The Intel Problem: CPU Efficiency & Power Consumption

The Intel Problem: CPU Efficiency & Power Consumption

Introduction and Comparison of CPUs

The speaker discusses the efficiency and performance of the 14700 K and 7800 X CPUs, particularly in gaming workloads.

Efficiency of 14700 K vs. 7800 X

  • The claim that the 14700 K is less efficient than the 7800 X is considered ludicrous, unless specifically referring to gaming.
  • When both CPUs are set at the same wattage, the 14700 K outperforms the 7800 X in every workload.

Exploring Power Consumption and Efficiency

The speaker addresses a tweet received about declaring the 7800 X as the best overall CPU of 2023. They discuss considering factors beyond all-core work for power consumption and efficiency.

Considerations for Power Consumption

  • There is more to discuss regarding power consumption and efficiency beyond all-core work.
  • The tweet, although defensive, raises a valid point worth exploring scientifically.
  • The goal is to examine how AMD vs. Intel lineup changes in terms of power efficiency when looking at gaming workloads.

Testing Methodology for Power Efficiency

The speaker explains their testing approach for evaluating power efficiency between AMD and Intel CPUs in various workloads, including gaming.

Testing Approach

  • A scientific approach is taken to determine power efficiency differences between AMD and Intel CPUs.
  • Tests include gaming power consumption without constraints, ISO power or FPS matching, and normalizing for the same power consumption or frame rate between competing chips.
  • Additional tests focus on spikier behaviors during game loading or asset rendering that can significantly impact CPU power consumption.
  • Previous testing mainly focused on fully loaded cores but did not cover real-world use cases; this new approach aims to address that.

Support for the Testing Effort

The speaker mentions a special PMD interposer they acquired and thanks viewers for supporting their testing efforts.

Support from Viewers

  • A special PMD interposer was purchased from Elm more labs to aid in the testing process.
  • The interposer took a month to complete and was entirely funded by viewers and supporters.
  • Viewers can support the content by purchasing the brand new gn15 allover print component design mouse mat from store.gamersaccess.net.

Challenges of Gaming Power Consumption Testing

The speaker discusses the challenges of including gaming power consumption in their testing methodology due to game variability and fluctuating power consumption within games.

Challenges of Gaming Power Consumption Testing

  • Gaming power consumption is challenging to measure due to game-to-game variability.
  • Different games exhibit different power consumption levels, even when fully CPU constrained.
  • Power consumption can vary within gameplay itself, depending on different scenarios such as CPU or GPU bound situations.
  • Long-term plans involve selecting representative games for gaming power consumption testing and integrating them into future methodologies.

Metrics for Efficiency Testing

The speaker explains the metrics used for efficiency testing, including FPS per watt, hours chart for Blender rendering, and myips per watt chart for 7zip testing.

Metrics Used

  • FPS per watt is measured both unrestrained (out of the box) and with controlled FPS or power in specific tests.
  • Baseline metric prioritizes out-of-the-box performance since it represents how most people use their devices.
  • A hours chart is provided for Blender rendering, including the 14700 K with controlled power to assess efficiency improvements.
  • A myips per watt chart is used for 7zip testing to evaluate efficiency based on millions of instructions per second per watt.

The transcript provided does not cover the entire video.

Efficiency Testing for Compression and Decompression Workloads

This section discusses the efficiency of CPUs in handling compression or decompression workloads, which can be more demanding than running games. The focus is on how different CPUs perform in terms of power consumption when pulling power through EPS 12vt cables.

Efficiency Testing Results

  • The AMD 7800 x3d CPU is highly power efficient, producing almost 10 frames per second (FPS) on average per watt of power consumed.
  • The AMD 7950 x3d and 5800 x3d CPUs also show strong results with around 8 FPS per watt.
  • Intel's I3 1200f CPU performs well in terms of frame rate but is less efficient compared to AMD's offerings.
  • Intel's best performer, the 4900 K CPU, has a high frame rate but lower efficiency compared to AMD CPUs.

Equalized FPS Testing at Locked Frame Rate

This section focuses on equalized FPS testing at a locked frame rate of 144 FPS. By limiting the frame rate, it allows for direct comparison of power consumption between CPUs without being influenced by external limitations.

Equalized FPS Testing Results

  • Both the AMD 14900 K and 7800 x3d CPUs have equalized FPS entries.
  • Even with equalized FPS, AMD maintains a significant advantage in efficiency over Intel.
  • The AMD CPU achieves an average of 4.9 FPS per watt, while Intel only reaches 1.9 FPS per watt.
  • Although the advantage has reduced compared to previous metrics, AMD still maintains a significant lead over Intel.

Addressing External Component Limitations

This section addresses one of the challenges in benchmarking, which is external component limitations. By using frame rate limits, the testing ensures that neither the CPU nor GPU is fully bound throughout the entire test. This helps provide more realistic and comparable power consumption results.

