How to Unlock Adobe Premiere CS5, CS5.5, CS6 and CC to use almost any
NVIDIA graphics card with CUDA acceleration.
Download Free software to do the unlock for you for both Premiere CS5, CS5.5, CS6 and CC
| More Articles for Adobe Products
|
 Article Updated on 02/19/2014 Software Updated on 02/19/2014 Copyright 2014, Studio 1 Productions Written by David Knarr of Studio 1 Productions FAQ section is near the end of the article. |
|
Starting with Adobe Premiere CS5, Adobe added the Mercury Playback Engine (MPE). The Mercury Playback Engine can use an NVidia video card’s GPU to accelerate playback, effects and rendering. However, Adobe has only “certified” a few video cards from NVIDIA. (See Note 1, About Adobe’s Certified Video Cards). According to Adobe, they only certify video cards that have gone through rigorous testing by Adobe, which takes a lot of time. With all of the different video cards using the NVidia chipset and the variations of these video cards from the various manufacturers, it’s easy to understand why Adobe doesn’t certify more video cards. If you don’t have a certified card or you want to use a non-certified NVidia video card, there is a way to edit a file called cuda_supported_card.txt and add your video card to a list, so Adobe Premiere CS5, CS5.5, CS6 and CC will allow your video card to run with GPU Acceleration. Some people call this a mod, hacking or unlocking the video cards, whatever you choose to call it, this will not make it «certified» in the eyes of Adobe. But, you will be able to use the GPU Acceleration feature of Adobe Premiere CC, CS6, CS5.5 and Premiere CS5. In this article I will show you two ways to add your video card to the list. One, is with a small program I wrote. The other way is by manually editing the file. If you are running other Adobe programs, such as: You can enable these programs to use the GPU, with the program I wrote or again by manually editing the list. There are thousand and thousands of Adobe Premiere CC, CS6, CS5.5 and CS5 users who are using non-certified NVidia video cards without any problems. In fact, we have 8 systems running non-certified video cards without any problems. NOTE: Some people on the different forums will tell you that you can just delete the cuda_supported_cards.txt, however, this can lead to some problems when you use the Adobe Media Encoder. When you delete this file the Adobe Media Encoder will NOT use the GPU. This is why we say to NOT delete the cuda_supported_cards.txt file. Starting with Premiere CC, the Adobe Media Encoder has it’s own cuda_supported_cards.txt file. I recommend using the program I wrote, as it will give you the option to update the cuda_supported_cards.txt list without having to manually edit these files. Note for Laptop Owners: On some Laptops models, you may not be able to use the GPU Acceleration with Premiere CS5, CS5.5, CS6 or CC. Please make sure you read the section on page 2 marked Special Note for Laptop Owners. This article covers a lot of information, not just how to unlock the video. It coves topics such as the video card requirements for the unlock to work, how to setup Premiere for GPU Acceleration, I show some benchmark results, talk about power supply requirements, video card cooling, how to do the unlock and more. Please read the article in it’s entirety. There is also a Frequently Asked Question section at the end of the article. This cover different problems and questions that you may have. Please make sure you read it.
