News:

Willkommen im Notebookcheck.com Forum! Hier können sie über alle unsere Artikel und allgemein über Notebook relevante Dinge disuktieren. Viel Spass!

Main Menu

Intel Core i9-12900K swats Ryzen 9 rivals and i9-11900K away with blazing CPU-Z single-thread benchmarks

Started by Redaktion, October 31, 2021, 12:31:53

Previous topic - Next topic

Redaktion

A couple of CPU-Z benchmarks have surfaced showing the Intel Core i9-12900K achieving astonishing single-thread results. The new Alder Lake hybrid processor smashed its way past anything by an AMD Ryzen 9 chip and also demolished the i9-11900K. However, it seems more than likely that such single-thread supremacy comes at a high power cost.

https://www.notebookcheck.net/Intel-Core-i9-12900K-swats-Ryzen-9-rivals-and-i9-11900K-away-with-blazing-CPU-Z-single-thread-benchmarks.576387.0.html

JayN

"But as most people interested in processor development and the CPU market will know, this Intel single-thread superiority always comes at a power cost."

Always?  Or just while boosting to run the synthetic compute stress tests?

Intel slides claim the Gracefield efficient cores operate at 80% less power than Skylake cores at ISO performance, so these chips apparently have both the capability to run at very low power while performing normal work and the capability to kick in high performance processing for demanding applications.

see tomshardware article "Intel 12th-Gen Alder Lake Release Date, Pricing, Benchmarks, Specs, and All We Know"

schembfs

Makes one wonder about the possibilities if Intel could build this using the newest 5nm or even smaller process'. Because the reality is, "Intel 7" its actually 10nm which is holding this architecture back. A denser 5nm design would allow for even bigger cache' and more cores or lower power requirements (cooler and more efficient).

n7

Quote from: JayN on October 31, 2021, 16:19:36
"But as most people interested in processor development and the CPU market will know, this Intel single-thread superiority always comes at a power cost."

Always?  Or just while boosting to run the synthetic compute stress tests?

Intel slides claim the Gracefield efficient cores operate at 80% less power than Skylake cores at ISO performance, so these chips apparently have both the capability to run at very low power while performing normal work and the capability to kick in high performance processing for demanding applications.

see tomshardware article "Intel 12th-Gen Alder Lake Release Date, Pricing, Benchmarks, Specs, and All We Know"

Total FUD as usual, this time changing 40% to 80% and from a marketing slide on top of that, with no third party verification whatsoever.  Keep pumping at loser INTC stock.


schembfs

Intel should swallow it's pride and setup a contract with TSMC(or Samsung) to build a limited edition (read: more expensive) Alder Lake in 5nm with 24 cores/48 threads (ala Threadripper) with a much larger cache. Make it a KF edition with no iGPU and market it to super users(rich people) and the server market. They really dropped the ball in the build process R&D and focused too much on architecture. Because now they have a product with an excellent architecture design but stuck in a 3+ year old build process (10nm). Result: Not power efficient / runs hot / wasted potential.

Oldmanelements

Quote from: schembfs on October 31, 2021, 20:33:03
Makes one wonder about the possibilities if Intel could build this using the newest 5nm or even smaller process'. Because the reality is, "Intel 7" its actually 10nm which is holding this architecture back. A denser 5nm design would allow for even bigger cache' and more cores or lower power requirements (cooler and more efficient).

You don't even have an clue.. TMSC 7nm is a farce and they specifically uses different standards to make their process seem better Intel 10nm is the equivalent transistor size of tmsc 7nm hence the Intel 7 designation.. you can't compare the arbitrary NM measurement between totally different processes..

LyntonB

@oldmanelements I don't think it's a farce mate, you're completely choosing to overlook the hundreds of watts more power to achieve a small lead over a processor family a year/two old now. It's nice to see them trying to keep up and keep competition going but Intel are well behind the curve right now

fbk

@LyntonB
"hundreds of watts"? 241w is the peak power draw for 12900K. Intel is finally putting the TDP figures behind them because they were misleading. Ryzen 5950x, despite having a 105w TDP can also draw well over 200watts.

Dk sharma

The titile should be " Intel Core i9-12900K swats Ryzen 9 rivals and i9-11900K away with blazing CPU-Z single-thread benchmarks by consuming 1 GW of power"

Robert T

I think my biggest issue is all the tech news sources acting like this is some huge monumental thing. OMG a cpu set to release in 2022 is showing signs of outperforming a cpu released in 2020; well no s***. If it didn't then everyone at Intel should be fired and replaced for incompetence.

Will

Quote from: schembfs on November 01, 2021, 01:08:04
Intel should swallow it's pride and setup a contract with TSMC(or Samsung) to build a limited edition (read: more expensive) Alder Lake in 5nm with 24 cores/48 threads (ala Threadripper) with a much larger cache. Make it a KF edition with no iGPU and market it to super users(rich people) and the server market. They really dropped the ball in the build process R&D and focused too much on architecture. Because now they have a product with an excellent architecture design but stuck in a 3+ year old build process (10nm). Result: Not power efficient / runs hot / wasted potential.


They need to sell there foundry's to tsmc and use them exclusively, IMO nobody holds a candle to tsmc fabs. There's a reason AMD dropped global for tsmc. Nvidia tried to go Samsung but now are back to tsmc and it bet there cards get a big boost.

TheBeast

The way people are throwing Nm around like it actually means anything tangible. To dumb it down Nm is for all intents and purposes a version number and just like software you wouldn't compare Microsoft Excels version number against Google Sheets. Fact is Intel 7 (10Nm if you will) has higher transistor density than TSMC 7.

Getting over the marketing term Nm and looking at the smaller details way way way beyond my simplistic sentence would bode well for anyone wanting to understand this further.

evolucion8

Quote from: fbk on November 01, 2021, 12:16:04
@LyntonB
"hundreds of watts"? 241w is the peak power draw for 12900K. Intel is finally putting the TDP figures behind them because they were misleading. Ryzen 5950x, despite having a 105w TDP can also draw well over 200watts.

Nope. Ryzen 5950X can go up to 142W at small intervals. Current Intel CPU is peaking nearly over 100W more. So Elder Lake failed.

Quick Reply

Warning: this topic has not been posted in for at least 120 days.
Unless you're sure you want to reply, please consider starting a new topic.

Name:
Email:
Verification:
Please leave this box empty:

Shortcuts: ALT+S post or ALT+P preview