I don't have a personal opinion, but I can provide an objective assessment of your Ryzen 2600 overclock.
I don't have a personal opinion, but I can provide an objective assessment of your Ryzen 2600 overclock.
Yesterday I increased the clock speed of my Ryzen 2600 processor. Not being an expert, I would appreciate your advice to confirm whether all the settings are correct or not:
HWinfo after 45 minutes of prime95 (small ftt):
https://imgur.com/rtZB7uV
https://imgur.com/coG4OrQ
Bios settings:
https://imgur.com/MisY4vo
I began configuring the ASUS TPU II, but at 3.95 GHZ I measured 1.331 V on core (76°). Therefore, I started testing with a multiplier of 39 (3.9 GHZ) and adjusted the CPU voltage. The CPU is cooled by a Mugen 5 tower air cooler.
The video of the processor shows a value of 1.21, so I began with an offset (+) of 0.1 (74°), then 0.08 (72°), 0.06 (69°), 0.05 (66°), 0.04 and 0.03. At 0.03 Prime95 crashed after 2 minutes, so I reset the offset back to 0.04 (0.04375 to be precise) and ran Prime95 (small ftt) for 50 minutes and OCCT (5 minutes). It appears stable with a maximum temperature of 64.8°.
For completeness, LLC is set to AUTO, SOC voltage to AUTO, and RAM voltage at 1.35v (my RAM are 3200c16 but set at 3000c16; higher I get the BSOD).
Regarding OC settings, did I configure them correctly? Are there any issues with the screenshots? 1.175v for 3.9 GHZ seems quite low, especially since the ASUS TPU II gave me 1331v at 3.95 GHZ.
Lastly, I attempted benchmarks using Geekbench and CPU-Z. I know that a 3.9 GHZ is not much, as the processor normally reaches a boost frequency of 3.9 GHz at stock speeds. However, the results show little to no improvement, which is a bit disappointing:
Geekbench single: ~ 4400
Geekbench multi: ~ 22000
These numbers seem consistent with the stock 2600 model, so it doesn’t appear to be a major issue.
Cpuz single: 443.8 (stock was 448)
Multi CPU: 3468 (stock was 3438)
Even lower in single processor, practically similar in multi.
Could this indicate something is wrong? Is the CPU throttling for some reason, or is 3.9 too low to achieve noticeable gains?
Thank you very much!!!
Synthetic benchmarks usually show improvement percentage same as percentage of OC. For example, if you OC by 10%, benchmarks will also show about 10% higher scores. Only problem is that it doesn't translate same way in let's say games. In a particular game and only if CPU is a bottleneck, you may see maybe only a couple of FPS better result. If there's no CPU bottleneck you may not even see any improvement.
OC mostly affects IPC (Instructions Per Cycle), in practical terms single core performance in such programs/games that are single threaded applications (working mostly on one core).
It seems to be operating smoothly, with x being a higher binned value beyond x. However,... you can still increase it further since voltages and temperatures remain safely within limits (1.425v and 75c).
CountMike's message indicates everything seems to be functioning properly, with x being a higher binned value. He mentions that increasing voltages and temperatures stays within safe limits (1.425v and 75°C). He also asks for an estimate of performance improvement if the overclock frequency is raised to 200 MHz.
Synthetic benchmarks usually show improvement percentage same as percentage of OC. For example, if you OC by 10%, benchmarks will also show about 10% higher scores. Only problem is that it doesn't translate same way in let's say games. In a particular game and only if CPU is a bottleneck, you may see maybe only a couple of FPS better result. If there's no CPU bottleneck you may not even see any improvement.
OC mostly affects IPC (Instructions Per Cycle), in practical terms single core performance in such programs/games that are single threaded applications (working mostly on one core).
CountMike :
Synthetic benchmarks usually show improvement percentage same as percentage of OC. For example, if you OC by 10%, benchmarks will also show about 10% higher scores. Only problem is that it doesn't translate same way in let's say games. In a particular game and only if CPU is a bottleneck, you may see maybe only a couple of FPS better result. If there's no CPU bottleneck you may not even see any improvement.
OC mostly affects IPC (Instructions Per Cycle), in practical terms single core performance in such programs/games that are single threaded applications (working mostly on one core).
Yes, I use it mainly to play, so I'm not very interested in squeezing every possible MHZ from the CPU if I do not have some real advantage.
For now it seems stable and temeperature / voltage are low so i'd like to settle for this minimum overclock, but does it make sense? I use more energy because it is always at 3.9 ghz since I set the multiplier or it go in idle? from what you say stock boost to 3.9 ghz or overclock on 6 cores to 3.9 ghz for gaming is the same.
Thank you!!!
CountMike summarized the situation by noting that with FXR, not every core would reach maximum speed, yet all cores stayed at a fixed frequency. This approach might provide some extra boost for non-game tasks, though it introduces many uncertainties. He suggested adjusting RAM timing instead, as performance is currently limited to 3000Mhz C16 (the spec), and he can't increase the frequency due to BSOD issues during startup, but lowering timings might be feasible since they could drop to around 200 MHz below the spec.
Certainly, look into JEDEC and XMP specifications for that RAM. CPU-Z can provide a general overview, and Taiphoon Burner will offer more detailed insights.