Software Mode is Faster than GPU acceleration, WHY?!

I just built my first computer with an i7 4790k and geforce gtx 770 gb graphics card.
For some reason software only mode keeps coming out ahead by about 20% faster in
render and playback times.
Why would this happen!? Someone please explain.

Thanks for the quick reply Kevin.
I read over the article and tried a new clip with some of the effects listed for CUDA acceleration and it played back seamlessly and
rendered 20x as fast with the graphics card.
Would your best advice just be to pay attention to which effects I'm using to determine which MPE mode to use?
Do you think that with new Premiere CC updates, the graphics card will play a bigger and bigger role in playback and rendering?
I guess I'm just feeling like maybe I should have invested more in my CPU and waited on the graphics card.
What do you think?

Similar Messages

Firefox seems/is faster than Safari but why?

I always found Safari to be faster than Firefox. Safari 4 compared to Firefox 3.5 seems slower. Some pages take forever to respond in Safari and don't in Firefox. I also think the screen now refreshes faster in Firefox. I've looked at the specs and Safari is supposed to be much faster but I don't seem to be experiencing it as so.
Although Firefox is nicer for the Mac in its 3.5 iteration, I'm used to Safari having used it since I returned to the Mac in 2003. Also, Firefox doesn't support the same level of services as Safari. In Safari, I'm constantly using services to grab info to SOHO Notes folders.
Any idea why Safari might be performing like this.

Greetings,
The reasons behind your experience could be many, but try this to see if it improves things:
Quit Safari if it's running and open System Preferences>Network.
In the DNS Server tab, make sure IPv6 is turned off, and enter these two DNS servers to replace the ones you currently have:
208.67.222.222
208.67.220.220
click Apply Now and quit System Preferences.
Next, do these things in order:
1. Go to Home/Library/Cookies and delete the Cookies.plist file.
2. Go to Home/Library/Caches/com.apple.Safari and delete the contents of that folder.
3. Go to Home/Library/Caches/com.apple.Safari/Web Page Previews and delete the contents of that folder.
4. If you don't want to use the Top Sites feature, open Terminal and paste in this command:
defaults write com.apple.Safari DebugSnapshotsUpdatePolicy -int 2
5. Go to Home/Library/Caches/Metadata/Safari and delete the contents of that folder.
6. Go to Home/Library/Caches/Safari and delete the contents of that folder.
7. Go to Home/Library/Safari and delete these files (if you have them):
Downloads.plist
History.plist
Form Values
LastSession.plist
WebpageIcons.db
Repair Disk Permissions, then restart Safari and see if the speed is improved.

3.1.3 is much faster than 4.0 why?

I have conducted several speed tests. Its very disappointing. The only time 4.0 is faster is when multitasking and when the Apps are already open and eating battery. Why is 4.0 firmware soooo much slower?

kesajen wrote:
Does anyone know how to go back to 3.1.3? I have 16GB 3GS, running iOS4, and my phone runs VERY slow!! It's so frustrating. I don't really care about multi tasking, or any other goodie that came with 4. I want to go back to the old OS.
Any help out there?
I know how but can't say

GPU Acceleration Explanation?

I am trying to come up with a convenient explanation of GPU acceleration and video editing.
I would like to put out some statements and get comments on them.
The major push for the development of graphic cards is to improve the performance of gaming PCs.
The graphics processing for a video game takes a 3D representation of a simulated world and and renders it into a 2D viewing screen for interactive game play.
The graphics processing for Pr takes compressed HD video data and renders it to smooth reviewable on-screen video for interactive editing.
When the editing is complete the, video processor reads the accumulated edited instructions and HD raw files to produce highly compressed HD video (media) files for DVD production and files to be uploaded for online viewing.
Powerful graphic cards (GPUs) have been developed for the gaming industry.
Video editing application developers are utilizing the computing power of the new GPUs to speed up video editing and media file production.
Without GPU assistance the CPU must do most of the video processing.
When the CPU usage reaches 100% the smooth review becomes choppy and irritating
Choppy review reduces the throughput of the editing session.
With GPU acceleration the GPU does most of the video processing and this allows the CPU to process video faster.
GPU acceleration also speeds up the production of media files.