Number 1 — Introduction A) Mercury Playback Engine GPU Acceleration To use the Mercury Playback Engine in Adobe Premiere CC, CS6, CS5.5 and CS5 in the GPU Acceleration mode, you need an NVIDIA graphics cards (aka. video cards) with at least 896 megs of video ram and preferably, one of the newer model video cards. I will explain more on this a little later in the article. When you install Adobe Premiere, and it does not find one of the “certified“ NVidia video cards, the Mercury Playback Engine is set to run in “Software Only” mode, using only the computer’s CPU and not the video card’s GPU. If you don’t have a video card with the NVidia chip set, then Adobe Premiere will run in the «Software Only» mode. The GPU Acceleration is for accelerated playback, GPU accelerated effects, deinterlacing, blending modes, scaling and rendering the previews and final output. Adobe Premiere CS5, CS5.5, CS6 and CC does NOT use the GPU for encoding or decoding the video, only the CPU is used for that. Number 2 — Maximum Render Quality Mode The Maximum Render Quality mode will maximize the quality of motion in rendered clips and sequences. So when you select this option, the video will often render moving objects more sharply. Maximum Render Quality also maintains sharp detail when scaling from large formats to smaller formats, or from high-definition to standard-definition formats. For the highest quality exports you should always use the Maximum Render Quality mode. Whether you are running the Mercury Playback Engine in the software mode or GPU hardware mode, you can turn the Maximum Render Quality mode on or off. Here is how to set the Maximum Render Quality. Basically, I recommend you always set the Maximum Render Quality mode to ON. Please Note: I receive quite a few emails from people who say they don’t see much, if any speed difference between the Software mode and the GPU mode. This is because, when they are testing in Software mode, the Maximum Render Quality mode is set to OFF and is rendering at a lower quality, which makes it run faster. When it is set to GPU mode, the Maximum Render Quality mode is automatically set to ON and rendering in higher quality. PLEASE make sure when you test your video card in Software mode, have the Maximum Render Quality is set to ON, so the test will be equal with the GPU mode. Otherwise, you will not see the true difference in speed between GPU accelerated mode and Software mode. Number 3 — Updates & Video Card Drivers When Adobe releases an update to Premiere, you will need to unlock the video card again, as they tend over write the cuda_supported_cards.txt file, which is the file that holds the name of your video card. Important — Do NOT use auto-updates for the video card drivers, they don’t always have access to the latest version. You need to go directly to the NVIDIA website yourself and download the drivers directly from them. Please go to www.nvidia.com and download the latest drivers. |
Before you install your new NVIDIA video card, you should remove the old video drivers that you were using. On Windows 7 go into the Control Panel and select Programs and Features. Scroll down the list of programs and remove the video driver that you are currently running.
Then power down the computer and install the new NVIDIA video card. Once that is done, power up the computer and download from the www.nvidia.com website the latest driver for your video card and install it. Then Reboot your computer.
DO NOT down load any Beta drivers. They may not be stable. Only download the WHQL drivers.
If you lose the on board sound in your computer, then read the FAQ article on how to fix it. Note: This FAQ article is for Adobe Premiere, but the on board sound fix will be the same.
Number 4 — Decoding
The decoding of video footage is still handled by the CPU and not the GPU. So if you are using heavily compressed video, such as AVCHD or h.264, you will need a fast processor, since the video card won’t help with the decoding of the video. Also, keep in mind, that if you use a non-Mercury Playback Engine enabled plug-in or effect, the GPU on the video card won’t help you with rendering that effect.
Different types of footage, such as DV, HDV, XDCAM, AVCHD, H.264, DSLR footage, RED, etc. will all play a big part in overall editing performance, as the CPU has to do the decoding of the video format. For example, HDV is easier to decode than AVCHD.
Number 5 — CUDA Cores
Each NVIDIA GPU has a certain number of CUDA cores, which is the computing engine in the NVIDIA GPU.
Now some of you might just stop here and think, «I will just buy a video card with the most CUDA cores». While this might seem like a good idea, it may not be for you. There are other factors that are involved than just CUDA cores. We will cover that later, right now I just wanted you to know that different NVidia graphics cards will offer a different amount of CUDA cores.
(Your video card should have 96 CUDA cores or more. I will cover more on this later in the article.)
Number 6 — Video Card Memory
The Mercury Playback Engine requires the NVIDIA video card to have at least 896 megs memory or more. Anything less, the Mercury Playback Engine will NOT work in the GPU accelerated mode. It will work in the Software mode, but you want it to work in the GPU accelerated mode.
Video cards come with different types of memory, such as DDR2, DDR3, and DDR5 type of memory.
Video cards with DDR2 memory will be to slow for the Mercury Playback Engine, causing problems with Premiere. If you have a older video card with DDR2 memory, replace it, plain and simple.
If you already have a video card with DDR3 memory, you may be fine with that.
If you are buying a new video card, the newer video cards come with either DDR3 or DDR5 memory:
- The lower end cards generally come with DDR3 memory
- The mid-level cards can come with either DDR3 or DDR5. In this case, go for the DDR5 version
- The higher end cards come with DDR5 memory.
DDR5 memory is faster than DDR3 memory, when all things are equal.
Below, I ran several tests using different video cards with DDR3 and DDR5 memory. You might be surprised that in one test the video card with DDR3 memory was faster than the video card with DDR5. Doesn’t make sense does it? I will explain after the test results.
