What method is most effective for verifying 3900x stability?
What method is most effective for verifying 3900x stability?
I performed an OCCT heavy test for two hours without encountering any crashes. Now I'm looking into ways to control my voltage during idle.
If you opt to game then the 3800x offers better performance. You achieve higher clock speeds without the extra delay from two CCDs. With the 3800x in games, you can experience up to 4.45GHz boost using water, while the 1.0.0.3 ABBA BIOS provides that. The challenge with the 3900x lies in the cooling needs. Games typically don’t demand 12 cores, usually 6-8 are sufficient. To match the 3800x’s clock speeds, you’ll need a chiller.
With my 3800x overclocked to its limit, I achieve a 11800 Time spy cpu using IF 1900 and RAM 3800.
https://www.3dmark.com/spy/9573602
In essence, I perform similarly to a 9900k in games with the same GPU. On average, a 9900k scores around 11730 points.
The 3900x is not ideal for gaming due to lower clock speeds, and for games an 8-core 9700k setup is generally adequate with overclocks of 5-5.1GHz or more.
The 3800x matches the performance of the 3900x in games.
In gameplay, the 3900x offers only marginal gains compared to the 3700x/3800x.
You can run phase change continuously, making a chiller a breeze.
For gaming he should rely on pbo and auto OC. For efficiency he needs to manually maximize all core clock speeds. The overclocking effort should focus mainly on RAM. If successful, an IF of 1900 can be stable. Otherwise, set the RAM to 3800 and fine-tune the timings as much as you can.
RAM overclocking significantly boosts performance far beyond just the core speeds. For instance, my current settings are cl15 (as close as possible with other configurations) at 1.44 volts on a Samsung B-Die.
When working with video encoding, use NVENC.
I've checked my bios settings and noticed that power saving features for idle cores are still enabled, yet hardware info keeps showing static voltage and frequency—clearly not functioning properly. I found the pstate options but prefer not to adjust them right now. Using the Ryzen balanced power plan isn't helping either, as it doesn't lower core clock at idle.
Is Prime95 still considered dependable for stress testing? I conducted a blend test across all cores and waited an hour, observing temperatures at just 58°C with the CPU using only 38W power in Ryzen Master. That definitely seems unusual. But since all CPU cores were pushed to their maximum, I’m questioning whether I’m spending unnecessary time on Prime95. I was considering running it all night because I had pushed my all-cores to 4175 MHz today.
Blend isn't that high in temperature, you should aim for 60c. I do it with the 3800x. Try using a HEVC load and see what happens. Significant changes in temperatures. Also, aida64 will stress more on the CPU but prime95 at 8k fft will really increase the heat.