No, I think it's just an overgeneralization (if the term oversimplification bothers you).
I know enough about GPU programming to be dangerous but I can't hope to "explain" how it all works. Consider that display driver releases easily exceed 100 megabytes in size (!!) to get a hint of how complex this realm is.
Take this as an example: In a CPU resident application you might create a Frame Buffer Object, then render some pixels into it using OpenGL commands and data from other buffers you have prepared, do some manipulations using shader programs written to run on the GPU, then you might combine the data from that frame buffer with other data using the CPU. You might then submit that frame buffer to an OpenCL program for processing. Complex graphics programs such as video games or applications written by Adobe do this and MUCH more all the time.
As a programmer you might not even know nor care where your data is at a given moment. The computer hardware, under control of the display driver, busily and dutifully transfers blocks of data back and forth at eyeblink speed from CPU to GPU to CPU to GPU as it is needed.
The CPU is good at doing things linearly - like making decisions and controlling the flow of the logic - and the GPU is good at doing repetitive things hugely in parallel - such as an operation on each pixel of an entire image. The best modern programs get both working simultaneously on the parts they do best.
And note that we're not even beginning to talk yet about multiple CPUs (multiple cores), or multiple GPUs.
When you think about it, it's kind of a wonder something this utterly complex could ever be made to work.
-Noel

Why do I no longer have Open CL GPU Acceleration, and have Mercury Playback Engine Software Only greyed out instead?

It seems I've lost all gpu acceleration options, and i am unsure why. I am on a mac pro late 2013, AMD fire pro d700 cards. The system was working fine for many months, then last week for the first time, i got the message that says the project was last used with gpu acceleration, but the present system will support MPE Software only because open cl is not available on this system. Any help would be greatly appreciated.

Ask in the Premiere Pro forum.
Mylenium

Preview slower with hardware GPU acceleration enabled than software only?

Sorry if this had been asked before.
But somehow my preview/playback was slower and choppy once I enable the hardware GPU acceleration(I mean I select the 'mercury playback engine gpu acceleration' under Project-Settings-General')
I followed the following tips to enable my graphic card, (thanks David)
http://www.studio1productions.com/Articles/PremiereCS5.htm
I was working on a clip of approx 3 mins long with majority of the footage(1920 and 720) from a DSLR, some 1920 clips from gopro hd2 and some 1440 clips from a P&S camera(scale to 1920 to match the sequence setting). main effects applied (magic bullet looks).
Before I enabled my graphic card and selected it from the project setting, the preview was ok, nearly smooth all the way except some sections with dynamic links, but once I turn it on, the preview became ridiculously choppy, I couldn't believe my eyes, so I turned it off (select mercury playback engine software only), did playback again, it became much smoother!(or just as smooth as it was when it was under that setting, it's just that comparison gave me a bad impression of my hardware gpu acceleration).
btw: I'm running on a low spec laptop
Core™ i7-2670QM 2.2GHz turbo to 3.1GHZ
8GB DDR3 memory
NVIDIA GeForce 555M 1GB graphics
750G HDD(I partition to 150G system drive and 600G media drive) 5400RPM (I know, it's slow)
currently have about 80G and 400G empty space on two drives respectively.
15.6” HD display (max resolution 1366x768)
I know the machine is not a proper workstation to work with CS6, but the hardware gpu acceleration was slower than the software only gpu acceleration was beyond my understanding, not a minor difference we are talking about, it was much much slower!
Have I done something wrong?
Any tips would be appreciated.

Here's my guess.
MB Looks runs on the GPU at all times, though not in the same way as Adobe's 'accelerated' effects. With CUDA off, the graphics card was free to process the Looks effects at full speed. With CUDA on, PP took some of those resources and Looks processed slower.
Remove Looks and replace it with PP's native accelerated effects and you'll see what all the hoopla is about.