Test Number 1
| Video Card | Number of Cuda Cores | Type of Memory | Memory Interface Width |
| GT 240 | 96 Cuda Cores | DDR3 Memory | 128 bit |
| GT 240 | 96 Cuda Cores | DDR5 Memory | 128 bit |
Results: The video card with DDR5 Memory was approx. 40% faster.
Test Number 2
| Video Card | Number of Cuda Cores | Type of Memory | Memory Interface Width |
| GT 545 | 144 Cuda Cores | DDR3 Memory | 192 bit |
| GT 545 | 144 Cuda Cores | DDR5 Memory | 128 bit |
Results: Both video cards produced almost the exact times.
Test Number 3
| Video Card | Number of Cuda Cores | Type of Memory | Memory Interface Width |
| GTX 260 | 192 Cuda Cores | DDR3 Memory | 448 bit |
| GTX 450 | 192 Cuda Cores | DDR5 Memory | 128 bit |
Results: The GTX260 was approx. 25% faster because it offers a much wider Memory Interface Width.
Explaining the results:
Test Number 1 — Both video cards had the same number of CUDA cores and the Memory Interface Width was 128bit on both cards. The only difference here was one video card had DDR3 memory and the other had DDR5. The DDR5 card was clearly faster.
Test Number 2 — Both video cards had the same number of CUDA cores, but look at the Memory Interface Width. The video card with DDR3 memory had a wider, 192 bit width, while the video card with DDR5 memory had only a 128 bit width. In this case, the wider 192 bit memory interface width made up for the lack of DDR5 memory, allowing it to produce almost the exact same rendering times.
Test Number 3 — Once again both video cards had the same number of CUDA cores. However, the the GTX260 with DDR3 memory has a memory interface width of 448 bits. That is over 3 times the memory interface of the GTX450. In this case, the wider memory interface width more than made up for the DDR3 memory speed, allowing the GTX260 to produce rendering time that were approx. 25% than the GTX450 video card with DDR5 memory.
Bottom Line: The older GTS and GTX 200 series of video cards with DDR3 memory came with a wider memory interface width that allowed them to make up for the slower speed of the DDR3 memory. They also had a higher Memory Bandwidth measured in GB/sec.
If you have a older NVidia card, upgrading to a newer one may not gain you that much. It will all depend on the total specs. of the video card.
Just remember the video cards needs to have at least 896 megs, 1 gig or more would be best.
Number 7 — Chart of NVidia graphics Cards
Here is a chart of the different NVidia Graphics Cards. This chart will open up in a separate window so you won’t loose your place here in the article.
It will show you how many CUDA cores are on each card, the Memory Interface Width, the Memory Bandwidth Speed, the Recommend Size of the Power Supply.
NOTE: The specs and power supply requirements listed in the chart are based on NVidia’s web site. PLEASE check with the manufacturer of the video card you plan on purchasing to see what their power supply requirements are.
NOTE: Adobe Premiere CS6, CS5 and CS5.5 does not support more than 1 GPU. So the GTX590 and GTX690, which has dual GPU’s, only 1 GPU and half of the CUDA cores will be used by Adobe Premiere. If you are looking at the GTX590 or GTX690, you would be better off with the GTX580 or GTX680.
Number 8 — Power Supplies
The Power Supply — Before you run out and buy an NVIDIA video card, you need to know how big your power supply is in watts. Different video cards will require that you have a minimum number of watts power supply. So open your computer, if you are comfortable doing that, otherwise, find someone who is.
Look on the label on the power supply for the number of watts it is rated. It may say something like 300 watts, 450 watts or higher. Once you know the watts, then you can select a video card that will work with your power supply.
For example, the NVIDIA GeForce GT440 will work fine with a 300 watt power supply. But, the GeForce GT580 will require a minimum of a 600 watt power supply.
This is why you need to know your computer’s power supply size before you run out and buy a video card. You don’t want to use a video card that your power supply can’t handle. For example, if you decide you really want a GeForce GT580 video card and you only have a 300 watt power supply, then you will need to upgrade your computer’s power supply.
If you don’t want to bother upgrading your power supply, then make sure you stick with a video card that will work with what every the size of your power supply is.
