The tighter lithography might contribute to instability problems in Intel chips.
The tighter lithography might contribute to instability problems in Intel chips.
HardwareTimes received their initial RMA approval. The replacement CPU also malfunctioned, and Intel refused to cover the cost under warranty. The situation has worsened the brand image significantly. Much worse than before. Many still remember how Intel damaged the C2000.FTDI reputation a decade ago, and it continues to affect them. The intel sales team is likely offering steep discounts on new agreements to offset the challenges of the 13th and 14th generation issues. Wendel estimates a roughly 20% failure rate from OEMs, which could cost millions to manage these returns. For Intel, this is extremely difficult. They face several tough choices: recalling older models with no high-end alternatives, pushing customers toward AMD (which is problematic), maintaining current stock (harming reputation and risking lawsuits), searching for a fix, creating new silicon, producing extra masks, updating manufacturing, and testing the IC. Depending on available staff, this could impact other projects. Fixing with 15th generation and supplying replacements would require new sockets, swapping mainboards, updating production lines, and retesting – all demanding time and engineering effort. Not every failure happens instantly, so improving testing to spot problematic CPUs early could help. Shipping only those that pass the latest tests offers a quick, low-cost solution and restores confidence.
It seems fixing the issue with CPU frames and slightly reducing mounting pressure mostly resolves the problems. A design flaw is likely obvious, though it's unclear if it can be addressed during production. Some suggest increasing the voltage or lowering the maximum boost, which might point to another design issue. Intel appears to be under pressure, as time is running short and they need to prevent a repeat in the next generation. People might eventually switch to AMD if reliability becomes a bigger concern.
It seems the node size might be affecting performance, but it doesn't appear to be the main factor. The i3s and i5s sharing the same node aren't showing issues, suggesting another cause. (Or perhaps size combined with another element.)
I already viewed the second clip. I only watched that part of the first one. Unless I overlooked something, there was no reference to Xeon. In the original video they mentioned running consumer CPUs on W chipset boards because higher clocks were better suited for the tasks. The key point is it challenges the idea of overly complex M.2 settings, though some reports might still be explained by other factors. I still think multiple elements are involved, and now more attention is being paid. Edit: Wendel is currently featured on PCWorld (The Full Nerd Podcast). The Intel situation is coming up next.