Why is JVM faster than CLR?

hi
i wrote a N-body algorithm in both Java and C# (shown below). i executed it using .NET CLR and JDK1.4.1. in JDK it is twice as fast as .NET (on win2000). now i am trying to find out why is it so??
the interesting thing is that i ran some other algorithms like FFT and graph alogrithms, and they are faster in .NET. so i want to find is there some operation in the below algorithm that is making it run faster in JDK.
in general, what can the possible reasons be for JVM to run faster than CLR?
thanks
double G = 6.6726E-11;
double difference = 0.0;
for(int i=0; i<numBodies; i++)
     accelarations[i] = 0.0;
     for(int j=0; j<numBodies; j++)
          if(i != j)
          difference = radii[i] - radii[j];
          if(difference != 0)
          accelarations[i] += masses/(Math.pow(difference, 2));
     accelarations[i] *= G;

Interesting N-Body problem that treats accelerations as scalars.
Anyway, if there is no optimisation for small integer powers in the Math.pow() method, then I'd expect almost all the time is used there or in its equivalent in .NET. Hardly a meaningful test of relative performance.
Try using (difference * difference) instead.
Sylvia.

No GPU acceleration in PPro w/Quadro 4000 using drivers above 4.0.19 on Mac. Any ideas why?

Can someone out there help me figure out why PPro 5.5.2 will not offer GPU acceleration from my Nvidia Quadro 4000 when running any driver higher than 4.0.19? It's the strangest thing, and I'm at my wit's end trying to figure out why I can't update the Nvidia driver without losing GPU acceleration in PPro.
Eric Deuser
Mac OS X Lion 10.7.3
Premiere Pro CS 5.5.2
Nvidia GeForce 8800 GT
Nvidia Quadro 4000
2 x 2.66 GHz 6-Core Intel Xeon
24 GB RAM

Bug report with likely fix: https://bugs.archlinux.org/task/43143

Why is kernel-2.6.9 (OEL-4) faster than kernel-2.6.18 (OEL-5) ?