In the chart above, we list a minimum power supply size needed for the each video card. This list above are based on NVidia’s web site recommendations.
We have received email about what the video card box states as the minimum power supply requirement. For example, the MSI N240GT (GT 240) video card with 1 GB of DDR5 memory, the MSI box says it requires a 350 watt power supply. So why does our chart list a 300 watt power supply?
Okay, if you look at the MSI box for the GT240 with DDR5 memory, it does say they recommend a minimum of 350 watt power supply. But, when you open up the installation guide it says the minimum power supply is 500 watts based on a PC configured with an Intel Core2Extreme Qx9650 processor. Wait a minute, the list above says 300 watts. Okay so what is going on here?
According to MSI, the main group of people who buy higher performance video cards are people who use the computer to play games. The GT 240 card falls into this category and the GT 240 card with 1GB of DDR5 memory is capable of being overclocked. (See Note 2 on Overclocking). MSI recommends a 350 watts if you are a gamer and are going to overclock the video card. If you won’t be overclocking the card and you really don’t need to for video editing, MSI said you will be fine with a 300 watt power supply.
If you are running a quad core system, such as one with the Core2Extreme Qx9650 processor, you will generally have a larger power supply than 300 watts any way and most likely it will be 450 to 500 watts or larger. This is why their installation guide recommends a 500 watt power supply. The more powerful the CPU, the larger power supply your computer will have, since the CPU pulls quite a bit of power in watts. For example, the quad core Qx9650 processor pulls around 65 watts and the I7-930 processor pulls 130 watts.
Also, I was informed by several other video card manufacturers that they put a higher minimum power supply requirement on their video cards, since they don’t know what other devices or the number of hard drives you have in your computer. This way, they will be on the safe side.
Number 9 — Video Card Performance and System Performance. PLEASE READ THIS SECTION CAREFULLY.
As I will show you below, your video card performance will vary greatly on your overall computer system. Everything from the type of CPU, the CPU speed, the amount of memory in the computer and more, will all play a part in your over all performance with Adobe Premiere.
Let’s take a look at a couple of benchmark tests using a project in Adobe Premiere. I used the PPBM5 benchmark test with 4 video tracks. This benchmark test was designed to test the performance of you total computer system by using a project for Premiere. Since the benchmark is designed to test your entire computer, I only used the benchmark to test the video card.
Our First test system (AMD FX-6300 Six Core)
- Motherboard — A Gigabyte GA-990FXA-UD3
- CPU — AMD FX-6300 6 Core CPU running at 3.5 Ghz with turbo to 4.1 Ghz
- 16 gigs of Ram — DDR3-1600
- 3 Western Digital Black Hard Drives. SATA 6.0Gb/s, 7200 rpm. One drive is a 1TB hard drive and there are two 500GB hard drives. 64MB cache on all drives.
- Windows 7 Home Edition
- 1 Gig or more of DDR 5 on all video cards tested, except for the GT-640 video card. It had DDR3 memory.
- Each test was run 5 times in the GPU mode and then averaged together.
- To measure the time, I used a stopwatch and the PPBM5 script
- Tests with the Mercury Playback Engine in Software mode was run once.
- You can read about this computer here.
MPE GPU = I was running the Mercury Playback Engine using GPU Acceleration.
MPE Software = I ran the test with the Mercury Playback Engine using Software only.