Addressing External Component Limitations Results

  • The equalized FPS testing eliminates concerns about external component limitations.
  • AMD CPUs still maintain a significant lead in efficiency over Intel, even with equalized FPS.
  • Frame rate limits help eliminate the influence of external limitations on efficiency testing.

Efficiency Testing with Power Lock

This section focuses on efficiency testing with power lock, specifically for the Intel 14700 K CPU. By limiting its power consumption, it allows for comparison with AMD parts and assesses whether power lock improves efficiency.

Efficiency Testing with Power Lock Results

  • The Intel 14700 K CPU at different power limits shows varying levels of efficiency.
  • When unlimited, it achieves around 1 FPS per watt.
  • With a power limit of 86 watts, it reaches 1.7 FPS per watt, showing an uplift in efficiency.
  • However, even with improved efficiency, the Intel 14700 K remains in the lower half of the chart compared to AMD CPUs.

Additional Testing with Heavier GPU Load

This section discusses additional testing conducted at a heavier GPU load by running Phantom Liberty at 1440p Ultra settings. The purpose is to observe how both CPUs perform when the GPU becomes a limiting factor.

Additional Testing Results

  • At higher GPU load, both CPUs' frame rates drop significantly due to GPU limitations.
  • The AMD 7800 x2d remains more efficient than the Intel 4900 K CPU under these conditions.

Due to limited information provided in the transcript beyond this point, further sections cannot be created.

New Section

This section compares the graphics efficiency and cost of different CPUs, specifically focusing on Intel and AMD processors. The comparison is done using frame rates in the game Starfield.

Comparing Graphics Efficiency and Cost

  • The chart shows that Intel leads in frame rate compared to AMD's x3d CPUs in Starfield.
  • Despite running at lower frame rates, AMD CPUs are more efficient than Intel CPUs.
  • The 7800 x2d has a higher efficiency with a result of 1.9 FPS per watt.
  • The I3 1200f also performs well in terms of efficiency.
  • The 14900 K gets 0.6 frames per watt.

New Section

This section discusses the impact of power consumption on frame rate calculations and compares the performance of Intel and AMD CPUs in Starfield.

Impact of Power Consumption on Frame Rate Calculations

  • Power consumption can bias frame rate calculations in favor of Intel if it tips the frame rate higher.
  • Despite running at lower frame rates, AMD CPUs remain more efficient than Intel CPUs.
  • The 7700x is less efficient due to its position on the volt frequency curve.

New Section

This section explores GPU scheduling and how it affects CPU efficiency for both Intel and AMD processors.

GPU Scheduling and CPU Efficiency

  • In this test, Intel is GPU-bound, meaning the GPU is busy almost the entire duration of the frame creation process.
  • However, this doesn't mean that the CPU isn't busy; it just means that the GPU is busier than the CPU.
  • Windows Telemetry may be encouraging Intel to continue blasting power at an elevated power state.
  • AMD's GPU busy time was not as bound up as Intel's in Starfield.

New Section

This section further examines GPU scheduling and power consumption for Intel and AMD CPUs in Starfield.

GPU Scheduling and Power Consumption

  • The CPU remains busy and in a boosted power state, even when the GPU is busy.
  • Intel's power draw remains high, potentially due to tasking the GPU with work like draw calls.
  • AMD, despite being in a similar GPU-bound state, does not exhibit the same power behavior as Intel.

New Section

This section analyzes AMD's GPU busy time in comparison to Intel's in Starfield.

Comparison of AMD and Intel GPU Busy Time

  • In Starfield, AMD's GPU was not as bound up as Intel's, indicating that it may be more efficient at handling workload distribution between the CPU and GPU.
  • The difference in power behavior between AMD and Intel could be attributed to various factors such as Windows Telemetry or architectural differences.

New Section

This section retests CPU efficiency by restricting frame rates for both the 7800 x3d and 14900 K CPUs.

Retesting CPU Efficiency with Restricted Frame Rates

  • Restricting the frame rate to 60 FPS reduces the efficiency of both the 7800 x3d (from 1.9 FPS per watt to 1.4) and the 14900 K (from 0.6 FPS per watt to 0.5).
  • Despite producing the same work and in-game experience, Intel draws more total EPS12V power than AMD.

New Section

This section summarizes the findings from testing CPU efficiency in different scenarios.

Summary of CPU Efficiency Findings

  • In this specific test scenario, whether Intel is GPU-bound or CPU-bound, it is less efficient than AMD in Starfield.
  • The x3d CPUs from AMD show higher efficiency compared to Intel's CPUs.
  • In Baldur's Gate 3, the 7800 X and 7950 X have the highest FPS, followed by the 4900 K.
  • The efficiency advantage again goes to the x3d CPUs from AMD.

