Skip to content
Why Encoder Compute Efficiency is Being Measured by Everyone

Why Encoder Compute Efficiency is Being Measured by Everyone

The ability to provide a high-quality viewing experience while cutting costs is coming to streaming video ops teams everywhere. Regardless of whether your business model revolves around free ad-supported streaming (FAST) or subscription-based premium content, transcoding is one of the most significant production-related costs that directly impacts video quality and operational efficiency.

Capital Costs (CAPEX): The Foundation of Cost-Efficiency

Capital costs are incurred when purchasing transcoding equipment. To effectively compare different transcoding solutions, it’s essential to normalize these costs using metrics like cost per stream or cost per encoding ladder.

If System B carries a CAPEX cost that is higher than System A, but in the case where System B offers a 10x capacity advantage, the  lower cost per stream will make it the more cost-effective option as there is never an end to OPEX based public cloud pricing. Your usage directly correlates to how much you pay, and when you use more on the public cloud, you are going to pay more.

Example Analysis

Consider two systems, A and B. While System B costs twice as much as System A, but System B offers a 1000% capacity increase, System B’s higher capacity leads to a greatly reduced cost per stream, demonstrating that a higher initial investment can result in significant long-term savings.

Why Encoder Compute Efficiency is Being Measured by Everyone - table 1
TABLE 1: While System B costs twice as much as System A, but System B offers a 1000% capacity increase, System B’s higher capacity leads to a greatly reduced cost per stream, demonstrating that a higher initial investment can result in significant long-term savings.

Operating Costs (OPEX): Power and Storage Efficiency

Operating costs are another crucial aspect of transcoding economics, encompassing both power consumption and storage costs. A more dense system, such as System B, which delivers more streams per unit of power, will offer lower operating costs over time.

Power Consumption

Power consumption is a major factor in operating costs. For example, if both Systems A and B consume the same power but System B delivers more streams, System B will be cheaper to operate. Table 2 illustrates the watts per stream for both systems, highlighting System B’s efficiency.

Why Encoder Compute Efficiency is Being Measured by Everyone - table 2
TABLE 2: If both Systems A and B consume the same power but System B delivers more streams, System B will be cheaper to operate.

Storage Costs

After purchasing transcoding systems, hosting them carries a cost. This is particularly relevant for third-party co-location services, though even in-house data centers incur these expenses. The denser system B proves much cheaper to house and power over five years, as demonstrated in Table 3.

Why Encoder Compute Efficiency is Being Measured by Everyone - table 3
TABLE 3: After purchasing transcoding systems, hosting them carries a cost. The denser system B proves much cheaper to house and power over five years.

Detailed Cost Analysis: Accounting for Output Codecs and Consistency

When evaluating transcoding solutions, it’s essential to consider the output codecs. Some systems may perform well with simpler codecs like H.264 but struggle with more complex ones like HEVC or AV1. Ensure that all systems are evaluated using consistent output parameters to make an apples-to-apples comparison.

Output Parameters

Different vendors might quote throughput at varying frame rates (e.g., 30 fps vs. 60 fps). Ensure that you normalize these values for a fair comparison. If a vendor quotes 60 fps, you can generally double the throughput for 30 fps streams, but always verify these figures in real-world testing.

Codec Complexity

As codecs evolve, they tend to become more computationally intensive. For instance, AV1 offers superior compression efficiency compared to H.264 and HEVC but requires significantly more processing power. Selecting a transcoder that can handle current and future codec demands ensures long-term cost efficiency and scalability.

As codecs evolve, they tend to become more computationally intensive. For instance, AV1 offers superior compression efficiency compared to H.264 and HEVC but requires significantly more processing power. Selecting a transcoder that can handle current and future codec demands ensures long-term cost efficiency and scalability.

Real-World Comparison: CPU vs. GPU vs. ASIC-Based Transcoding

There are three main types of transcoding solutions: CPU-based (software), GPU-based, and ASIC-based. Each has its own cost implications.

