What's scary is that I think the owner of userbenchmark actually believes that statement. Which might explain how he's so out of touch that he thinks his own crap doesn't stink and deserves to be locked behind a subscription. I'm just sad that there might be a not insignificant number of people that pay for it.

The real Neanderthal social media account is the one writing that review.

Instruction and data caches have a real, tangible benefit. Although there is a point of diminishing returns, more L3 cache is absolutely worth a 10% clock speed trade-off for consumer systems. Fetching memory from the bus is an order of magnitude slower than fetching from cache, and the processor has to perform other work or stall while it's waiting for that.

But, knowing the bias of the reviewer, they're probably running DDR4 at 5200 MT/s (2000 over JEDEC specs) on their Intel systems to make up for the lack of cache while thinking, "just buy a more expensive processor and RAM, you brain-dead cretins."

Ok so I am about to build a new rig, and looking at the specs the X3D does seem less powerful and more expensive than the regular 7950.

While I completely agree that this guy seems extremely biased and that he comes off like an absolute dickbag, I don’t think the essence of his take is too far off base if you strip off the layers of spite.

Really, it seems like the tangible benefit of the X3D that most people will realize is that it offers similar performance with lower energy consumption, and thus lower cooling requirements. Benchmarks from various sources seem to bear this out as well.

It seems like a chip that in general performs on par with the 7950x but with better efficiency, and if you have a specific workload that can benefit from the extra cache it might show a significant improvement. Higher end processors these days already have a fuckton of cache so it isn’t surprising to me that this doesn’t benchmark much better than the cheaper 7950x.

Not if you remember that the writers are being paid by Intel. Then, it all comes together.

You're expecting rationale from someone who just made crazy statements because their feeling are hurt.

Uhh.... Aren't... Aren't these two statements kinda contradictory?

No no, you see; it performs reasonably consistency under varying real world conditions but for a CPU to truly shine it needs to handle all workloads, including unrealistic synthetic ones.