AMD Announces Zen 4C Cores Coming To Ryzen Laptops

[pegasus]

Zen4c cores are rather amazing for high throughput scenarios. You can easily get 10k of them per rack and cooling them is not an issue. It's sooo nice to see millions of jobs chewed through per day

[pieman]

Yeah kind of odd how AMD went the opposite direction.
Makes me wonder if they'll make a CPU with a group of C cores, standard cores, and X3D cores.

according to moore's law is dead podcast, that's amd's plan for future zen generations. at least, with the grouping of normal size zen core and c variants. one ccd normal size for higher frequencies, higher power consumption, and the other ccd full of c variants for lower frequency but lower power consumption. amd isn't against intel's idea of power and efficiency cores. amd just doesn't like the mismatching isa's.

[qarium]

I don't understand: other than lower boost clocks, what's actually different about these cores? With AVX-512 and SMT, what's actually sacrificed? And, what's preventing these cores from boosting higher?

i can tell you some details for example the sram cells in ZEN4 it is 8T cells means they use 8 tranistors to make 1 bit of sram
in ZEN4c they use new highend TSMC 6T cells who only use 6 tranistors to make 1 bit of sram.

this alone save 25% of chip area but the 8T sram cells are faster in the meaning they can handle higher clock speed,
the other difference is that the zen4C only have the half of L3 cache...

[qarium]

The only reason it seems to be more power efficient is because of lower clocks, run a normal Zen 4 at the same 4C clocks and efficiency should be very similar.

keep in mind it is sram cells 6T vs 8T sram cells... 25% more tranistors for the same cache size burn more energy.

so it is not only the lower clocks what makes zen4c more power efficient.

[Anux]

this alone save 25% of chip area

according to my link above it only saves low single digit % of chip area and L1-6T compared to L1-8T saves 32%.

keep in mind it is sram cells 6T vs 8T sram cells... 25% more tranistors for the same cache size burn more energy.

Yes that might explain the small difference in their efficiency graph.

[qarium]

according to my link above it only saves low single digit % of chip area and L1-6T compared to L1-8T saves 32%.
Yes that might explain the small difference in their efficiency graph.

another reason is that high clock tranistors need to be bigger in size to reach the high clock.
this means the bigger the tranistor the more electrons you need to make the on/off switch of the tranistor..
this is another reason why the ZEN4c is more power efficient they have smaller tranistors and you pay for this with much lower clock speed.

anyway i think AMDs solution is much better than intels big.little(efficient) design

the point that intels solution is different ISA for the big cores and different isa for the little cores makes it a bad idea

AMD solution with the same ISA for both cores is much better design also AMDs double pumped 256bit emulation of AVX512 is much better than what intel has.

right now AMD has 2 different asymetric concepts one is big.X3D and the other is zen4Big.zen4cLittle...

in the future AMD will have a 3way asymetric design with Big.X3Dbig.Zen5cLittle... all 3 with the same ISA...

we have basically 3 kind of apps one that like high clock speeds and one that likes much cache and one that only like many cores at low clock speed.

with such a asymetric design: Big.X3Dbig.Zen5cLittle. AMD would catch every single user-case.

[M@GOid]

Those things will be great on the next wave of portable game consoles. Valve must be looking at those for a Steam Deck 2.

[Anux]

this is another reason why the ZEN4c is more power efficient they have smaller tranistors and you pay for this with much lower clock speed.

Do you have any reference to this? I've heard it the first time, my assumption till now is that transistors are same size but closer together.

[Jedibeeftrix]

still think a 5nm 8core Zen4c / 8CU RDNA3.5 APU would make a great steamdeck 2 chip.

[qarium]

Do you have any reference to this? I've heard it the first time, my assumption till now is that transistors are same size but closer together.

i did it read in some zen4c article,... if i remember correctly it was this article: https://www.semianalysis.com/p/zen-4...-to-hyperscale
but just take your words and think about it: bigger tranistors or same sitze tranistors but closer together results in the exact same result you need more chip area for the same tranistors...
its litterally the same the result is the same.
for me it makes no difference if the tranistor is smaller or only closer together the result is the same higher density for zen4c...

also about the electron flow you need for the same result if your distance is higher between the tranistors the electrons cause more heat on the wire...