Hi,
as long as RHEL-5 and then OEL-5 have been released, I have been wondering why my own programs, compiled and run on RHEL-5/OEL-5, are slower than the same programs compiled and run on RHEL-4/OEL-4 on the same machine. This is really barmy since gcc-4.1, shipped with RHEL-5/OEL-5, is very aggressive compiler and produces faster binary code than gcc-3.4.6, shipped with RHEL-4/OEL-4. I verified this hundred times testing both compilers on RHEL-4/OEL-4 and RHEL-5/OEL-5. The 4.1 compiler always produces faster executable on the same OS.
The problem is obviously in kernel-2.6.18. There is something in the kernel (maybe scheduler?) that slows down the execution of programs. But what? I experimented with changing various kernel boot parameters (eg "acpi=off" etc), even tried to recompile the kernel many times with various combinations of config parameters, and nothing helps. Thus, I'm still wondering whether the problem is solvable by disabling one or more config parameters and recompiling the kernel, or is deeply embedded in the main kernel code.
Is there anybody in this forum who experienced the same, say running OEL-4 before migrating to OEL-5?
Here are two examples showing different execution times on OEL-4.5 (kernel-2.6.9-55.0.5.0.1.EL.i686, gcc-3.4.6-8.0.1) and OEL-5 (kernel-2.6.18-8.1.10.0.1.el5, gcc-4.1.1-52.el5.2). The first example is trivial but very sensitive to overal system load and kernel version. The second example is "Sieve of Eratosthenes" - the program for finding prime numbers (CPU bound).
EXAMPLE 1.
/* Simle program for text screen console */
/* very sensitive to overall system load */
/* and kernel version                     */
#include <stdio.h>
int main(void)
    register int i;
    for(i = 0; i < 1000000; i++)
     printf(" %d ", i);
    return 0;
/* end of program */
$ gcc -O2 -o example1 -s example1.c
$ time ./example1The average execution times on OEL-4.5 and OEL-5 are as follow:
Mode      OEL-4.5         OEL-5
real      0m3.141s        0m4.931s
user      0m0.394s        0m0.366s
sys       0m2.747s        0m4.563s
----------------------------------As we can see, the program on the same machine, compiled and run on OEL-4.5 (gcc-3.4.6 and kernel-2.6.9) is 57% faster than the same program compiled and run on OEL-5 (gcc-4.1.1 and kernel-2.6.18), although gcc-4.1.1 produces much faster binary code. Since the times the process spent in user mode are almost equal on both OS, the whole difference is due to the time the process spent in kernel mode. Note that kernel mode (sys) is taking 66% more time on OEL-5. It tells me that "something" in the kernel-2.6.18 slows down the execution of the program.
In the second example OEL-4.5 is also faster than OEL-5, but the differences in execution times are not so drastic as in the first example.
EXAMPLE 2.
/*           Sieve of Eratosthenes           */
#define GNUSOURCE
#include <stdio.h>
#include <stdlib.h>
#define MAX_PRIME_AREA 100000
#define REPEAT_LOOP 10000
int main(void)
    int prime, composite, count;
    char *sieve_array;
    if ((sieve_array = (char *) malloc( (size_t) (MAX_PRIME_AREA + 1))) == NULL)
     fprintf(stderr,"Memory block too big!\nMemory allocation failed!\a\n");
     exit(EXIT_FAILURE);
    for(count = 0; count < REPEAT_LOOP; count++)
     for(prime = 0; prime < (MAX_PRIME_AREA + 1); prime++)
             *(sieve_array + prime) = (char) '\0';
     for(prime = 3; prime < (MAX_PRIME_AREA + 1); prime += 2)
         if (! *(sieve_array + prime) )
          *(sieve_array + prime) = (char) 'P'; /* offset prime is a prime */
             for(composite = (2 * prime); composite < (MAX_PRIME_AREA + 1); composite += prime)
              *(sieve_array + composite) = (char) 'X'; /* offset composite is a composite */
        /* DO NOT COMPILE FOR TEST !!!
        fprintf(stdout, "\n%d\n", 2);
        for(prime = 3; prime < (MAX_PRIME_AREA + 1); prime += 2)
            if ( *(sieve_array + prime) == 'P' )
                fprintf(stdout, "%d\n", prime);
    free(sieve_array);
    return 0;
/* End of Sieve of Eratosthenes */The average execution times on the same machine on OEL-4.5 and OEL-5 are:
MAX_PRIME_AREA     Mode         OEL-4.5         OEL-5
                   real         0m9.196s        0m10.531s
   100000          user         0m9.189s        0m10.478s
                   sys          0m0.002s        0m0.010s
                   real         0m20.264s       0m21.532s
   200000          user         0m20.233s       0m21.490s
                   sys          0m0.020s        0m0.025s
                   real         0m30.722s       0m33.502s
   300000          user         0m30.684s       0m33.456s
                   sys          0m0.024s        0m0.032s
                   real         1m10.163s       1m15.215s
   400000          user         1m10.087s       1m14.704s
                   sys          0m0.075s        0m0.079s
---------------------------------------------------------Does this ring a bell with anyone? Any clue why?
N.J.

