C. to 875MHz, driver 275.33 (the fastest in my system)

33sec. with GPU on
160sec. CPU only

How about that:
https://dl.dropbox.com/u/39278380/render-test2010.png

VP 12, 486
Windows 7 64-bit
RAM: 24gb
Intel i7 970 3.2 GHz 12mb cache
GeForce 9600GT

GPU on, 5 secs.
CPU only, 31 secs.

WOW, Robert, how on Earth did you get that??? What is it in the build that gets you those blistering results?

Stunned . . .

Grazie

Are you kidding me? Is that real? Yeah, what on earth...

@rstrong, please go tools - cleanup pre-rendered video and do the test again.

Here's some interesting results, well I think so:-

1] GPU OFF : Set aside RAM 8gb = 2:59

2] GPU OFF : Set aside RAM 0gb = 13:45

3] GPU OFF : Set aside RAM 0.1gb = 3:47

4] GPU ON : Set aside RAM 0.1gb = 53secs

5] GPU ON : Set aside RAM 8gb = 46secs

And now for SHIFT+B:-

6] GPU ON : Set aside RAM 8gb = 50secs

7] GPU OFF : Set aside RAM 8gb = 3:01

So, GPU ON and set aside RAM at 8gb for optimum.

Render = 46 seconds

SHIFT+B = 45 seconds

It would appear everything "render" is now being handled by my nVidia GTX560ti 16gb card. Is this possible? Have SCS got it right?

Cheers

Grazie

ps: Would still love to hear from Rob?

I think it's high time we get a new Rendertest-2013 other wise we'll be measuring in milliseconds.

C'mon John do the honours.

If your entire projects fits into the preview RAM then the second time you render it it will be rendered out of the RAM. For real testing, you need to clear your preview RAM each time you render or set the preview RAM to the default of 200 to have comparable results. But it is fun watching the render to be done in seconds and the only bottle neck is the speed of your drive. I wonder how fast it would render to a RAMDisk.

Old 2500k with slight overclock to 4.2GHz
No GPU in the system.
VP10e

2:36

I guess I am baffled and maybe some of you experts on here can tell me what is my problem.
When I run this test with strictly my CPU and run the test at default ram my test was 211 sec
With GPU and ram default it was 38.
With GPU and ram at 800 which seems to be my sweetspot it was 31 sec.
My problem is that when I do normal rendering of video I do using the GPU is my slowest option. What gives here? In fact it is sometimes 3 times as slow. I am at this point taking HDV 1080 down for DVD which is what everyone still wants. When I do this my best option is to use the CPU. I like the results of the GPU with this though. Any advice on what to do or what is the problem? Thanks.

Which version of Vegas and template you use makes a difference too. Some templates, codex don't do GPU acceleration. AVC templates must be enabled to use GPU acceleration for example. If I leave the default AVC Internet template as it is, Automatic, it will not use the GPU and render times go up. The GTX5xx series are currently the best ones to use for Vegas.

I do most of my rendering to DVD Arch. NTSC Widescreen video stream. Are there any settings I need to change or can change?

I just rendered the rendertest-2010 using VP12, MPEG-2 DVDA NTSC Widescreen video stream template and it took 20sec. Are you rendering video and audio or audio seperate? You should actually render the audio seperate usng the AC3-Pro Stereo DVD template that will import into DVDA without recompression. Audio rendering does take it's time too and it can't be GPU accelerated.

Is Hyperthreading enabled in your bios?

The reason I revived this thread is that I am disappointed in the rendering performance of V11 & V12 vs. V10e and I am disappointed in the lack of high-end CPU & GPU developments in the last (3) years. It seems like all of the R&D is going into low-cost, low-power, CPUs & GPUs for tablets, smartphones, media centers, etc... Even Apple has not updated the Mac Pro line in a timely fashion...

About (3) years ago I purchased a 6-core 980x that when overclocked allows me to render a typical (2) hour HD project in about (2) hours using Vegas 10e which utilizes 100% of my CPU during renders. The exact project in V11 & V12 takes much longer with the same CPU and the results from GPU rendering (GTX570) are hit & miss but never as fast as 10e...

It's time to upgrade my oldest workstation & make it my flagship vs. the 980X (which will serve well for many years - just not as the primary editing rig...) HOWEVER, there just isn't a single CPU option right now that is all that much faster. I have been using Vegas for almost 15 years and this is the first time that a span of (3) years does not allow me to double my rendering speeds for the same/less money than I spent previously. My only option to double rendering speeds is to move to a costly dual Xeon which would cost 5X as much...