New Section

This section discusses performance and efficiency in Blender using different CPUs.

Performance and Efficiency in Blender

  • The 14700 K outperforms the 7800 xvd when run in a configuration that most users won't use.
  • The 14700 K produces a longer render time but becomes more efficient when limited to a specific power budget.
  • In terms of compression performance represented as mips per watt, the 7800 xvd leads this chart, followed by the 7950 XD. Intel's 14900 K performs worst in this test.

New Section

This section focuses on Photoshop performance and efficiency with different CPUs.

Photoshop Performance and Efficiency

  • In Photoshop, the 7800 x2d outperforms the restricted 14700 K in overall score and efficiency points per watt.
  • Efficiency points per watt correlate to Watt hours as this test is largely time-based.

New Section

This section examines high averages for specific intervals throughout the benchmark suite.

High Averages for Specific Intervals

  • Highest non-consecutive points are analyzed for both a 10-second interval and a 30-second interval.
  • These intervals help identify spikes but do not account for transients at a microsecond level.

Power Consumption Analysis

This section discusses the power consumption of different CPUs during various workloads and gaming scenarios.

Power Consumption During Workloads

  • The average power consumption of CPUs during actual workload is lower than the power consumption during loading intervals.
  • Loading applications, changing levels, and pulling assets for a save game can fully load the CPU briefly.
  • The Ryzen 5 1200F has the lowest overall power draw among the tested CPUs.
  • In Starfield, the average peak power consumption was 68 Watts for the Ryzen 7 800X, 79 Watts for the Ryzen 5 800X, 206 Watts for the Ryzen 7 14700K, and 244 Watts for the Ryzen 9 14900K.

Power Consumption During Gaming

  • In-game power consumption varies depending on the CPU model.
  • Balder's Gate III had generally high focus on CPU usage but still showed high spikes on Intel i7 and i9 categories.
  • The Ryzen 7 800X had a significant percentage higher power consumption compared to its average in-game load test.

Cost Analysis

This section analyzes the cost implications of different CPUs' energy consumption.

Total Cost per Year

  • The difference in total cost per year depends on electricity rates and gaming duration.
  • At an electricity cost of $0.10 per kilowatt-hour, there is a minimal difference in total cost between CPUs when gaming heavily (e.g., Cyberpunk).
  • For users with more expensive electricity rates or longer gaming durations, energy efficiency becomes more relevant.

Variable Electricity Rates

  • Users with variable electricity rates should consider their most likely gaming duration to calculate accurate costs.
  • Higher electricity rates can significantly impact annual energy costs.

Efficiency Comparison

This section compares the efficiency of AMD and Intel CPUs in different scenarios.

  • AMD CPUs generally offer better gaming performance and lower power consumption compared to Intel CPUs.
  • The efficiency advantage of AMD is evident even when power or FPS is locked.
  • In Blender, the Ryzen 7 14700K performs better when power is constrained compared to the Ryzen 7 800X, but this is not true for every production benchmark.
  • Overall, Intel operates at a less efficient point in its voltage-frequency curve.

Conclusion

This section concludes the analysis on CPU power consumption and efficiency.

  • Power consumption can be controlled by modifying voltage and power targets, but out-of-the-box settings show that Intel CPUs operate less efficiently.
  • AMD CPUs offer better overall efficiency in most games tested.
  • The decision on whether efficiency matters depends on individual preferences and electricity costs.

Efficiency of the 1200f Processor

The 1200f processor is remarkably efficient compared to other processors in its voltage frequency curve. It performs well between the 7800x2d and 14700K or 14900K range, where the frequency increases significantly without a significant increase in voltage.

Efficiency of the Volt Frequency Curve

  • The volt frequency curve shows that as the voltage requirement increases towards the back end, the frequency starts to go down.
  • Processors like i7 and i9 are already at the less efficient part of the curve, where there is a significant jump in voltage requirement.
  • Testing has shown that the 1200f processor is efficient in terms of power consumption.

Future Metrics and Inclusion in Reviews

The testing process has been educational and informative, providing insights into processors and their performance. There are plans to derive more metrics from this work and include them in long-term reviews. Suggestions for representative titles for long-term reviews are welcomed.

Deriving More Metrics from Testing

  • The testing process has taught a lot about processors and their performance.
  • There is potential to derive more metrics from this work.
  • Long-term reviews can include one or two representative titles with charts based on these metrics.

Baseline Power Consumption of AMD Processors

  • AMD's baseline power consumption tends to be relatively high compared to other processors.
  • This is particularly notable in processors like Threadripper, where idle power consumption can reach close to 100 Watts.

Selection of Games for Long-Term Reviews

Two games will be selected for long-term reviews based on popular demand. One game will be on the high-end, and the other can be chosen by viewers. The GN Mega charts page will be updated with power consumption data.