First Benchmark Test Using CS6 (Updated 03/04/14)
| Video Card | Number of Cuda Cores |
Time Line Render MPE GPU On |
Time Line Render MPE Software |
MPE GPU On Export to MPEG-2 DVD |
MPE Software Mode Export to MPEG-2 DVD |
MPE GPU On Export to H.264 Blu-Ray |
MPE GPU OFF Export to H.264 Blu-Ray |
| GT-240 | 96 | 12 Seconds | 83 Seconds | 241 Seconds | 235 Seconds | 182 Seconds | 178 Seconds |
| GTX-550 Ti | 192 | 9 Seconds | 83 Seconds | 162 Seconds | 235 Seconds | 84 Seconds | 178 Seconds |
| GTX-570 | 480 | 7.6 Seconds | 83 Seconds | 155 Seconds | 235 Seconds | 91 Seconds | 178 Seconds |
| GT-640 (w/DDR3 Mem.) | 384 | 14 Seconds | 83 Seconds | 262 Seconds | 235 Seconds | 196 Seconds | 178 Seconds |
| GTX 650 Ti | 768 | 8 Seconds | 83 Seconds | 156 Seconds | 235 Seconds | 86 Seconds | 178 Seconds |
| GTX 650 Ti Boost | 768 | 7 Seconds | 83 Seconds | 141 Seconds | 235 Seconds | 82 Seconds | 178 Seconds |
| GTX-660 | 960 | 7 Seconds | 83 Seconds | 139 Seconds | 235 Seconds | 81 Seconds | 178 Seconds |
| GTX-670 | 1344 | 6.5 Seconds | 83 Seconds | 135 Seconds | 235 Seconds | 79 Seconds | 178 Seconds |
| GTX-680 | 1536 | 6 Seconds | 83 Seconds | 133 Seconds | 235 Seconds | 77 Seconds | 178 Seconds |
| GTX 750 Ti | 640 | 7.5 Seconds | 83 Seconds | 153 Seconds | 235 Seconds | 85 Seconds | 178 Seconds |
| GTX-760 | 1152 | 6.5 Seconds | 83 Seconds | 135 Seconds | 235 Seconds | 78 Seconds | 178 Seconds |
| GTX-770 | 1536 | 5.5 Seconds | 83 Seconds | 131 Seconds | 235 Seconds | 76 Seconds | 178 Seconds |
| Quadro 4000 | 256 | 9 Seconds | 83 Seconds | 164 Seconds | 235 Seconds | 98 Seconds | 178 Seconds |
Our Second test system (I7-920) Using CS6 (Updated 12/8/12)
- Intel I7-920 Quad Core
- CPU running at 3.05 Ghz
- 16 gigs of RAM
- Dual Seagate 7200 rpm SATA 3.0 GB/s hard drives
- Windows 7 Home Edition
- 1 Gig or more of DDR 5 on all video cards tested, except for the GT-640 video card. It had DDR3 memory.
- Each test was run 3 times in the GPU mode and then averaged together.
- To measure the time, I used a stopwatch and the PPBM5 script.
- Tests with the Mercury Playback Engine in Software mode was run once.
I used the same PPBM5 benchmark project that uses 4 video tracks, as I did above.
| Video Card | # of Cuda Cores | Time Line Render MPE GPU On |
Time Line Render MPE Software |
MPE GPU On Export to MPEG-2 DVD |
MPE Software Mode Export to MPEG-2 DVD |
| GT-240 | 96 | 11.0 Seconds | 110 Seconds | 174 Seconds | 174 Seconds |
| GT-440 | 96 | 11.0 Seconds | 110 Seconds | 174 Seconds | 174 Seconds |
| GTX-470 | 448 | 9.8 Seconds | 110 Seconds | 95 Seconds | 174 Seconds |
| GTX-545 | 144 | 10.8 Seconds | 110 Seconds | 166 Seconds | 174 Seconds |
| GTX-550 Ti | 192 | 10.5 Seconds | 110 Seconds | 159 Seconds | 174 Seconds |
| GTX-570 | 480 | 9.4 Seconds | 110 Seconds | 90 Seconds | 174 Seconds |
| GT-640 (w/DDR3 Mem.) | 384 | 10.5 Seconds | 110 Seconds | 163 Seconds | 174 Seconds |
| GTX-660 | 960 | 9.4 Seconds | 110 Seconds | 88 Seconds | 174 Seconds |
| GTX-680 | 1536 | 9 Seconds | 110 Seconds | 84 Seconds | 174 Seconds |
| Quadro 2000 | 192 | 11.2 Seconds | 110 Seconds | 160 Seconds | 174 Seconds |
| Quadro 4000 | 256 | 10.7 Seconds | 110 Seconds | 156 Seconds | 174 Seconds |
Explaining the Test Results — (Please read this section very carefully!!!!!)
In case you are wondering, Adobe Premiere uses ALL of the CUDA cores on the video card.
