I wonder what the gains would be if it reaches the rumored 5.5ghz on a live sample my guess is another 5% which won't really put Zen 4 in all that great of a spot to win over Raptor Lake.

ES sample usualy runs slower than retail chip i expect another 10% increase over zen3.

Kool64:

I wonder what the gains would be if it reaches the rumored 5.5ghz on a live sample my guess is another 5% which won't really put Zen 4 in all that great of a spot to win over Raptor Lake.

Where are you getting 5% from? 4.4Ghz to 5.5Ghz is a 25% increase in instructions executed.

Richard Nutman:

Where are you getting 5% from? 4.4Ghz to 5.5Ghz is a 25% increase in instructions executed.

yeah but what I'm talking about is the % increase in performance which almost certainly does not equate to a 1 to 1 per clock. so if a Zen 4 chip is 10% better than the best Zen 3 chip at 4.9 maybe 5.0 as the 5950x usually hits no problem we can expect maybe 5-10% more with a Zen 4 chip running at 5.5.

Undying:

ES sample usualy runs slower than retail chip i expect another 10% increase over zen3.

of course we'd expect retail to be better i'm just worried about the gains on clock speed when we take into account the architecture is generally worked out as it's just a port of Zen 3.

Kool64:

yeah but what I'm talking about is the % increase in performance which almost certainly does not equate to a 1 to 1 per clock. so if a Zen 4 chip is 10% better than the best Zen 3 chip at 4.9 maybe 5.0 as the 5950x usually hits no problem we can expect maybe 5-10% more with a Zen 4 chip running at 5.5.

So long as you're doing a CPU-intensive workload and don't have any other bottlenecks, there is a 1:1 increase. Seeing as this is a 6-core on DDR5, RAM is unlikely to be a bottleneck. I know nothing about Basemark so perhaps the CPU might not be fully utilized (where an increase in clock speed wouldn't help). However, 5.5GHz isn't sustainable, so, there could still potentially only be a 5% increase overall if it doesn't boost to 5.5 for very long. Perhaps you were confusing CPU frequency with one of the following: Power consumption doesn't change 1:1 with frequency Increasing RAM frequency doesn't cause a 1:1 improvement, at least not until it meets the bandwidth demands of the CPU Additional cores doesn't yield a 1:1 improvement in most scenarios

Kool64:

I wonder what the gains would be if it reaches the rumored 5.5ghz on a live sample my guess is another 5% which won't really put Zen 4 in all that great of a spot to win over Raptor Lake.

It doesn't matter in any useful program if it reaches 5.5ghz in 2 ms. What we want to see is what frequency it can maintain 🙂 My 5950x is more like 4.6ghz all core with not too much voltage. Far away from pbo oc boost at 5150mhz like 2ms LOL. Fun times with zen 4 for shure 🙂

nizzen:

It doesn't matter in any useful program if it reaches 5.5ghz in 2 ms. What we want to see is what frequency it can maintain 🙂 My 5950x is more like 4.6ghz all core with not too much voltage. Far away from pbo oc boost at 5150mhz like 2ms LOL. Fun times with zen 4 for shure 🙂

of course we'll probably see 5.2 5.3 average if 5.5 is even a thing to begin with.

schmidtbag:

However, 5.5GHz isn't sustainable, so, there could still potentially only be a 5% increase overall if it doesn't boost to 5.5 for very long.

This essentially sums up what I was saying

Undying:

ES sample usualy runs slower than retail chip i expect another 10% increase over zen3.

from your lips to Lisa Su's orders