RAM Requirements for Adobe CS5.5

This article is the second of a series on configuring your Windows workstation for producing with Adobe CS 5.5. The focus of this particular article is the optimal RAM configuration for both a single CPU and dual CPU system for producing with Premiere Pro and Adobe Media Encoder.

Z400.png

Figure 1. The HP Z400, our single CPU, 4/8-core contestant.

As an overview, I tested with two similarly configured systems from HP, who sponsored this testing, one with a 2.67 GHz 4-core CPU (an HP Z400), the other with two 2.67 GHz 4-core CPUs (an HP Z600). Both systems used the same graphics card, an NVIDIA Quadro FX 4800. I ran multiple tests, encoding sequences created from different camera formats, from DV to Red, into multiple outputs, from MPEG-2 for DVD to H.264 for YouTube and Blu-ray. I ran the tests with three different RAM configurations, 6 GB, 12 GB and 24 GB.

Why these configurations? Because the CPUs on these systems have three channels of memory, so it’s most efficient to use a configuration that divides equally by three. With six memory slots on the Z400, the 6 GB configuration used 3 x 2 GB memory DIMMs, 12 GB used 3 x 4 GB DIMMs, while 24 GB used 6 X 4 GB DIMMS. On the Z600, with two CPUs, the memory would be configured as 6 x 1 GB, 6 x 2GB and 6 x 4GB. Certainly you can configure the systems with configurations that don’t divide evenly by 3, but these were the most natural configurations.

To put these configurations in a bit of perspective, the cost difference between 6GB and 12GB on the HP site is $360, while the difference between 12 GB and 24 GB is $960. Not a huge amount of cash in either case, but certainly you’d certainly like to see some performance boost before making the additional investment.

z600-trim.png

Figure 2. The HP Z600, our dual=CPU, 8/16-core contestant.

Except in one or two discrete cases, I didn’t see a whole lot of performance difference in the results. Perhaps this relates to the fact that encoding is a comparably slow operation. That is, if RAM isn’t available, the system will use the hard disk for temporary storage instead. In a real time operation, this could really slow things down, but when encoding in slower than real time, paging back and forth to disk isn’t that much slower than working with RAM. For whatever the reason, in my tests, which involved strictly rendering, I didn’t see much difference in most test cases.

Let’s jump in with the dual-CPU Z600.

The HP Z600

Table 1 summarizes the test results for the Z600. On the extreme right is the difference between the computer configured with 12GB as compared to 6 GB, while the middle column represents the performance difference between the 24 GB configuration and 12 GB. Figures in red indicate a slow down in performance between the two platforms, while figures in green are performance increases in excess of 15%. The obvious thought being is that that any slow down in performance is bad, while you’d need to see some significant benefit to invest the additional cash for more RAM, particularly the $960 to increase from 12 GB to 24 GB.

Z600 summary.png

Table 1: Performance differences between the three RAM configurations for the HP Z600.

As you can see in the table, increasing memory from 6 GB to 12 GB improved performance in all cases, but only crossed the 15% threshold in projects involving Red source footage. Given that this upgrade only costs $360, it seems like a reasonable investment.The same can’t be said for the $960 cost to upgrade from 12 GB to 24 GB, which slowed performance in five of nine test cases, and never accelerated performance more than 3%.

Additional Considerations

Note that my test projects contained a mix of source materials and applied effects. In all instances, when Dynamic Link with After Effects was involved, RAM requirements jumped significantly. For example, Figure 3 shows the memory usage of a DV-based project with no Dynamic Link effects (the screen is from HP’s useful Performance Advisor software that’s included with each workstation). As you can see, memory usage was flat-lined just below 6 GB.

Z600 -DV - nehalem_tr.png

Figure 3. Memory usage for a DV project with no Dynamic Link (image from the HP Performance Advisor software).

Figure 4 shows a DV-based project with a Dynamic Link to After Effects for its greenscreen effect. As you can see on the left, After Effects consumed an extra 6 GB of RAM or so, projecting RAM usage to the 12 GB mark. Notwithstanding this additional memory usage, however, the performance boost from 6 GB to 12 GB was only about 1%.

