Ryzen 7 3800x with Gigabyte B450 AORUS ELITE V2 running continuously on Linux
Ryzen 7 3800x with Gigabyte B450 AORUS ELITE V2 running continuously on Linux
Enable PBO in BIOS and adjust the voltage slightly lower. Try setting PPT to 333W, with EDC and TDC at 230A. Apply a minor undervolt using negative offset only. Begin around -0.0125V or one notch. This helps the CPU run cooler, maintaining higher clock speeds for longer. Avoid extreme settings as they may cause performance drops or instability. Undervolt values differ based on motherboard and CPU, so experiment to find the best setting.
that 3800x should be performing quite well as it stands
if you aim to overclock make sure your PC has ample airflow (not just for the CPU)
make gradual adjustments, most of these chips are quite powerful from the factory and don’t offer much room for improvement.
you could also consider selling the 3800x and purchasing a 3900x used if you really need to speed up file compression...
just watch some YouTube videos and see where to begin
Activate PBO in BIOS and apply a slight undervolt. Some basic PBO configurations that could assist: configure PPT at 333W, with both EDC and TDC set to 230A. Apply a very minor negative offset undervolt—starting around -0.0125V or one notch. This allows the CPU to operate cooler, helping it sustain higher clock speeds over time. Proceed cautiously; excessive undervolting may reduce performance or cause instability. Undervolt values differ based on motherboard and CPU, so some trial and error is necessary to reach the ideal UV.
Enable the following features: AMD Cool n Quiet, Advanced C States, Processor CPPC, and CPPC Preferred Cores.
PBO increases thermal output, but its boost algorithm safeguards the processor by reducing clock speeds when temperatures rise. Therefore, maintaining high boost clocks while PBO is active relies heavily on effective cooling. Given continuous operation, consider opening the case to use a room fan near the CPU cooler intake.
Aim for temperatures below 80°C for optimal performance; mid-to-high 80°C is acceptable for sustained use. Avoid exceeding 90°C long-term, as it approaches the Tjmax of 95°C. Also, adjust memory clock settings accordingly. After completing these adjustments, particularly after memory tuning, perform a stability test to ensure no corrupted files are generated.
I wouldn't rely on a fixed, manual, overclock for any Ryzen 2nd or 3rd gen CPU. While some users find it effective, it requires extensive adjustments to discover a safe voltage and frequency that won't cause CPU issues. This is particularly crucial for a system operating continuously with heavy all-core tasks. PBO keeps the boost algorithm active, which safeguards the processor. Moreover, a well-configured system using PBO should match the performance of a manually overclocked setup in all-core scenarios and even outperform it in lighter threaded tasks.
If you're prepared to increase CPU stress with PBO, raise the PBO Scalar setting to 5X. This loosens the algorithm's protection thresholds. I wouldn't exceed 5X unless temperatures remain in the low to mid-80s.
For RAM overclocking beyond what XMP settings suggest, use a tool like Ryzen DRAM Calculator. It assists in determining ideal timings for higher RAM speeds while maintaining stability, though it can be somewhat complex to apply correctly. I wouldn't stress about clocks surpassing 3600 unless you need to desync the IF for greater speeds.
Hello, I made a small adjustment and observed some changes. I applied PBO, set PPT to 333W, and adjusted both EDC and TDC to 230A. Voltage remained unchanged, and I configured Scalar 5X. After 12 hours of operation across all cores, the results are:
CPU Usage - 35-60%
Temp - 65 °
Code:
cpu MHz : 2537.951
cpu MHz : 3900.000
cpu MHz : 4437.716
cpu MHz : 4435.262
cpu MHz : 3900.000
cpu MHz : 2800.000
cpu MHz : 3900.000
cpu MHz : 2800.000
cpu MHz : 4437.235
cpu MHz : 3900.000
cpu MHz : 2200.000
cpu MHz : 2200.000
cpu MHz : 2800.000
cpu MHz : 2800.000
I'm not sure about the tool you're referring to, but it seems to be a typical average clock speed per thread. This approach doesn't accurately reflect the performance of a modern multi-threaded processor. The rapid changes in core clocks can make it misleading, especially since each virtual thread shares only one physical core.
It could also indicate better compression code efficiency rather than overall processor speed, as it implies heavy use of just a few cores (like three at over 4.4Ghz) while the rest run lightly.
To get a clearer picture, you should run a standard benchmark and measure how long it takes to finish. A typical test would take around ten minutes. I usually check performance using a Handbrake video.
The fastest compression results often come from increasing to higher settings like 3900 or 3950 cores, which makes better use of available resources. However, be mindful of the motherboard VRM temperature, as it might not improve much beyond that point. The report also warns that scaling performance further may not be as effective as expected.
I've constructed my machine using this setup:
https://docs.google.com/document/d/18a5_...qqQU/edit#
The person mentioned uses: Intel Core i7-10700, producing around 2.76 TiB/day with the same parts but a different CPU and motherboard, achieving 2.20 TiB/day. Is it realistic that the 10700 offers about 20% better performance than the 3800x?
I believe the answer would be positive if the code is tuned for Intel, particularly if it leans heavily toward Intel at the cost of AMD compatibility. Examining the thread clock data suggests it might be designed for light threading and could struggle with high-core counts like 8 cores or 16 threads. It seems tailored for a 4 core/4 thread processor, highlighting Intel's performance edge in IPC.