The flip side is that I guess I can keep my money & not upgrade for a while... However, I miss the days when I edited in SD only and a simple processor upgrade doubled or tripled my rendering speeds... It was great the first time a 2 hour SD project rendered in under 45 minutes. Those were the days...

I feel totaly the opposite. The rendertest-2010 took 105sec in VP10e versus 34sec in VP12. I still have 10,11 and 12 on my PC. I dont see a GTX570 in your system specs. Did you recently upgrade your computer? For me GPU acceleration is stunning in terms of improvement; I guess that shows how different systems can be.

LightADs --@rstrong, please go tools - cleanup pre-rendered video and do the test again.

Okay so I did cleanup pre-rendered video this time........

GPU......18 sec's
CPU......31 sec's
Preview ram was set to 1024mb

Win7 64-bit
RAM: 24gb
Intel i7 970 3.2 GHz
GeForce 9600GT

I agree with TheRhino on the CPU front and I'm totally frustrated on the GPU front. While I like my new system, I didn't feel the bang was worth the trouble. I felt as if I was going through the motions on my four-year upgrade cycle. My expectation is to double my rendering performance.

I wish intel had been successful with the Prescott and could have pushed the clocks much higher wihout burning the house down. (I also wish I hadn't bought 10,000 shares of that dot-com stock in 2001.)

Parallelism, though it has been around for a long time, has not been the perfect solution to increased rendering performance. When the parallelism is spread across dissimilar processors (a.k.a. GPU cores) it becomes even more problematic to the end user. It sometimes reminds me of the days when we first were able to buy CD burners and each drive manufacturer had its own interface card.

> "GPU......18 sec's"
> "CPU......31 sec's"

Robert, are you following the instructions and rendering to MainConcept MPEG-2 / HDV 1080-60i with Video Rendering Quality set to BEST?

I find it hard to believe that your Intel i7 970 6 Core 3.2 GHz is TWICE as fast as my Core i7-3930K Sandy Bridge-E 3.2 GHz! Also I can't believe that a GeForce 9600GT doubles your performance.

Can you explain a bit more about your setup? Are you overclocking or did you not follow the render test instructions?

~jr

I don't have a GTX 570 installed on my 980X system, but I recently built a new workstation for a client/friend. He liked my 980X setup so we went with the newer 3930K & GTX 570. After it was all said & done his system is only about 20% faster than my 3 year-old 980X. (I went light on the overclocking but chose the best GPU drivers based on others' results...)

Until he is able to find a license for V10 we could only test V12. We found that it is actually SLOWER than my (3) year-old 980X system running V10e. V10e uses 100% of my CPU during rendering whereas V12 only uses 60%... His GPU does OK but it does not work with every codec & he has to overcome the learning curve of when to use the GPU and when not to. IMO this is frustrating...

I just thought that (3) years later I could spend the same $2,000 on a system and have 1.5X to 2X the performance. I have (4) workstations in our studio and the slowest one is only a Q6600. Every few years I remove the guts from the slowest workstation & make it my main editing rig. This is the longest I have gone between upgrades.

Hi JR,

Yes, I followed the instructions exactly as the readme that DSE provided, which seem to be fairly simple. But am I missing something?
I copied the file from the timeline in VP10e and pasted it into VP12. Was that okay?
When I rendered in VP10, I got a GPU time of 130 sec's. Something must be different when I render in VP12.

As far as my system goes, it's a home build, and very simple. I don't do any overclocking ( I really don't know how). The only thing really custom is the water cooling system, which is refrigerated. I water cool the ram, the GPU and chipset, as well as the CPU & HDD's. Water temps are approx. 50 degrees F. But I really don't know if cooling makes a difference.
I delete all pre-rendered files, and minimize the preview window before rendering, as well as doing a reboot before each test.
I'm surprised the video card performs so well, I thought it would need something newer to make a difference.

As fas as I remember the Max Number of Rendering Threads has a significant influence on the rendering time.

Kind of that a high number (some 10-12) was best for CPU alone - while a low number (around 2-4) gave an optimum for GPU assisted rendering.

So - when the New Rendertest is done - the testers maybe should also note the Max Number of Rendering Threads that gives the best result for their setup.