SoylentNews

The Worst CPU & GPU Purchases of 2019

Intel Core i9-9900KF / i9-9900KS

Essentially what Intel has done here is create hype around a new product that’s not new at all. They’re charging users more money to cherry pick the best silicon, while reducing the overall quality of the 9900K range by limiting it to parts that can’t easily run at 5 GHz or beyond.

As a result, a few months out from the 9900KS release people started to notice how poorly new 9900K processors were overclocking and binning specialist ‘Silicon Lottery,’ had to drop the 9900K altogether. Typically as a CPU ages the manufacturing process that it’s based on will mature and this leads to a higher chance of ending up with better quality silicon, so typically you’d see parts like the 9900K that usually overclocked to 5 GHz, start to do it more frequently as times goes by.

Our concern with the 9900KS is that Intel now has the option to release a new CPU series and send reviewers the best silicon available at the time while also selling it to early adopters initially. After which point they activate an aggressive binning process, saving top tier silicon for an upcoming special edition series and sell it at a premium.

The chance of purchasing a lower quality silicon chip is always a possibility, but with this change your chance of winning the silicon lottery goes from say ~30%, to zero.

While we’re not wrapped with the idea of the 9900KS, there is also the 9900KF model. What we have here is a 9900K that overclocks no better, the integrated graphics are disabled, and it costs no less. Asking to pay full price for a defective 9900K is no joke.

Intel is so tight on 14nm supply right now that they’re selling everything, parts once destined for the bin are now binned as special versions without iGPUs. In our opinion, they’d be better off selling them to overclockers without the [integrated heat spreader], Intel could save a few bucks there, and overclockers would appreciate having to avoid the delidding step.

When you look at benchmarks for CPUs on sites like cpubenchmark.net, you're hopefully seeing an average score based on hundreds of tested samples, with the score fairly representing expected performance. But cooling and thermal differences, RAM speed, the number of RAM channels in use, and other factors could create large performance variations between the same chips. For example, the Intel Celeron N4000 in some 13-inch laptop should perform significantly better than in a thermally constrained device such as Walmart's landfill quality 10-inch EVOO tablet.

Now, not only do you have to be wary of benchmark conditions and reviewers getting sent excellently binned chips, but you should also keep in mind that early benchmarks could be artificially higher if a company's binning shenanigans includes selling lower quality chips with the same name later.

Related: That's Ryzen AF: Some Old AMD Chips Might Be Getting a 12nm Makeover