CPU-Based Transcoding (Software)

CPU-based transcoding relies on the central processing unit (CPU) of the host machine to perform all the necessary computations for video transcoding. This method, often referred to as software transcoding, uses general-purpose processors to handle the encoding and decoding tasks.

Cost Implications:

  • Initial Costs: Typically, the initial cost can be lower since it utilizes standard server hardware.
  • Operating Costs: Higher due to significant power consumption and cooling requirements. CPUs are not optimized specifically for video processing, making them less efficient in terms of energy and processing speed.
  • Scalability: As demand increases, more CPU resources are needed, leading to higher costs and potentially more servers, which increases space and power requirements.

GPU-Based Transcoding

GPU-based transcoding uses Graphics Processing Units (GPUs) to perform video transcoding tasks. GPUs are designed primarily for rendering graphics, but their parallel processing capabilities make them well-suited for the computational demands of video encoding and decoding.

Cost Implications:

  • Initial Costs: Higher than CPU-based systems due to the specialized nature of GPU hardware.
  • Operating Costs: Moderate, as GPUs are more energy-efficient for transcoding tasks than CPUs but still consume considerable power.
  • Scalability: Better than CPU-based solutions. GPUs can handle multiple streams simultaneously, reducing the number of required servers and lowering space and power usage.

ASIC-Based Transcoding

ASIC-based transcoding employs Application-Specific Integrated Circuits (ASICs), which are custom-designed hardware dedicated to transcoding tasks. These chips are optimized for specific functions, such as video encoding and decoding, providing high efficiency and performance.

Cost Implications:

  • Initial Costs: Can be high due to the specialized nature of ASICs, but this is offset by their efficiency.
  • Operating Costs: Lowest among the three options. ASICs consume significantly less power and generate less heat, leading to lower cooling costs and overall operational expenses.
  • Scalability: Excellent. ASICs offer high-density processing, enabling the handling of many streams in a compact form factor. This reduces the need for additional hardware and associated costs as demand scales up.

Each transcoding solution offers different advantages and trade-offs. CPU-based solutions are flexible and cost-effective initially but incur high operational costs. GPU-based solutions strike a balance with better performance and moderate operating expenses. ASIC-based solutions provide the best long-term cost efficiency and scalability, making them ideal for high-demand, continuous transcoding operations.

continuous transcoding operations. Mayflower Case Study

A real-world comparison involving Mayflower’s use case demonstrated significant cost savings with NETINT’s ASIC-based transcoders compared to CPU and GPU solutions. Table 4 summarizes these findings, showing that the Quadra T2 delivered the lowest cost per stream and highest density, resulting in reduced power and storage costs.

Why Encoder Compute Efficiency is Being Measured by Everyone - table 4
TABLE 4: Quadra T2 delivered the lowest cost per stream and highest density, resulting in reduced power and storage costs.

ASIC vs. CPU Transcoding: In-Depth Comparison

ASIC-based transcoding solutions provide a significant edge over CPU-based systems in both capital and operating expenses. Let’s delve deeper into these comparisons.

Cost Efficiency

ASIC-based transcoders, like those from NETINT, offer substantial savings in both CAPEX and OPEX. Table 5 shows a detailed comparison of the cost per stream and overall five-year costs between ASIC and CPU-based solutions. The findings highlight an 85% reduction in capital costs and up to 90% savings in operating expenses with ASIC-based systems.

Why Encoder Compute Efficiency is Being Measured by Everyone - table 5
TABLE 5: A detailed comparison of the cost per stream and overall five-year costs between ASIC and CPU-based solutions.

Performance and Efficiency

ASIC-based transcoders are designed specifically for video processing tasks, offering higher density and efficiency. This results in lower power consumption and reduced storage requirements. The streamlined design of ASICs allows for more effective use of resources compared to general-purpose CPUs, which must handle a variety of tasks.

