Latest AGESA Combo-AM4 V2 1.0.8.0 achieved 3500x success rate with overclocking.
Latest AGESA Combo-AM4 V2 1.0.8.0 achieved 3500x success rate with overclocking.
Today I updated the BIOS on my board, and I was surprised to find a solid overclock around the standard 4.2ghz with surprisingly high voltages. Cinebench R20 at 4.4 all cores ran smoothly at 1.325v in BIOS, and Prime95 worked fine. After a one-hour stress test, it crashed. It feels like I just got a 3600X processor? Anyone have any ideas? Thanks. P.S. No need to worry, lmao! So here are the temperature readings from that run.
I piqued my interest in what OC options exist at 1.325v. I attempted it with CCD1, but it struggled to increase the clock speed without crashing. It also isn’t a fully validated stable OC, which suggests the die has solid performance potential—capable of reaching 4.4 on 1.325v.
It's important to mention the voltage is configured in BIOS while the CPU started with default settings initially. It ran at 4.4 at 1.425, then we reduced it during the FPU stress test and kept lowering it until it stabilized. Only about an hour of testing, but results look stable so far. The original BIOS didn't boot on 4.4 before 4.2 was the limit; it required at least 1.375. With a 3900x, the load line is probably set around 2 with 1.4V in stock clocks. Ryzen Master might default to around 4.3. I haven't experimented much, but watch the voltage drop in Ryzen Master. Not an expert, but I've overclocked a CPU using a pencil before.
I didn't set up an LLC at all. Perhaps I should explore that option. While adjusting voltage and frequency in Ryzen Master, I used 1.375V and temperatures hit 92°C, so I'm holding off for now.
What's the latest on cooling? I'd suggest sticking to temps under 75°F. My Corsair 240mm AIO isn't cutting it with an air cooler—real performance drops around 4.4°C. Updated Sept 6, 2020 by Leon86
Reaches a maximum of 4525GHz with a peak of 4550, likely won’t work on R20, probably not stable beyond 4.4 GHz
dark rock pro 3’s stock settings are already around 80c-82c, that’s cool. I’m wondering how high I can push it with an extra CCD—seems like it won’t really help. I probably won’t hit those temps until I bring my PC inside where it’s cooler, while the monitors and other gear stay in the warmer room.
It's great to discover such a thread. I recently purchased a 3500X earlier this week, which is a 2020 production chip. I'm able to run all cores at 4.4Ghz but with a lower voltage (1.25V) compared to yours (1.35V). I'm using a B550M motherboard with the latest BIOS. The system performs well in AIDA64 and Cinebench, as well as during demanding CPU gaming sessions. Interestingly, last year I bought a 3500X with a mid-2019 production date, which struggled to exceed 4.25Ghz at higher voltages. I suspect improvements have been made since then, as the newer 3600 models can reach 4.5Ghz. The X variant might be differently configured, possibly due to increased L3 cache size to 32MB. I'm planning to experiment this weekend to see if I can further boost the stable 4.4Ghz clock or push it up to around 4.5Ghz with higher voltage. I'm cooling it with a 240mm AIO, and my peak temperatures are in the low 60s. I'm open to any tips that could help increase the power output and reach those higher frequencies. Thanks!
Your thread title mentions the 3500X, but the screenshot displays a 3600X chip profile with 6C 12T. When reviewing the R20 multi-core score, it isn’t a 3500X. I can reach 3000 points, which is quite strong, though the absence of SMT in the 3500X limits its comparison to the 3600X clearly. Thanks for clarifying this.