Z600 -DV - after effects.PNG

Figure 4. Memory usage for a DV project with  Dynamic Link (image from the HP Performance Advisor software).

Other factors that affected memory usage included projects with picture-in-picture effects, which often more than doubled RAM requirements, though again, producing only minimal difference in encoding times between the different configurations. There was little format-related difference in RAM uage, with the exception of the 4K Red footage, which automatically pushed memory useage above 12 GB (Figure 5). As you saw in Table 1, projects using 4K Red footage are the only slam dunks in the decision to upgrade from 6 GB to 12 GB, but 24 GB produced only minimal additional performance boost.

Z600-  Red - no effects_trim.png

Figure 5. Memory usage for a single track Red project with no effects (image from the HP Performance Advisor software).

Let’s take a quick look at the Z400 results.

The HP Z400

Table 2 shows the results for the Z400. Interestingly, there were really no test cases where upgrading beyond 6 GB appeared compelling.

Z400 summary.png

Table 2: Performance differences between the three RAM configurations for the Z400.

For the most part, this can be attributed to the fact that a single CPU system uses less RAM than a dual CPU system. For example, Figure 6 shows the Z400 and the Z600 both rendering the same Red project at the same point in the project. As you can see, the Z400 is consuming around 9 GB of RAM while the Z600 is over 12 GB. While not universal, most comparisons show a similar pattern.

both red.png

Figure 6. Memory usage for a single track Red project with no effects on the Z600 (on the left) and Z400 (image from the HP Performance Advisor software).

The other factor is that as a single-CPU system, the Z400 was slower than the dual-CPU Z600 in virtually all tests. Since the CPU is more of a bottleneck, it makes it even less likely that paging to hard disk rather of RAM would slow overall system performance.

Recommendations

Assuming that your primary consideration is rendering time for the Adobe Media Encoder, in most instances, I would start with 6 GB for a single CPU system, and 12 GB for a dual-CPU system. I would consider adding an additional 6 GB if working with 4K Red footage or projects dynamically linked to Adobe After Effects.

Caveats

Sometimes it’s time to shoot the author and ship the article, wouldas, and shouldas be damned. This is one of those times. If I could start over, though, I would throw in some preview tests to see if RAM made a difference there. I don’t think so, because preview is nothing but a render on the timeline, but I don’t know for sure.

The other concern would be users who run multiple programs simultaneously. If you were editing multiple humongous images in Photoshop, or running After Effects while rendering, either of these programs (and many more) could consume RAM and force more paging to hard disk. The permutations of testing these scenarios, however, can get pretty convoluted, so I didn’t pursue them. However, if you’re running a RAM intensive program in the foreground while rendering in the background, additional RAM beyond the recommended figures could very well help.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


About Jan Ozer

Avatar photo
I help companies train new technical hires in streaming media-related positions; I also help companies optimize their codec selections and encoding stacks and evaluate new encoders and codecs. I am a contributing editor to Streaming Media Magazine, writing about codecs and encoding tools. I have written multiple authoritative books on video encoding, including Video Encoding by the Numbers: Eliminate the Guesswork from your Streaming Video (https://amzn.to/3kV6R1j) and Learn to Produce Video with FFmpeg: In Thirty Minutes or Less (https://amzn.to/3ZJih7e). I have multiple courses relating to streaming media production, all available at https://bit.ly/slc_courses. I currently work as www.netint.com as a Senior Director in Marketing.

Check Also

Five star review for video quality metrics course.

New Five Star Review for Video Quality Metrics Course

The Computing and Using Video Quality Metrics course teaches encoding pro to compute and use video metrics like VMAF, PSNR, and SSIM.

Figure shows the different components to live streaming latency.

The Quality Cost of Low-Latency Transcoding

While low-latency transcoding sounds desirable, low-latency transcode settings can reduce quality and may not noticeably …

NAB Session on AI in Video Streaming

Like most encoding professionals, I’ve followed AI-related streaming advancements for the last few years. I’m …

Leave a Reply

Your email address will not be published. Required fields are marked *