GPUs and flash memory are used in numerous media and entertainment applications today and the number is rapidly growing. The need to store and transcribe huge amounts of data through image and data processing is becoming overwhelming. The more GPUs and flash memory available, the quicker the data can be used. GPU appliances supporting multiple NVIDIA GPUs and Intel Phi coprocessors and Flash memory appliances supporting multiple Fusion-io flash memory cards are quickly becoming the best and most economical way of accomplishing this tremendous feat.
In 2003, the first commercial 4K camera was released. Ten years later, Google released a new video coding format called VP9 specifically for 4K video. A few years from now, most people will have a 4K capable TV in their house. When it comes to filming in 4K, it takes a lot more storage than older video formats. There are a lot of factors to consider when determining how many GB or TB of storage is needed when shooting in 4K: how many cameras, how many hours the cameras will be shooting, frames per second, video compression type, etc. “One hour of standard definition DV footage requires approximately 12.7GB of storage; approximately 217MB per minute. By comparison, one hour of RAW 4K content requires close to 110GB of storage; approximately 2GB per minute.” If a single Hollywood movie has more than 200 hours of raw footage, that equates to more than 22TB of data before any post production work has been done. With the emerging large format 8KUHDTV standard, that storage requirement is expanding even further.
During post production, the raw footage needs to be rendered or edited, which can be done with a variety of GPU accelerated applications. With traditional video storage, the data is stored on SSDs and transferred from PCIe RAID controllers via SATA or SAS back to PCIe for the GPUs, creating a bottleneck during the transfer. It’s better to use PCIe flash memory Instead of using traditional SSDs because the data is being transferred across the PCIe bus eliminating the bottleneck. The OSS Flash Storage Array (FSA) offers up to 200TB of fast flash storage, which allows plenty of capacity for hundreds of hours of raw 4K footage and the vast amounts of edited or rendered footage as well. In addition, because the flash is PCIe based the rendering and editing programs can access the stored footage much more quickly than SATA/SAS SSDs.
Video rendering is the process of using computer programs to create a two-dimensional image from a three-dimensional representation. These images are highly complex as they have to take into account all of the aspects of the object: viewpoint, lighting, texture etc. GPUs are commonly used in rendering because they are able to help a CPU perform complex rendering calculations. In addition to the movie and television industry, rendering is used in architecture, and the video game industry. Many of these rendering processes are computationally intensive.
A lot of rendering servers tout powerful CPUs. But GPUs can really boost the performance of a rendering server. CPUs have a few, very fast cores that are optimized for sequential tasks. But rendering processes consist of a large number of tasks that need to happen simultaneously. When running such a large parallel process like this, GPUs are far more efficient because they contain thousands of smaller cores designed for handling multiple tasks simultaneously. The more GPUs you have available, the more power available for computing rendering calculations. One Stop Systems has the densest solution on the market with the High Density Compute Accelerator (HDCA). If one GPU with thousands of cores can speed up rendering processes, imagine what 16 GPUs could do! You’re probably thinking about sixteen GPUs taking up a lot of space but our HDCA is only 3U. The HDCA can add a huge amount of power to any rendering system.
Here are some rendering applications that can be sped up by using GPUs:
While the HDCA is a great solution for large companies that need a lot of compute power, an individual or small company that does rendering may not need 16 GPUs. Smaller companies often use laptops or desktop computers instead of high end servers. Instead of adding a large appliances to a laptop or computer, it's possible to add a smaller, portable appliance. The One Stop Systems CUBE Expansion Appliances can fit anywhere from one to four (double-wide) GPUs and can be plugged into a computer or laptop. It’s an easy, inexpensive way to boost compute power. The CUBEs can accommodate any PCIe add-in card, so adding a flash storage card, video editing card, audio card etc. would also be possible on this smaller scale.
Here are some rendering applications that can be sped up by using one GPU:
Color grading is the process of changing and enhancing the color of a movie, video, or picture in a digital color suite. There are two different types of color grading systems: hardware-based and software-based systems. The hardware based systems already utilize GPUs because of how powerful they are at performing parallel processes. However, many of the hardware based systems are incredibly expensive. For example, Da Vinci Resolve had a one GPU based system that was $200,000 and a 16 GPU system that was $800,000.
