JamesSneed:

This is the perfect price if you don't want to sell too many chips. I doubt many will buy it considering Zen4 isn't that far off.

Horus-Anhur:

When HU did their tests with CPU cache, they found little improvement when going from 12MB to 20MB. So there is definitely something with those numbers from AMD.

user1:

price is a little too high I think, but we'll see , could just be that it isn't going to be very good supply . not all architectures are equal, first off, there are different caching strategies, in amd's case the l3 is a "victim cache" which scales with size in a predictable way, intel doesn't necessarily do this type of caching, they employ a more intelligent caching scheme for their l3, though they have used the l4 on broadwell-c as a victim cache in the past. secondly, cache scaling depends greatly on the workload, perfect example is the apus vs the chiplet designs, despite only having 16mb of cache, for most workloads, you're looking at -5%, however for certain workloads like in games, it can be as much as -20%. I will also add, that zen3 was designed with the Vcache in mind. so there is potentially headroom there that is not fully exploited, A reasonable assumption could be surmised that they simply didn't need to add the cache since they already compete well enough without it, and it isn't cheap.

Horus-Anhur:

You are right. But we are talking about a a game going from almost no improvement with cache increase, to an increase of 36%. For a CPU, 36% increase is huge.

user1:

well if we use a crude rule , like say doubling the cache size increases the performance by 20% best case, and you add a clock speed /powertarget bump, Its not inconceivable. There is also the part where minimum size matters, so if you have a process where increasing the cache from 32 to 64 means that it runs in cache almost 100% of the time, you can get disproportionately large increases in performance.

Horus-Anhur:

I can agree that several games will have this increase in performance. But Watch Dogs never showed great scaling with cache. We would have to be very lucky for Watch Dogs Legion to have some main thread data that perfectly fits inside those 96MB of cache, resulting in the CPU just breezing trough data. Clock speeds on the 5800X3D will be lower. Which makes this supposed result of 36% even stranger. It has a base clock of 3.4GHz and a boost clock of 4.5GHz with a 105-watt TDP. That’s a 400MHz reduction in base clock and a 200MHz reduction in boost clock compared to a stock Ryzen 7 5800X.

user1:

I thought they only doubled the cache , but its actually triple, from 32 to 96mb, if we assume that same rule, then it puts the best case uplift at +30%, curious that the clocks are lower, we'll just have to see I guess, maybe there are some changes to the silicon aswell ( b2 revision).

MonstroMart:

Depends on a lot of thing. Let's just you have an old AMD platform able to accept the 5800X3D. Let's say price of DDR 5 remains high. I could definitely see someone buying that as a last upgrade for their old AM4 platform while waiting for the price of DDR 5 to come down a bit.

Undying:

Amd so late to the party. Who needs a 5600 now when you can get a faster 12400. Same goes for 5700X.

Horus-Anhur:

When HU did their tests with CPU cache, they found little improvement when going from 12MB to 20MB. So there is definitely something with those numbers from AMD.

KissSh0t:

Will there be lower model X3D chips? like a 5600X3D for example? or is the plan for only the 5800X3D?

bobnewels:

Has anyone posted in this thread even going to buy the 5800X3D or this just a complaints thread,if so my tea got warm while playing elden ring.

Undying:

Amd so late to the party. Who needs a 5600 now when you can get a faster 12400. Same goes for 5700X.