News Ryzen 7 7800X3D Smashes Core i9-14900K in Factorio Gaming Benchmark

[bit_user]

I think @waltc3 say so is that you insist on argue such as cache size is not very important.

I don't think I ever said it wasn't very important. Whether it is, really depends on your workload - that's the point I've been repeating over and over.

This is an obvious point and all of us knows this.

Not from the sound of it.

But if I say to you to choose from a CPU with 1MB of cache and another with 4MB, at the same clock, architecture, etc.. what CPU do you think go faster ?
Generalising, more cache is better, do you agree ?

Well, almost 20 years ago I bought a Pentium 4 Prescott, because I thought the double-size L1 & L2 caches would make up for its other deficiencies (i.e. 31-stage pipeline, compared to Northwood's 20-stage). That wasn't my only reason, but it turned out not to be a great decision.

We had another natural experiment with the X3D chips - 3x the L3 cache at slightly reduced clock speeds. Some games and apps preferred more cache, while others preferred more clock speed. If "cache is king", that shouldn't have happened, right? If "cache is king", then the larger cache version should run everything faster! ...but it didn't!

Cache is not king. It's more of an unreliable prince, if we're being reductionist. If we're being nuanced, then you also have to look at latency, bandwidth, how it fits into the cache hierarchy, associativity, etc. On that last point, AMD has been playing some games by comparing its L3 against Intel's, yet I believe AMD's L3 is more like a L2.5. It seems to be the case that Zen CCDs will only populate their own slice of L3, whereas Intel CPUs have a truly global L3 that's equally shared by all the cores.

 

[NinoPino]

...
Well, almost 20 years ago I bought a Pentium 4 Prescott, because I thought the double-size L1 & L2 caches would make up for its other deficiencies (i.e. 31-stage pipeline, compared to Northwood's 20-stage). That wasn't my only reason, but it turned out not to be a great decision.

We had another natural experiment with the X3D chips - 3x the L3 cache at slightly reduced clock speeds. Some games and apps preferred more cache, while others preferred more clock speed. If "cache is king", that shouldn't have happened, right? If "cache is king", then the larger cache version should run everything faster! ...but it didn't!

Cache is not king. It's more of an unreliable prince, if we're being reductionist. If we're being nuanced, then you also have to look at latency, bandwidth, how it fits into the cache hierarchy, associativity, etc. On that last point, AMD has been playing some games by comparing its L3 against Intel's, yet I believe AMD's L3 is more like a L2.5. It seems to be the case that Zen CCDs will only populate their own slice of L3, whereas Intel CPUs have a truly global L3 that's equally shared by all the cores.

This is the reason because I wrote "generically" and "with same architecture, clock, etc...".
I understand your point but "more cache is better" is a good slogan. Also "cache is king" is not bad.😃

 

[bit_user]

This is the reason because I wrote "generically" and "with same architecture, clock, etc...".

Ah, but there's usually a tradeoff. Even "with same architecture, clock, etc...", larger caches have higher latencies.