Let’s take a look at the results. When looking at the results above, remember the AMD system is a 6 core system, while the I7-920 is a 4 core system. Also please note, the older AMD processors do not have SSE 4.1+ instruction set support, where the newer AMD FX series and Intel processors do have the SSE4.1 instruction set that Adobe Premiere uses. Now this doesn’t mean you can’t use an AMD processor or an older Intel processor, you can, it will perform slower than processors that have the SSE4.1 support.
Timeline Rendering. As you can see in the above benchmark tests, there is a huge difference in timeline render speeds between the Mercury Playback Engine running in GPU mode vs. Software mode on either computer.
Another thing you will notice is there is NOT a big difference between a NVidia card with 96 cuda cores vs. one with 480 cuda cores, when rendering the timeline.
Exporting to MPEG2-DVD format. Here is where things get interesting. You will notice on both systems, that the more cuda cores the faster it is to export to the MPEG2-DVD format with the Mercury Playback Engine (MPE) in the GPU acceleration mode vs. the MPE in software mode.
On the AMD FX-6300 system, you will notice the GT-240 and GT-640 render slower with the GPU turned on. The reason is the AMD FX-6300 processor can render the video faster than the video card can since these are lower end video cards. However, if you have an AMD Dual Core or Quad Core Processor, the GT-240 and GT-640 will render faster than the CPU.
With the Intel I7-920 system, a video card with 96 cuda cores, was a few seconds slower in the GPU mode than if you used the MPE in software mode. The reason for this is, on this Intel system, the CPU could actually export the MPEG2-DVD footage faster than the video card with only 96 cuda cores. Once I put in a GTX-545 with 144 cuda cores, then the video card was able to export faster than the software mode. Just like on the AMD system, on the Intel system, the more cuda cores, the faster the export to MPEG-2 is.
Important Note: When exporting to MPEG-2, the more ram you have the faster the exporting time will be. When I fist did the MPEG-2 rendering test, I tried it with only 8 gigs of memory. Once I upgraded it to 16 gigs, the MPEG-2 test were about 40% faster. By adding more system memory, you can actually speed up the time it takes to export to MPEG-2 DVD with what ever NVidia video card you are using.
Exporting to h.264 format. Just like when I did the exporting test to MPEG-2 format, AMD FX-6300 system could render the h.264 video a few seconds faster than the GT-240 and GT-640 video cards. Again, with a Dual Core or AMD Quad Core system the GT-240 and GT-640 will be able to render faster than these CPUs. But remember, with a higher end CPU the GT-240 and GT-640 will not be able to keep up with the CPU.
Important Note About the NVidia 600 series of Video Cards: The NVidia 600 series of video cards have a lower memory interface than the 500 series. However, due to the new architecture (or design) of the 600 series of video, can turning in slightly better results with lower specs. In addition, they consume less power and run cooler. NVidia slightly crippled the 600 series of video card by reducing the memory interface width and reducing the memory bandwidth (data transfer rate). They increase the number of CUDA cores on the 600 series of video cards, which in some application can make up for the crippling.
Important Note About the NVidia 700 series of Video Cards: The NVidia 700 series of video cards have better specs over the 600 series of video cards. They have a faster GPU clock speed, more cuda cores, a wider memory interface width and a higher memory transfer rate over the 600 series of video cards. The GTX-700 series of video cards are the ones I would go with at this time.
Your results will vary from mine simply because our systems are different.
From the chipset on the motherboard, to the memory chips, type of video card, etc. all of these things will make a difference in the benchmarks. If you run your own tests, make sure you run your tests with the Maximum Render Quality set to ON for both Software mode and GPU mode.
Here are some factors that will come into play for overall system performance, they are:
CPU Cores and Clock Speed — The more CPU cores you have and the higher the clock speed the better. Remember, the decoding of your video is handled by the CPU and not the GPU. Having a newer Intel processor or an AMD FX series processor will help with decoding heavily compressed video formats such as AVCHD and h.264.
Again if you CPU doesn’t have SSE4.1 support, it just means it will decode the video a little slower.
Hard Disk — A rotation speed of 7,200 RPM’s is the minimum you want.
When I put together the AMD FX-6300 system, the motherboard has SATA 6 Gb/sec hard drive connections. I first set it up using the SATA 3 Gb/s internal hard drive and ran ATTO Disk Benchmark program to see what the data transfer rate would be. With the SATA 3 Gb/s drive it gave me 122 MB per second read transfer rate and a 111 MB per second write transfer rate.