Energy Consumption

Table 6 further emphasizes the energy efficiency of ASIC-based transcoders. By reducing the power needed per stream, these systems significantly lower operational costs and contribute to a smaller carbon footprint.

Why Encoder Compute Efficiency is Being Measured by Everyone - table 6
TABLE 6: By reducing the power needed per stream, these systems significantly lower operational costs and contribute to a smaller carbon footprint.

Cloud vs. On-Premises Transcoding: Cost Considerations

While cloud-based transcoding solutions offer quick deployment, they often come with higher long-term costs compared to building your own transcoding infrastructure.

Cloud Cost Analysis

Using AWS’s Elemental MediaLive or EC2 instances for 24/7 operations can be significantly more expensive than on-premises solutions. For instance, the annual cost of using AWS for Mayflower’s system was projected at $33,775,056 over five years, vastly exceeding the cost of an ASIC-based solution.

TABLE 7: The annual cost of using AWS for Mayflower’s system was projected at $33,775,056 over five years, vastly exceeding the cost of an ASIC-based solution.

Practical Applications: Building Your Own Transcoding Infrastructure

For service providers aiming to enhance profitability and competitiveness, building an in-house transcoding system is often the most cost-effective approach. Leveraging ASIC-based solutions not only reduces costs but also minimizes environmental impact due to lower power consumption

Implementation Tips

  1. Evaluate Output Needs: Ensure your chosen transcoder can handle your current and future codec requirements.
  2. Consistent Testing: Conduct real-world tests to verify vendor claims, focusing on throughput and power consumption under your specific operating conditions.
  3. Consider Density: Opt for dense transcoding solutions to maximize efficiency and minimize space and power usage.

Scalability

One of the significant advantages of ASIC-based transcoding solutions is their scalability. As demand for streaming services grows, these systems can easily scale up to accommodate more streams without a proportional increase in costs. This scalability is crucial for businesses that experience fluctuating or rapidly increasing demand.

Environmental Impact

With growing awareness of environmental sustainability, choosing energy-efficient transcoding solutions can also help reduce the carbon footprint of data centers. ASIC-based transcoders, with their lower power consumption per stream, contribute to greener operations, aligning with corporate sustainability goals.

Integration and Compatibility

Another consideration is the ease of integration with existing infrastructure. ASIC-based transcoders are designed to seamlessly integrate with various streaming platforms and workflows, ensuring smooth transitions and minimal disruption to ongoing operations.

Achieving Cost-Effective Transcoding

The economics of transcoding play a critical role in the profitability of streaming services. By understanding and strategically managing both capital and operating costs, businesses can significantly enhance their operational efficiency and cost-effectiveness. Building your own transcoding infrastructure, particularly with VPU-based solutions, presents a compelling case for long-term savings and improved performance.

To explore how NETINT’s VPUs can revolutionize your transcoding operations, schedule a meeting with our experts today. By leveraging advanced technology and strategic insights, you can transform your video delivery systems into a model of efficiency and cost savings.

Picture of Mark Donnigan

Mark Donnigan

is a veteran of the video ecosystem, working with disruptive innovation companies like NETINT to increase video codec standards and streaming video technology adoption. In addition to working at the forefront of building one of the world's first T-VOD services and driving early HEVC and AV1 adoption, Mark contributed actively to the development and growth of the digital locker initiative, Ultraviolet, breaking device-based content walled gardens, allowing consumers to enjoy video on any device, any time, and in any location. As a technologist and ecosystem developer, Mark's work building cloud-deployed and hyper-scale WebRTC, live, metaverse, and cloud gaming applications gives him a unique view of the OTT and video streaming landscape.

WATCH ON DEMAND: ASIC TRANSCODING FOR HIGH-VOLUME PLATFORMS

ACCESS NOW:  ASIC-Based Transcoding
for High-volume Use Cases
Including social media, broadcast, interactive platforms, and service providers