An hour? Hard to believe or is your hardware that
old?An hour? That's a super good time for 3 kernel
packages (i686, xen and PAE) with all modules, plus 3
kernel-devel packages, plus debuginfo package of
150-580 MB where smart people at Red Hat decided to
put uncompressed vmlinux image which is necessary for
kernel profiling and debugging. Ah, I had a different kernel make process in mind.
Oracle doesn't ship
debuginfo package. Of course, this is when I build a
"complete suite" of kernel rpm packages using
unmodified spec file. And, to be honest, it takes
much more than an hour, maybe even two hours. Another
thing is compiling single i686 kernel without
building a package. But it also takes at least half
an hour. Anyway the time significantly depends on how
many modules are selected to be built in.That what I was looking for.
What's your time to build a single kernel (which
version?) with default set of modules ? On which
hardware ? I've only access to a root server right now, which is
cat /proc/cpuinfo | grep "model name"
model name      : AMD Athlon(tm) 64 Processor 3700+with about 2GB of RAM
free -m
             total       used       free     shared    buffers     cached
Mem:          2024       1957         67          0        368       1291
-/+ buffers/cache:        297       1727
Swap:         3827         24       3803under
uname -a
Linux base 2.6.22-gentoo-r5 #5 PREEMPT Mon Sep 10 22:32:37 CEST 2007 i686 AMD Athlon(tm) 64 Processor 3700+ AuthenticAMD GNU/LinuxThis is what i did
cd /usr/src/linux
make clean
time nice -n 19 genkernel --lvm2 --makeopts="-j2" --oldconfig all
* Running with options: --lvm2 --makeopts=-j2 --oldconfig all
* Linux Kernel 2.6.22-gentoo-r5 for x86...*
mount: /boot mounted successfully!
* config: >> Running oldconfig...
* config: --no-clean is enabled; leaving the .config alone.
*         >> Compiling 2.6.22-gentoo-r5 bzImage...
*         >> Compiling 2.6.22-gentoo-r5 modules......
real    17m30.582s
user    16m8.480s
sys     1m9.000sWhat could helped here was that I've switched off some modules and (maybe) the use of ccache.
C.

Is write mode faster than append mode?

Hi All,
I have to write some data after selecting from a table. I have to call six procedure to write data. Each of the procedure writes around 100-150MB data.
If I generate six files in write mode it takes around 11 minutes. (using utl_file.fopen('/tmp','csv','w');)
but if i generate 1 file (combining all) using (using utl_file.fopen('/tmp','csv','a');) It has passed 55 minutes an size of file is still around 500 MB and still growing.
Can anyone explain why append is so slow comapred to write ? or there is any other reason .. note that we have lot of space on server.
Regards,
Amit

What you've written does not compute.
On my laptop I can write that much data far faster than what you've indicated you are seeing.
What version number?
How is the data being selected?
How much time does the SELECT take without the WRITE?
When I have large amounts of data to write my method of choice is to concatenate them into a CLOB and then write it out using DBMS_ADVISOR.CREATE_FILE
http://www.morganslibrary.org/reference/dbms_advisor.html

Have functional difference with GPU acceleration and software processing of Mercury Playback Engine?

Used Decklink SDI and Premiere CS6 to testing.
I have a question~Have functional difference with GPU acceleration and software processing of Mercury Playback Engine?
(Can GPU acceleration export same used 59.94i external monitor export?).
IN specification，Is it need“software processing”？ whether other restrictions?
If Anyone know please tell me.
thank you.

Thank you for your help, and sorry about didn't make it clear.
Use Decklink SDI and Premiere CS6 to test.
In GPU mode of Mercury Engine, the signal form SDI is 1080i, but even fields will be missing.
Take a simple test:
Make quick slide for reproduce animationin in quadrangle(1080i 29.97), but it can't paly smooth and same gone to 30P.
If in Software mode, the action can be smooth.
Although use Software mode can be ok,but if GPU does not realize,it will take long time to render.
system info:
macpro 10.8.1
premiere cs6.0.2
blackmagic driver 9.6.4
GTX285
sequence of premiere
Presets/blackmagic/8bit YUV/interlace/59.94i
Have any problems with my System or the GPU mode can no external monitor output in 59.94?
IN specification，Is it need“software processing”？ whether other restrictions?
thank you!

Why is Mac Pro 2.66 only 1.3x faster than 2.7 G5 on CPU intensive stuff?

I produce DVDs so my Compressor DVCam -> MPEG2 encoding is the most time consuming task. Take the MacWorld benchmarks, I was dissappointed the QC 2.66 was a third faster than a DC 2.7 G5 running Compressor.
I would have expected almost 2x as fast, basically halving encoding times. The Mac Pro took 107s vs G5 137s only 1.28x as fast OR put another way jobs complete in 78% of the time taken for the G5.
This is key reason for me to have just sold a G5 DC 2.3...but I'm dissappointed with these early indicators. Would it be reasonable to assume Apple have not optimised Compressor for Intel - surely not at this late stage?
G4 Dual Gigabit Mac OS X (10.4.7) ATI 9800 Pro