Next, I installed a Western Digital Black 1 TB hard drive that has a SATA 6 Gb/sec interface and ran the test hard drive test on the drive. It gave me 161 MB per second read transfer rate and a 153 MB write transfer rate.
So, if your motherboard has SATA 6 Gb/sec drive connection, then you should use SATA 6 Gb/sec hard drives for better performance.
If your motherboard only supports SATA 3 Gb/sec drives, then you should be using SATA 3 Gb/sec hard drives with the largest amount of cache for the best performance with your system.
If your motherboard only supports SATA I/1.5 Gb/sec drives, then your data transfer rate will be much slower. You may want to consider upgrading your computer to get better hard drive performance.
Raid drive systems will improve performance. Also, the amount of cache the hard drive has can make a difference. The more cache the better.
RAM Memory — The ram speed and latency will play a part in the overall performance. Plus, the more memory you have in your computer the better the overall performance will be. Going from 8 gigs to 16 gigs, we saw a speed improvement of about 40% when encoding to a MPEG2 DVD. The more system ram the better when exporting to MPEG-2.
The Video Formant — Different video formats put different demands on the CPU and GPU. For example, if you have an AVI clip on your timeline and you export it to an AVI file with no effects or transitions then the GPU will get around 1% usage. However, if your timeline has the same AVI clip with no effects or transitions and you are exporting to an MPEG2 DVD then GPU will see much more usage.
The Video Card Spec that is most often over looked.
GPU and CUDA cores — While the more CUDA cores your video card has, this size of the Memory Interface Width, the Memory Bandwidth and having DDR5 memory on the video card, the better performance the video card has to offer. Remember, just because the video card has more to offer, doesn’t mean the rest of the computer system will take advantage of it. There are times the video card’s GPU will be waiting on the rest of the computer to feed it the data to process. When the GPU gets a hold of the data, the more CUDA cores the faster it will process the data. Then it hands this data back to the CPU. The wider the memory interface width and the higher the memory bandwidth, the faster the data will move back to the CPU.
Two specs that most people overlook when selecting or recommending a video card are the Memory Interface Width and the Memory Bandwidth. Let’s take a look at the chart below:
| Video Card | Cuda Cores | Memory Interface Width |
Memory Bandwidth | Clock Speed |
| GTX640 | 384 | 128bit | 28.5 GB/s | |
| GTX650 | 384 | 128bit | 80 GB/s | |
| GTX670 | 1344 | 256bit | 192.2 GB/s | 915 MHz |
| GTX760 | 1152 | 256bit | 192.2 GB/s | 980 MHz |
| GTX770 | 1536 | 256bit | 224.3 GBs | 1046 MHz |
The first two video cards listed above, the GTX640 and the GTX650 has the same number of CUDA cores and the same Memory Interface Width. These are the two specs I see a lot of people on the forums talk about. I have also seen a few people say they will give you equal performance since these two specs are the same. That’s is just NOT true.
Just take a look at the column for the Memory Bandwidth. The GTX640 only has a memory bandwidth of 28.5 GB/s, where the GTX650 has an 80 GB/s memory bandwidth. Three times more memory bandwidth than the GTX640.
I ran an informal test using these two video cards. I slowed down the clock speed on the GTX650 to match the GTX640, trying to make things as equal as possible. I rendered the timeline using the PPBM5 benchmark and the result are:
GTX640 took 14 seconds
GTX650 took 11 seconds
Clearly the GTX650 with the higher Memory Bandwidth made a difference, when everything else was equal. But, so many people forget about the Memory Bandwidth spec.
For the next set of cards listed, the GTX670, and the GTX760, I rendered the timeline again and here are the results:
GTX670 took 6.5 seconds
GTX760 took 6.6 seconds
With both of these video cards they have the same memory interface width and the same memory bandwidth. However, they have a different number of CUDA cores and the clock speed is different. What this is showing me is the GTX760 with it’s higher clock speed, but lower number of CUDA cores will give you the same performance as the GTX670.
The last card on the list, the GTX770 is to show you that while different video cards may have the same Memory Interface Width, they may not have the same Memory Bandwidth.