Selection of Games

  • Viewers are requested to suggest two games for long-term reviews.
  • One game should be on the high-end, such as Cyberpunk or Starfield.
  • The second game can be chosen by viewers through discussion in the comments section.

Updating GN Mega Charts Page

  • The GN Mega charts page will be updated with power consumption data.
  • Gaming numbers and idle power consumption will also be included based on viewer requests.

Excitement for Future Possibilities

The testing process has opened up new possibilities for reviewing CPUs. There is excitement about exploring more aspects and ideas related to CPU performance.

Underrepresented Aspect of CPU Reviews

  • Reviewing power consumption during different workloads has been an underrepresented aspect in CPU reviews.
  • This testing process allows for more comprehensive exploration of CPU performance.

Updates and Future Ideas

  • The GN Mega charts page will be updated with the latest data in the coming weeks.
  • Feedback and ideas from viewers are welcomed to further enhance the testing process.
  • Support can be provided through store.gamersnexus.net or patreon.com/gamersnexus.

Timestamps have been associated with relevant bullet points as per the given transcript.

Video description

GRAB A BRAND NEW GN15 MOUSE MAT! https://store.gamersnexus.net/products/15-yr-mouse-mat (black/blue/yellow and available in signed or unsigned!) or consider buying one of our GN15 metal emblem glasses to support our work! https://store.gamersnexus.net/products/gn-3d-emblem-glasses A lot of you have requested that we run power consumption tests for gaming on CPUs, so we've finally done that! And alongside fulfilling that request, we also wanted to tackle a comment from Greg, our recurring antagonist commenter, who "requested" CPU efficiency testing. These benchmarks look at the efficiency and raw power consumption of Intel vs. AMD CPUs. There's a particular focus on the AMD Ryzen 7 7800X3D and Intel Core i7-14700K and 14900K CPUs, as these are the most recent and directly comparable / best gaming CPUs. Watch the CPU Sample Size video! https://www.youtube.com/watch?v=PUeZQ3pky-w Learn about Intel's Snake Oil: https://www.youtube.com/watch?v=xUT4d5IVY0A Read our CPU power consumption mega charts (these ones aren't there yet!): https://gamersnexus.net/megacharts/cpu-power Like our content? Please consider becoming our Patron to support us: http://www.patreon.com/gamersnexus You can also find Wendell from Level1 Techs here: https://www.youtube.com/c/Level1Techs RELATED PRODUCTS [Affiliate Links] AMD Ryzen 7 7800X3D on Amazon: https://geni.us/bah9 AMD Ryzen 7 7800X3D on Newegg: https://howl.me/cle8h3CtNL9 Intel Core i9-14900K CPU on Amazon: https://geni.us/Etbz3 AMD Ryzen Threadripper 7980X CPU on Amazon: https://geni.us/6HfTHBO AMD Ryzen Threadripper 7980X CPU on Newegg: https://howl.me/ck3kepz4Noi AMD R9 7950X on Newegg: https://howl.me/ck2HGLczYCA TIMESTAMPS 00:00 - TwitterBall X 00:25 - BET 01:23 - Tests Included 03:22 - The Plan & Testing Methodology 06:54 - F1 2023 CPU Gaming Efficiency 08:12 - Locking the Framerate (F1) 10:25 - AMD's Floor - I/O Die Requirements 10:44 - Cyberpunk: Phantom Liberty Heavy Gaming 12:15 - GPU-Bound CPU Efficiency in Gaming 13:28 - Starfield Efficiency 14:19 - GPU Utilization vs. CPU Power Discussion 15:28 - 60FPS Locked 16:21 - Baldur's Gate 3 17:03 - Blender Power Efficiency (Wh) 17:27 - Compression Efficiency (MIPS/W) 17:47 - Photoshop Locked to 86W 18:10 - Highest Power Consumption Spikes 21:04 - Conclusions & Actual Electricity COST 25:59 - Future Plans for This Testing ** Please like, comment, and subscribe for more! ** Links to Amazon and Newegg are typically monetized on our channel (affiliate links) and may return a commission of sales to us from the retailer. This is unrelated to the product manufacturer. Any advertisements or sponsorships are disclosed within the video ("this video is brought to you by") and above the fold in the description. We do not ever produce paid content or "sponsored content" (meaning that the content is our idea and is not funded externally aside from whatever ad placement is in the beginning) and we do not ever charge manufacturers for coverage. Follow us in these locations for more gaming and hardware updates: t: http://www.twitter.com/gamersnexus f: http://www.facebook.com/gamersnexus w: http://www.gamersnexus.net/ Host, Testing, Test Lead, Video Editing: Steve Burke Video Editing, Testing: Mike Gaglione Video Editing: Tim Phetdara Guest: Wendell of Level1 Techs