R5 1400 offers reduced energy usage.
R5 1400 offers reduced energy usage.
You're experiencing higher idle power consumption despite optimizing settings. The voltage drops during idle, but the CPU still draws more than desired. Since you're already using lower voltages in BIOS and haven't changed hardware, consider checking for additional power-saving features like Intel's Power Management or Ryzen's Turbo Boost controls. Also, ensure your motherboard supports aggressive idle power throttling and that no background processes are unnecessarily active. If possible, explore firmware updates or advanced BIOS tweaks to fine-tune efficiency.
You're unlikely to significantly reduce idle power. The CPU core remains inactive during idle. To achieve under 20 watts idle, consider a lower-power setup such as a mini PC or laptop. Also, other components on the board are likely consuming substantial power.
A nucleus would be perfect, but what makes you think this is necessary?
Summit Ridge's core operates with 20 watts. Reduce memory frequency significantly to lower fclk.
The processor is outdated and uses 14nm technology, which limits efficiency. Switching to a 3100 or 3300X offers better performance. Undervolting can enable 7nm operation, allowing lower voltages without significantly reducing clock speed and improving power efficiency compared to 14nm.
It's about twenty dollars a month at that rate, which adds up to roughly one hundred twenty dollars a year. I’d prefer a yearly cost of around two fifty dollars if things run smoothly.
Generally, getting several PCIe ports from a laptop or notebook is tough or even impossible. I'm swapping out the i5 4590T, which operates at turbo-off 2GHz without SMT, drawing around 20W and idling under 5W. Notebooks usually have limited PCIe options, which makes sense—I think my approach is reasonable. I’m running it at XMP 3200, so I’ll compare DDR3 1600 with the timings on my current setup and adjust the DDR4 to match performance. This should help reduce uncore usage and overall power consumption. From a quick check, DDR4 2133 C10-C11 seems slightly faster than DDR3 1600, so I’ll see if tuning it down brings improvements. Once built, I’ll test switching from stock speeds to 2GHz low memory speeds to assess any impact on power draw and performance.
CPU Power Information: These parts are hard to locate and often cost $150 or more with long shipping times from China. If it’s in the $150 range, it’s expensive; domestically, especially used, could be closer to $200.
The i5 4590T uses a 14nm process but is efficient, so broad details about manufacturing aren’t crucial. This seems more about architecture than power efficiency. You’re likely right—2GHz will work fine for my needs. At $0.08 per watt, it’s not a huge concern. I aim to keep things running as efficiently as possible for long-term power usage.
I’m rebuilding my mining rig from 2 GPUs to 4 GPUs, and the only available board with enough PCIe slots is the B450. My only compatible AM4 CPU is the R5 1400. An APU Athlon 200GE has limited PCIe lanes, so I’m uncertain it will suit this project.
It should work well at 2ghz and low voltage. I managed this with my 220ge, but at 4ghz it dropped to half the cores. I own a Ryzen 5 1400 in a B450M-A. Based on the specs, reducing it to dual-core with four threads might help. Keeping the CPU at stock speed could halve the power draw.
I haven't experimented with that before. When I tried it on my 2700X with PBO still active (it's now off), the idle consumption for 4C and 6C matched what it did for all 8. The reduction in idle power happened when I turned off PBO and XFR, which cut idle draw from roughly 40W to about 20W. For my 1400 model, the idle usage is nearly comparable to passive cooling—just a budget down draft heat pipe Snowman cooler I purchased earlier. It’s labeled “120mm,” but it’s actually just the fan; it’s a 92mm cooler with a 120mm attachment. I chose it mainly because it’s affordable, the heat pipes cover the entire Ryzen IHS, and the spacing keeps dust from clogging quickly compared to denser models.
Running memory at the standard 2400 MHz reduced idle consumption to 10–13W, which is improved. Peak power usage is now around 25W, similar to what I saw with the i5 4590T. The downcore control offers settings like 2+0 or 1+1, likely to disable a full CCX or one core per CCX. I prefer keeping all four cores running since I’ll have four cards in the system—three are Nvidia, and I’ve faced odd problems when stacking more cards than cores. While mining isn’t the same, hitting my power target is better.
Under load at 2GHz, it drops to about 0.7 volts, and idling falls to around 0.4 volts. It’s stable, passed a 24-hour Prime95 test with memory at 3200, so I’ll run it another day at 2400 with the same settings before rebuilding.