News Ryzen 5 9600X benchmarks show doubled cache bandwidth improvements — leaked AIDA64 benchmarks point to much faster L1 and L2 cache

[abufrejoval]

Yes, it does matter. Also, not everyone wants the "fastest" X3D gaming chip. Most gamers would be interested more in the vanilla parts. And there is this price gap difference as well.

I'd slightly modify that statement and say that with the newer cores not every gamer needs the fastest X3D chip any more, practically relevant returns might be diminishing and increasingly they might just be happy with what the get, or risk more running into GPU walls instead.

While the benefits of top clocks for anything not gaming might still offer valuable returns at this point.

In my experience, upgrading Ivy Bridge to Kaby Lake i7s to 5800X3D for the gaming rigs of my kids did provide a nice performance uplift, especially at current prices for Zen 3.

I've now used falling prices to upgrade a 5950X to a 7950X3D and see if the two types of CCDs were really all that different for gaming, but failed to detect really awsome real-world differences between the two with my RTX4090.

The V-cache-less Zen4 cores on the 2nd CCD simply don't suck all that bad and the VC enabled Zen4 cores on the 1st CCD fail to completely alter the gaming experience all over the field.

In fact, the differnce may be somewhat measurable, but it's not that noticeable in my personal gaming experience on an RTX 4090 I bought for CUDA.

But then I run things at only up to 144 Hz variable frame rates, because my 43" 4k screen can't do better. While the X3D might enable faster frame rates, especially at lower resolutions, those would either be far beyond 144Hz where I don't care or they hover around the 70Hz mark with both, because even an RTX 4090 with DLSS3 and frame generation won't run faster e.g. on ARK Ascend. I've compared with both, BIOS disabling of CCDs and Project Lasso, to make sure I tested properly.

I mostly observe far less difference than all the enthusiastic reviews seemed to indicate. But upon closer look, most of these are benchmark obsessions at resolutions and frame rates that are pure drag racing, not "driving on the Autobahn from A to B".

I went with the 7950X3D mostly for curiosity and because I didn't want to do worse at the occasional game than my kids with their 5800X3D. And I'm just glad it's still a bit faster than the 5950X it replaced at work loads, beause that's paying for the purchase.

But I'm not holding my breath on upgrading CPU or GPU at this point, because finally I got enough power for the 4k screens and even for the HP Reverb headset. More doesn't turn into better that easily from here, or would require something really EPYC.

 

[TerryLaze]

Yes, it does matter. Also, not everyone wants the "fastest" X3D gaming chip. Most gamers would be interested more in the vanilla parts. And there is this price gap difference as well.

Yeah, that's my point, no matter which one people are interested in, most will wait for both types to come out so that they can compare them and decide what to get.
Nobody has any FOMO right now and would rush out to buy the first available version because everybody already has a pretty decent CPU.

 

[Deleted member 2731765]

I've now used falling prices to upgrade a 5950X to a 7950X3D and see if the two types of CCDs were really all that different for gaming, but failed to detect really awsome real-world differences between the two with my RTX4090.

The V-cache-less Zen4 cores on the 2nd CCD simply don't suck all that bad and the VC enabled Zen4 cores on the 1st CCD fail to completely alter the gaming experience all over the field.

Speaking of CCDs, I've noticed that some games prefer cache (CCD0), while others prefer frequency (CCD1). However, having both CCDs enabled is also important in some cases. But disabling one of them has really helped in some of the games.

https://fudzilla.com/news/gaming/55688-amd-ryzen-9-7950x-improved-by-disabling-ccd