Terpstar,
I was wondering if you have had a chance to use Motion yet. I have a MBP, and using Zapfino fonts with SciFi Glow crashes my system every time. I would be interested to see if this is the case on other intel based systems. This has led to a failure of my main logic board twice over the last month. See my thread:
http://discussions.apple.com/thread.jspa?threadID=614641&tstart=25
Also, of the two GB ram I have installed, FCP doesn't seem to utilize more than 100MB of RAM. Although the VM size is several GB for the app. I noticed that in order to utilize both cores on my MBP, Airport had to be turned off.
Also, as Ned Snowing was saying, there is no doubt that there are going to be many software bugs that must be sorted out. Especially since this program is being adapted for intel macs, and not re-written.

How can I open a project that has been created with Mercury Playback Engine GPU Acceleration software, on my trial version of Premier Pro CC?

Hi,
I'm working with a colleague who is using Mercury Playback Engine GPU Acceleration software on his PC version of Premier Pro CC. I am using a mac and have a trial version of Premier Pro CC. When I try to open a project file that he has sent me I get an error message-
"Missing renderer: This project was last used with Mercury Playback Engine GPU Acceleration (CUDA), which is either not available or not certified on this system. Mercury Playback Engine Software Only will be used."
When I click OK and locate the files I get a message saying "File format not supported".
Can anyone tell me what to do so that we can share project files?
Many thanks,
Jessie

Shouldn't really be the MPE at fault here ... what is the codec of the footage/sequence/project?
Second, can you create a new project in PrPro, then in the media browser, import that sequence from the other project?

Mercury Playback GPU acceleration in CS5.5 not available, Mercury Playback Software only available

Getting a warning screen each time I open an existing project in PP CS5.5 that says GPU acceleration is not available on my system...when it has always in the past. Editing on a HP Z400 with the NVidia Quadro 2000. I can't figure out how to get GPU acceleration going again. The software only option makes for a painfully slow and clunky edit. Any ideas are gladly welcomed. Hot and heavy deadlines looming...you all know the drill. The warning window I get reads "This project was last used with Mercury Playback Engine GPU Acceleration, which is not available on this system. Mercury Playback Engine Software Only will be used." And there is only an "okay" button to close the window. Also when I try to start a new project GPU Acceleration option is not accessible...defaults to the software setting and it's grayed out.

Hi Joe,
Have you updated your video drivers lately? Any OS updates?
Thanks,
Kevin

Why is my 2008 Macbook faster than my 2012 MacBook Pro?

I got a basic, 13Inch Macbook back in 2008. It has been used, nonstop, for the past almost 5 yrs since i bought it. I never turn it off, it always gets used on a bed/couch, tonnes of windows/programs constantly open. Its full of crappy DLed programs, movies, music and whatnot, has only recently been updated to 10.6 OS X. Its dying, the charger barely works, and it over heats and blacksout sometimes and yet, it runs so much faster than my end of 2012 13in MacBook Pro, 2.9GHz model. Slower to the point that i still use my old one and let Hubby take the new one to the Construction site with him. Ive used it maybe once or twice since i bought it in November. Booting up is slower, general use is slow, opening programs is slower. And its got almost nothing on it!
Just wondering why its slower when its newer and supposedly better? I thought i was upgrading?

Wipe the drive on the new system and Reinstall OS X. Factory installs aren't all they are supposed to be. Sometimes corrupted from the Get Go.
Only by wiping the drive and doing a Fresh Clean Install will you know if your slowness was caused by some type of OS corruption or posibly a hardware problem.
Also you don't mention what RPM the drive is in your older Mac but Apple puts really Slow 5400RMP drives in the newer models by default. These Slow 5400RPM drives will slow down the whole system. Slow to boot. Slow to load programs and files into those programs and slow to save back to the drive. Also OS X is constantly writing to and reading from the hard drive so a slower drive will again slow down the whole system.
If it is less then 14 days old you can return it No Questions Asked for a full refund and then maybe try another, different, machine.

Software Mode is Faster than GPU acceleration, WHY?!

Similar Messages

Maybe you are looking for