Software-based systems are much cheaper and can be used on existing infrastructure, or new, reasonable priced infrastructure could be purchased to run the software. As you can imagine, the quality and efficiency of the software based systems are dependent on the hardware used. Da Vinci Resolve’s 16 GPU system would be much more powerful than a server that can only support one or two GPUs. However, attach this server to One Stop Systems’ reasonably priced High Density Compute accelerator, and you can get 16 GPUs connected to one server, in only 3U of rack space. We partner with NVIDIA to provide NVIDIA GPUs for our customers because they are the most powerful GPUs on the market for applications such as color grading.
Here are some color grading applications that can be sped up by using multiple GPUs:
These large scale solutions may not work for individuals and small companies who use laptops or desktop computers coupled with color grading software. Instead of these large appliances, One Stop Systems CUBE Expansion Appliance would be a great solution. The CUBEs are small, inexpensive appliances that fit anywhere from one to four (double-wide) GPUs and can be plugged into a computer or laptop. It’s an easy way to boost compute power without having to buy large, expensive, unnecessary equipment. The CUBEs can accommodate any PCIe add-in card, so adding a flash storage card, video editing card, audio card etc. would also be possible on this smaller scale.
Here are some color grading applications that can be sped up by using one GPU:
Because of the advancements in technology, anyone with a laptop can do video editing. There are many different software programs out there, including ones that are pre-installed on most computers, that can cut, trim, re-sequence, and add sound, transitions and special effects to video. There are a lot of consumers who do it as a hobby, but there are just as many who do video editing as a job. Whether you work at a large firm or independently or for a small company, you want the best equipment so you can create the best possible video or film.
Large video editing companies are more likely to use high end computers and servers for their video editing needs. A video editing server running Adobe Premier Pro CC could use multiple GPUs to add performance and speed. The One Stop Systems Compute Accelerators can fit anywhere from one to 16 GPUs. Adding this product to existing video editing software would help exponentially boost the application’s performance and not take up a large amount of space (1U-3U). In addition to needing compute power for video editing, a company may also need storage. Video files, especially high resolution ones, can end up being very large files and may be slow to load and save. To solve this problem, a large video editing firm could also add an OSS Flash Storage Array to a server and add a tremendous amount of very fast storage to their system. Video files would load and save much faster and accessing multiple files at the same time would be no problem.
These large scale solutions however, may not be possible for everyone in the video editing industry. Individuals and small companies usually use a laptop or desktop computer coupled with video editing software. These applications can often be enhanced using one GPU to boost compute performance and make the editing process faster and smoother. However, GPUs don’t fit into laptops and most high end GPUs won’t fit into a computer or won’t have sufficient power or cooling if they do fit. For most video editors in this situation, 16 GPUS (in the High Density Compute Accelerator) would be too expensive, and unnecessary for the task. Instead, One Stop Systems CUBE Expansion Appliance would be a great solution. The CUBEs are small, inexpensive appliances that fit anywhere from one to four (double-wide) GPUs and can be plugged into a computer or laptop. It’s an easy way to boost compute power without having to buy large, expensive, unnecessary equipment. The CUBEs can accommodate any PCIe add-in card, so adding a flash storage card, video editing card, audio card etc. would also be possible on this smaller scale.
Here are some video editing applications that can be sped up by using one GPU:
Our compute accelerators support from one to sixteen double-wide PCIe cards and can be cabled up to four host computers through PCIe x16 Gen3 connections each operating at 128Gb/s. The all-steel construction chassis house power supplies, fans, and a system monitor that monitors the fans, temperature sensors and power voltages. Front panel LEDs signal minor, major or critical alarms. The compute accelerators are transparent and do not require software except for the drivers required by the PCIe add-in cards. Compute accelerators are the best appliance for applications that require a large amount of compute power.Learn More
Our flash storage arrays support from one to thirty-two single-wide PCIe cards and can be cabled up to four host computers through PCIe x16 Gen3 connections each operating at 128Gb/s. The light-weight, compact chassis house power supplies, fans, and a system monitor that monitors the fans, temperature sensors and power voltages. Flash storage arrays are the best appliance for applications that require a large amount of fast, flexible storage.
To read an independent review of the 3U FSA done by Tom's IT Pro, click here.Learn More
The CUBEs are Desktop Expansion Appliances that have anywhere from one to eight GPUs, Flash Memory, Coprocessors, or any other combination of PCIe Add-in Card requested. The CUBEs can accomodate, full-length (13.25"), mid-length (9.5"), or short PCIe cards (7.5"). All of the various CUBEs provide ample cooling and power. The CUBEs can connect to a laptop or computer via ExpressCard, PCIe or Thunderbolt.They are easy to install and no additional drivers or software are needed other than those required by the PCIe I/O cards.Learn More