Intel Prepares Raptor Lake Designs With 24 Cores and 32 Threads, More E-Cores This Time

[Dr. Dro]

Obviously AMD hates Argentina.

https://www.fullh4rd.com.ar/prod/12425/micro-amd-ryzen-5-3600
AMD RYZEN 5 3600 $35.637,00

https://www.fullh4rd.com.ar/prod/17680/micro-amd-ryzen-5-5600x
AMD RYZEN 5 5600X $42.290,00

https://www.fullh4rd.com.ar/prod/18814/micro-intel-core-i5-11400f-sin-video
INTEL CORE I5 11400F $34.890,00

Those prices are only moderately higher than what currently practiced by Brazil's largest hardware store and AMD's largest licensed retailer (KaBuM!), where I bought my R9 5950X and my 3090. That store is asking roughly the equivalent of 2200 BRL for the 5600X, this processor can be obtained at 1750 BRL currently, for a smaller hardware store, that is not so bad. The price with Mercado Pago payment processor is probably due to the commission Mercado Libre charges, whenever I sell stuff through there, they take 17% in the premium plan (giving buyer ability to finance in 12 installments + free shipping and featuring my listing in ads).

The numbers seem quite stratospheric, but in reality, it's more of Americans often not being aware of how good they've got things. It's mostly a result of the combination of it being a small store, taxes and the Argentinian economy's very poor performance. In general, though, the COVID-19 pandemic's economic recoil in South America (including Brazil) has been generally felt recently because of our devaluing currency. 1 USD is currently trading for 5.60 BRL and just north of 101 ARS.

Thats intel's ultimate goal.. imagine the multi core performance of a 8+100 core cpu insane

As long as the prices are kept in check, I welcome this development with open arms.

 

[ShurikN]

Did you not see the E-core review that W1zz did? At 4k...

I have. Have you? The application performance is poor compared to P-cores.

At 4k in gaming even bulldozer performs well... Can't believe you used 4k gaming as an argument for cpu performance.

 

[Aquinus]

I have. Have you? The application performance is poor compared to P-cores.

At 4k in gaming even bulldozer performs well... Can't believe you used 4k gaming as an argument for cpu performance.

It shows that they're not nearly as useless as you're claiming. Also Bulldozer would probably use a lot more power doing the same thing.

Would you like to show us on a doll where Intel hurt you?

 

[Punkenjoy]

It shows that they're not nearly as useless as you're claiming. Also Bulldozer would probably use a lot more power doing the same thing.

Would you like to show us on a doll where Intel hurt you?

At 4K, you are really GPU limited meaning a smaller/lower performance CPU will be enough to do the task.

I do not think that it prove they are useful at all in gaming. It just prove that at 4k, people should really focus on upgrading their GPU before CPU.

In game where you really need high IPC, the E-core would suck anyway.

In game that are CPU limited due to the main core having too much thing in serial, they wouldn't be utilized, and another P core would beat them due to the very high latency in scenario where they would be utilized. Also it wouldn't matter if you have 8 or 24 e core in 4K anyway, so i am not sure why someone would want to pay for all those extra cores.


That do not mean they cannot run games, they can. After all, PS4 and Xbox one had CPU that were way worst than those E-cores and managed to do fine (at 30 fps most of the time, true..)

But this is not ideal for gamers.

The E core have been designed to run background task and this is why they have some drawback like latency and lower frequency. It wouldn't matter as much for these kind of task but for high refresh rate gaming, it would not be a good thing.

 

[efikkan]

I'm unaware of any CPU made, be it AMD, Intel, or IBM POWER, or ARM, that did data-sharing in "close" caches (L1). All data-sharing is in LLC (last-level cache).

I'm referring to this:

(source)

Maybe , just maybe, sometime they will come up with 16 cores single ringbus.
But that means a 20 ring stop ringbus.
Will core to core latency become a huge concern ?

I don't think so, core-to-core communication is fairly infrequent compared to e.g. cache and memory accesses. (For comparison even a single cache miss is much more costly, and more frequent) Even most heavily multithreaded tasks will probably be bandwidth starved before this becomes a problem.

 

[londiste]

The application performance is poor compared to P-cores

Not looking it up right now but IIRC E-cores ended up at around 3300X which is not a bad result at all.
Vs 3300X E-cores were losing ground in single thread - due to lower boost clocks - and gaining ground in multi thread - due to 8c8t vs 4c8t.

 

[Mussels]

Not looking it up right now but IIRC E-cores ended up at around 3300X which is not a bad result at all.
Vs 3300X E-cores were losing ground in single thread - due to lower boost clocks - and gaining ground in multi thread - due to 8c8t vs 4c8t.

I recall seeing that as well, and it adds to the concept that AMD were looking at mixing and matching Zen generations as their own hybrid variant

So a Zen4 CPU could have 6x zen 4 cores, and 8x zen 3 cores (with tweaked power efficienct settings)

It'll be interesting to see where this ends up, as zens architecture really does seem better suited to mixing different generations together than intels version