AMD's new cheap gaming chip is 25% faster than Intel's more expensive alternative — Ryzen 7 5700X3D is faster than Core i5-13600KF in early reviews, even on aging AM4 platform

Pierce2623 said:

While I’m definitely an advocate of AMD’s 3dvcache, testing against a stock 13600k at 5.1GHZ is kind of unfair when it’s practically guaranteed to do 5.6 all core and maybe 5.7. The vcache parts can’t be multiplier overclocked so you’re limited to whatever max clock AMD sets it at. In fact, with having to also do work on my PC, I’d only get one of 2 CCD vcache chips because I also need the higher clocks of the non-vcache ccd.

Eh, I mean they're within the same price range, the 5700X3D will retail for $250 vs the current pricing of $286 for the 13600KF. It would be worth looking at whether that extra $36 dollars is worth it, especially with AM4 platforms having similar pricing to LGA 1700.

Pierce2623 said:

While I’m definitely an advocate of AMD’s 3dvcache, testing against a stock 13600k at 5.1GHZ is kind of unfair when it’s practically guaranteed to do 5.6 all core and maybe 5.7. The vcache parts can’t be multiplier overclocked so you’re limited to whatever max clock AMD sets it at. In fact, with having to also do work on my PC, I’d only get one of 2 CCD vcache chips because I also need the higher clocks of the non-vcache ccd.

On the contrary, testing one CPU that is overclocked and one at stock clocks is 100% unfair. It's nearly always been the case to compare things like for like in (reputable) product reviews. However, providing the overclocked results adds value to the review, and the potential buyers gain some insight on their planned purchase. The fact remains that most K SKU's remain at stock clocks, overclocking is far less common than people think.

For some context (Using an unfortunate automotive analogy that doesn't quite fit) it used to be VERY common for automakers to provide review vehicles with (undisclosed) expensive performance tires (and sometimes other modifications) that were not provided on the (stock) retail models. Tires have a MASSIVE effect on performance and fuel economy and the choice of a better performant tire can skew results favorably for an otherwise noncompetitive product . Certain review magazines (yes, this predates the internet) began calling out the practice and it mostly ended, though there is definitely some tomfoolery still. Cars are not CPU's but this does highlight WHY products are tested AS SHIPPED TO RETAIL. Whether it's clock speeds, RAM timings, single or dual channel, cooling devices and the list goes on.

Now, that all said it is perfectly OK, and encouraged for a reviewer to afterwards modify settings or add better cooling or memory, even overclock as a value add to the review. It's practically standard practice these days. For those of us who are enthusiasts it's often this information that is most valuable.

On the current "review", the wins/losses definitely highlight how cache can affect performance. A simple OC wasn't going to net a win in every situation for the i5, but for the closest ones it could be the tie breaker. Regardless, they are two mostly non competing markets. The AM4 X3D variants are aimed towards enthusiasts looking to upgrade existing rigs to modernish performance standards, not new system builders. Those people will get the i5, or go to a modern AM5 platform which DOES directly compete with the Intel offering.

Like how do you people come up with these titles?
Why up to 25% ?
The biggest difference is 33% while the average is 9% why the 25% ?

TerryLaze said:

Like how do you people come up with these titles?
Why up to 25% ?
The biggest difference is 33% while the average is 9% why the 25% ?

You clicked didn't you?

Sadly most of the time when there are cheap chips like this they are paired with the worst of the worst of other PC equipment. Even if it is possible for the chip to handle it, the other gear won't hold up and will slow it down.

BUY GAMING PC!
It has XYZ chip in it for performance!
Chip was tested under ideal conditions. This PC is not ideal.

Pierce2623 said:

While I’m definitely an advocate of AMD’s 3dvcache, testing against a stock 13600k at 5.1GHZ is kind of unfair when it’s practically guaranteed to do 5.6 all core and maybe 5.7. The vcache parts can’t be multiplier overclocked so you’re limited to whatever max clock AMD sets it at. In fact, with having to also do work on my PC, I’d only get one of 2 CCD vcache chips because I also need the higher clocks of the non-vcache ccd.

True they can't be multiplier unlocked past a certain point but they can be forced to stay at boost without thermal issues and then they can still be overclocked higher using the eclk the only problem is that crystal also is in charge of pcie and SATA clocks... But that doesn't mean it can't be used by skilled overclockers and plenty of patience.

I've seen someone get it up to 5.9 GHz but so far my best efforts top out at 5.5 but I don't think I have the best bin. That said the gains were there and temps were still very manageable and even more so if I disabled CCD1 but I usually don't bother because I haven't really had any scheduler issues and even games that get put on the non vcache ccd run great on my 4090 because it's running at 5.9ghz and my memory is a 6400mt lot overclocked to 7000mt CL28-28-30-80 +0.036 mv (I think it's 36 anyway lol) which actually shocked the hell out of me because my first kit and it's store exchanged replacements topped out at 6400mhz which was it's expo profile.