So if you read something on the forums and they are only talking about the number of CUDA cores and the Memory Interface Width, make sure YOU look into the Memory Bandwidth spec before buying a video card.
Please understand, do to all of the various computer configurations (ie. amount of RAM, BUS speeds, hard drive speeds, type of video card, the type of RAM on the video card, the CPU type, the speed of the CPU, etc. ) your performance results will naturally vary from others. This is not due to Premiere, but do to the way your computer is configured and to the video format you are working with.
Bottom Line: Will you see a performance increase? Yes.
How much, will vary with the factors listed above. Now the performance increase I am talking about is between having the Mercury Playback Engine in Software mode vs. GPU Acceleration mode and having the Maximum Render Quality set to ON when comparing between the Software mode and GPU Acceleration mode.
Note: Some people have reported they have only seen a small increase in performance, while others say they have seen up to 12 time faster performance. This is because of the video format, the effects and transitions they are using. Each can play a big part in how much speed increase you will see. If you use a simple video format with no effects or transitions, don’t expect to see a huge increase in performance.
Note: these test results are from our systems, your results will vary from ours, simply because of the hardware differences.
Quadro or GeForce Video Cards
The only reason to use a Quadro video card with Adobe Premiere is if you are using a 10 bit monitor like the HP Dreamcolor or similar or you need SDI output. Otherwise, the Quadro’s are under powered and over priced.
Let’s take a look at some of the Quadro cards.
The Quadro 2000 — This video card only has 192 CUDA cores and a 128bit memory interface. Basically, it is just an GTS 450 with a slower clock speed. In other words, the GTS450 would be slightly fast. Also, the GTX 550 Ti, with it’s 192 CUDA cores and 192 bit memory interface would be faster than the Quadro 2000, due to the wider memory interface and slightly faster clock speed.
The Quadro FX 3800. This card is now 3 generations old and is based on the GTX 260, but with only a 256 bit memory interface. The GTX260 has a wider memory interface at 448 bit and would produce faster results than the FX 3800.
The Quadro 4000 — This video card is based on the same GPU that was used on the GTX 470. However, it performs much slower than the GTX470. In fact, the performance level is like the GTX 460 SE. Even a regular GTX 460 (not the GTX 460 SE version) would give you better performance due to the Quadro 4000 have only 256 CUDA cores, while the GTX 460 has 336 CUDA cores.
Quadro 5000 — This is based on a GTX 465, with a wider memory interface 320-bit memory bus giving it an edge over the GTX 465. However, it would be slower than a GTX 470 or GTX 570..
Quadro 6000 — This video card is on par with the GTX470, although the Quadro 6000 is much more expensive.
I do not recommend a Quadro video card, unless you have a specific program that requires a Quadro video card or if you have the HP Dreamcolor monitor or similar 10bit monitor or you need SDI output from the video card.
You will get better performance for a lot less money with the GeForce cards.
Please continue to the Next Page for more information and the Unlock procedure……
Unlocking the NVIDIA video card will not make it «certified» in the eyes of Adobe. Adobe would like you to use a «certified» NVIDIA video card, because these are the ones they have put through rigorous testing with Premiere CC, CS6, CS55.5 and CS5.5.
There are a thousands of users out there that are using this unlock technique on their NVIDIA cards with no problems at all and that includes us, at Studio 1 Productions.
Mercury Playback Hack Premiere CS5 Premiere CS5.5 Premiere CS6 Premiere CC
Adobe Premiere CS6 Mercury Playback Engine
The MSI N240GT series of graphics cards (including the N240GT-MD512-OC/D5 and N240GT-MD1G) allows the user to adjust both the voltage and the overclocking configurations via the Afterburner overclocking software from MSI to increase GPU clock up to 30%. The core clock can go from 550MHz to 625MHz.
We do NOT recommending overclocking ANY video card when working with Adobe Premiere CS6, CS5.5 or CS5.
If the video card comes with Factory Set Overclocking you will be okay. However, we have experienced, along with other users reporting, that when they would overclock video card, Premiere would crash after using it a few minutes. The crashing can be from high video card temps and/or overclocking the video card to a speed that is beyond what the video card can handle for long periods of time.
Блог ZVi.by 