Author: Anonymous d43all

Expire: Never

The text translates to: [spoiler]"Spending Money for Real-World Testing Intel 13th and 14th generation CPUs are questioned for their stability. On the mainland Chiphell forum, a player (kmdkai) spent money to test the stability of the 13900K and 14900K CPUs. The results showed that the quality of the 13900K was already poor; under the motherboard manufacturer's Auto settings running productivity software, less than half could run continuously for a week without crashing, restarting, or reporting errors. Surprisingly, Intel insisted on pushing the limits of the 14900K even higher. In real-world testing, the Auto setting was stable only 20% of the time. Moreover, Intel's introduction of the Baseline Profile did not fully solve the problem. This article has become a hot topic recently, believe it or not. According to kmdkai, he runs a studio, so his company needs several dozen high-end computing PCs, and occasionally he assembles and sells computers. Therefore, he has encountered hundreds of 13900K and 14900Ks over the past two years, including both retail and wholesale chips. He posted the receipts of his recent CPU purchases to prove his claim. Starting from the 13th generation using the 13900K, and then switching to the 14900K when it was released, kmdkai shared his experiences. He has used ASUS motherboards primarily, including Z690, B660, Z790, B760, and checks the ASUS official website every week for BIOS updates hoping they might solve the problem. The testing standard is to run real productivity software continuously for a week without crashes, restarts, or errors. The software includes Matlab, R, FEA, VC, VS, large office batches, etc. In summary: About 40-50% of 13900Ks can be used reliably with all AUTO default settings (out of 10, 4 to 5 can be used with all default settings). Adjusting the Loadline anti-drop voltage setting can increase the probability by about 10-20%. Setting the voltage to worst / failsafe on the Z690 / Z790 boards, or using B boards can increase the probability by another 20%. Only about 20% of 14900Ks can be used reliably with all default settings (out of 10, only 2 can be used with all default settings). Adjusting the Loadline anti-drop voltage can increase the probability by less than 10%. A special feature of the 14900K is that it cannot be used reliably on the Z690 / Z790 boards, even if the failsafe voltage mode is set. Therefore, he primarily pairs it with the B660 / B760 boards. Under the same load, the power consumption and temperature of loose chips are generally about 10-20% higher than retail chips, so they are more likely to hit power or temperature limits. Loose chips have slightly better stability than retail chips. The conclusion only applies to brand new loose chips. Second-hand loose chips are likely to have been picked over, even if there are no signs of pressure. A slightly higher voltage for loose chips may be a lifesaver. The most outrageous 13900K loose chip he has seen ran at 100 degrees under a load of 180W, but it was more stable. However, the 14900K loose chip was completely useless. Recently, Intel added a baseline profile setting, but testing found it completely useless. It's equivalent to setting the failsafe voltage, default CEP, enabling the current wall, and power wall on the Z690 / Z790 for you. It doesn't help improve stability at all. After the Z690 / Z790 boards adopt this setting, they effectively become B660 / B760 boards. Before the introduction of the Inte Baseline profile, B boards were more stable than Z boards. But after Z boards adopted the baseline option, there was no difference in stability. He has only bought 3 pieces of 14900KS so far, which is too much of an intelligence tax, so he will not comment on the 14900KS for now. He shared his personal experiences as a productivity user, which is like spending money to come to a conclusion for everyone. Increasing the pressure is definitely the right solution, but how to increase it to a stable point, especially for productivity that requires long-term testing, is too time-consuming. Sometimes you might test for 2 days without any problems, but problems occur on the third day, which is very time-consuming, so every time he encounters a test failure, he immediately sells it. If you are just a general user, there is no need to worry about my extreme productivity conclusions. You can be more lenient. After all, general users can tolerate infrequent instability. For example, his personal computer is a 14900K+Z version, and under all default conditions, it only crashes and restarts once a month at most. In addition, he stated that the problems are mainly concentrated on the 13900K and 14900K. He has bought dozens of 13700Ks, all of which are as stable as old dogs, without a single problem. Therefore, he believes that if you want to use it with peace of mind, the 13700K is the most stable. He has not touched the 14700K because its cost-effectiveness is not as good as the 13900K, so he will not comment on the 14700K."[/spoiler]

Attached Files


JavaScript is not enabled in your browser. Most features and paste content is missing . Switch to full experience by editing url from /nojs/[link] to /share/[link]