Having 7950x3D feedback!
Having 7950x3D feedback!
Hey guys, had 7950x3D for few weeks now, upgraded from 7950x since it cost me almost nothing.Decision was made since 9950x was underwhelming. However it's not that simple with 7950x3D even if it's configured correctly:fresh win install, fresh reinstall of chipset drivers after revo unistaller, correct bios settings and etc. In the end somehow it's still slower than vanilla 7800x3D, even though 7950x3D has 7800x3D chip inside, it doesn't performed atleast on par even though x3D ccd was working during workload. I encountered strange performance penalty even if non x3D ccd was off in bios, performance was even lower than not disabling ccd at all which didn't make sense. So finally figured out that disabling non x3D cpu and then reinstalling chipset driver kinda fixed problem and now simulated 7800x3D works ALMOST like vanilla 7800x3D, so conclusion is that even if you set everything to AMD guide it doesn't mean cpu will perform properly, extracting all performance it has. It still required tinkering and simply disabling and enabling ccd's doesn't do a trick. I hope with 9950x3D AMD will just use both ccds with X3D else it will again require tweaking, sadly. Overall when it works properly it's amazing cool chip!
Search for contributions from @Agall regarding the 7950X3D. He has conducted thorough testing and shared updates on the topic.
3D v-cache driver appears only when needed; disable CCD1 in UEFI by setting it to 0-0-0-0-0-0-0 or use the CPPC Preferred Core option as 'cache'. Running 7950x3D offers similar results without better performance compared to 7800x3D if CCD1 is turned off. The 3D v-cache boosts CCD speed by 200MHz, which works with adequate cooling. It can sustain 5.25GHz across multiple cores when using direct-die cooling and proper tuning. I traded my desktop for an upgrade path with an RTX 4080, so I retained my 4090 but moved to an Ally X eGPU setup. I spent roughly a year testing and fine-tuning it, including replacing the cooling solution.
Thanks for the reply, but unfortunately I haven't achieved better results with my chip than vanilla 7800x3D (framechasers or techyescity form latest video). Even with CO-25 (-30 works but might crash in light loads) and 6200mhz cl28 plus subtimings dual rank ram, and now after reinstalling the chipset drivers twice, vcache usage is possible even when ccd0 is off. I’ll repeat myself: the surprising part is it just isn’t enough to just turn off ccd1 to make it faster like 7800x3D again. Until the chipset driver gets reinstalled while ccd1 remains disabled, performance drops further. This odd behavior might be specific to my Asrock B650e PG Ripley Wi-Fi. I’ll keep investigating with older BIOS and some tinkering. For now it runs within 1-2% of what vanilla 7800x3D could do before this chipset driver double install on 7950x3D started showing a 10-20% performance drop in the same games, even after tuning. I still hope 9950x3D will have both ccds with x3D to avoid these issues and software bugs.
I am analyzing the data to identify similarities and differences between the outcomes.
Uncertain about speed comparison since I only have a 7950X3D; however, setting preferred cores to Drivers and using Process Lasso to restrict the CCD usage—especially for BM Wukong benchmark that wasn't released—appears effective. Only CCD0 is utilized and clocks typically stay around 5GHz on average. For the 9950X, some tech writers claim it performs well, matching the gaming performance of the 7800X3D without any cache complications. But it's quite expensive.
Standard FAR CRY 6 test, often shared by many tech reviewers. For instance, gamers nexus ran FHD at high settings with their 7800x3D achieving around 223fps, while 7950x3D hit 198fps. Framechasers managed 245fps under the same conditions. My regular 7950x3D at default settings reached about 203fps after tweaking the chipset and BIOS, which nearly matched the official numbers. I wasn’t sure how to reach 245fps. Turning off CCD1 didn’t help—it would drop below 200fps or worse, which seemed unreasonable. But as I mentioned, reinstalling the chipset while keeping only CCD0 enabled boosted performance to 240fps, aligning more closely with the vanilla 7800x3D results. It’s strange how much effort it takes just to understand why FPS fluctuates so much.
I would run Cinebench R23 and compare the results with my thread and others. If it looks good, it might mean the problem isn’t with your CPU but with Windows or software, possibly even a faulty GPU. Running a gaming test is probably checking too many things at once.
Using synthetic tests that don’t understand vcache feels irrelevant for me. The main goal is gaming, and even with tricks like process lasso or custom windows builds, issues remain—like CPU affinity problems. Even when everything seems set up correctly and ccd0 works while ccd1 doesn’t, performance drops below 7800x3D. My feedback helped clarify why core parking isn’t perfect with official methods, and avoiding it is still challenging. I personally had better results with 7950x for two years, reaching around 40,500 points in Cinebench. At 5.7–5.8GHz, it struggled to match 7800x3D even at lower settings. The simple fix was turning off ccd1 and reinstalling the chipset twice—this restored proper performance. Thanks for your insights!