Are The New Zen2 Refresh CPUs Worth It?

[r.linder]

The short answer: No.

The long answer: No, you're better off with the pre-existing Zen2 or Comet Lake CPUs or waiting for Zen3, because:

1) Performance
While the new refresh CPUs, assuming silicon lottery is on your side, can get the advertised boost increase on a single core, most multi-threaded loads will have pretty much the same clocks as it would with the CPU it "replaces."
For example, while the 3900XT can reach 4.7 on average on a single core, the average clock speed across all cores on all core load is less than 100 MHz higher than the 3900X, and it's usually closer to 4~4.1 GHz. Given the price difference between Zen2 and Zen2 refresh, it's not really worth it because you'll pretty much never be running single core loads anyway. Take it from a 3900X user; I never see 4.6 GHz outside of single core loads. Paying more for the XT isn't a good idea, because all you're getting is an extra 100 MHz on a single core which doesn't really help you much at all.

2) Value
Ryzen's value is largely from the low cost, not just raw performance. These higher priced chips aren't much better at all, and due to the price they actually pale in comparison to the 3600 and 3700X's value. One may make an argument for "more performance" but if you don't care about power consumption, then Comet Lake's raw performance over Zen2 would probably be a good idea if you don't want to wait for Zen3's release by the end of the year and don't need/care about the multi-core performance from AMD's Ryzen CPUs over competing Intel CPUs. These XT chips do not replace any CPUs in price or production, they're just stacked on top of the product stack. The 3600XT looks good, but depending on what you're running on your system at any given time, the extra cores of the 3700X (only around 40$ more) will probably be more valuable and useful, and when the time comes when 6 cores become less viable, 8 core users will be fine.

Conclusion: It's no surprise to most people that the refresh CPUs are a pointless flop, and that users buying Ryzen should be focusing on the more valuable chips (based on what they need), such as the Ryzen 3 3100 (or 3200G if you don't need a dedicated GPU), Ryzen 5 3600, Ryzen 7 3700X, and Ryzen 9 3900X.

[r.linder]

Rumpelcrutchskin님이 먼저 게시:

Kurumi Tokisaki님이 먼저 게시:

By then, Intel will have the next big thing. AMD will never beat Intel in gaming performance, funny how people have been rooting for AMD for over a decade but they've never come close.

Yeah Intel is prepping the 11th gen and it`s still 14 nm... can you smell the desperation in the air.

If it's still 14nm then it's probably going to be another refresh, and that means pretty much the same IPC while AMD's IPC is expected to increase at least as much as it did from Zen+ to Zen2 (which was 12%)

[Pharaoh (Revolution Idol)]

Rumpelcrutchskin님이 먼저 게시:

Kurumi Tokisaki님이 먼저 게시:

By then, Intel will have the next big thing. AMD will never beat Intel in gaming performance, funny how people have been rooting for AMD for over a decade but they've never come close.

Yeah Intel is prepping the 11th gen and it`s still 14 nm... can you smell the desperation in the air.

I think both of the following statements should be viewed as pretty much fact by now:

- AMD is closer than ever to parity with Intel in recent memory. Their products are now very competitive.

- Die shrinks aren't all they're cracked up to be. I stopped getting excited after 22nm. 65nm to 45nm was the last massive leap imo. That was a night and day difference in efficiency and performance.

[r.linder]

JohnChronic420님이 먼저 게시:

Rumpelcrutchskin님이 먼저 게시:

Yeah Intel is prepping the 11th gen and it`s still 14 nm... can you smell the desperation in the air.

I think both of the following statements should be viewed as pretty much fact by now:

- AMD is closer than ever to parity with Intel in recent memory. Their products are now very competitive.

- Die shrinks aren't all they're cracked up to be. I stopped getting excited after 22nm. 65nm to 45nm was the last massive leap imo. That was a night and day difference in efficiency and performance.

It's not so much the die shrink itself, it's just that Intel has promised a change for a number of years and still hasn't done so for various reasons and excuses. There are also a number of theoretical improvements to be made with die shrinks, one of which being power efficiency. TSMC's 7nm node in Ryzen 3000 chips is pretty damn efficient, considering a 12 core CPU uses half as much power as Intel's 10 core 10900K on average and at max load.

[Autumn_]

JohnChronic420님이 먼저 게시:

Rumpelcrutchskin님이 먼저 게시:

Yeah Intel is prepping the 11th gen and it`s still 14 nm... can you smell the desperation in the air.

I think both of the following statements should be viewed as pretty much fact by now:

- AMD is closer than ever to parity with Intel in recent memory. Their products are now very competitive.

- Die shrinks aren't all they're cracked up to be. I stopped getting excited after 22nm. 65nm to 45nm was the last massive leap imo. That was a night and day difference in efficiency and performance.

Node shrink =/= die shrink.

[Pharaoh (Revolution Idol)]

Seems like you're still picking up what I'm laying down though.

Edit: I get what you're saying though. Sometimes the extra density is used to design a new die that's the same size but with more complexity. Intel was known for a while for just die shrinking though.

[Pharaoh (Revolution Idol)]

Escorve님이 먼저 게시:

It's not so much the die shrink itself, it's just that Intel has promised a change for a number of years and still hasn't done so for various reasons and excuses.

Lol, yeah I guess they have been talking their big 10nm game up for literally years now.