Support NeoGAF

[KyoZz]

At CES 2021, Intel allegedly announced its 11th Gen Core Rocket Lake-S series during the CES 2021 press briefing.
Official announcement slides have been leaked as well, which shows almost 14% IPC Boost with 11th Gen CPU lineup. Some screenshots have also been shared by Geeknetic from the said presentation, showcasing Intel’s Rocket Lake-S CPU’s performance against AMD.

As mentioned previously, Intel’s new Rocket Lake-S CPU beats an undisclosed AMD 12-core Ryzen in the Metro Exodus game. Now the company’s flagship Rocket Lake Core i9-11900K processor has been tested in a variety of games in CPU limited situations.
Testing was done at 1080p resolution, and yes it includes the Cyberpunk 2077 game as well.
According to Intel the 8C/16T Core i9-11900K reclaims the gaming performance crown from AMD’s Ryzen 9 5900X at 1080p, albeit even by a small margin. Intel ran these tests with an EVGA RTX 3080 XC3 graphics card, and the company claims the lead ranges from 2% to 8%.

The gaming benchmarks which were tested include Total War: Three Kingdoms, Gears of War 5, Metro Exodus, Cyberpunk 2077, Watchdogs: Legion, Farcry: New Dawn, and Assassin’s Creed Valhalla.
All games were tested on the 1080p graphics settings to truly test the capabilities of the CPU. There appears to be a difference in performance range from 8% in Total War: Three Kingdoms game, which is a very CPU intensive title, to 2% in Assassin’s Creed Valhalla.
On an overall aggregate basis, Intel’s Core i9-11900k is about 4.4% faster than AMD’s Ryzen 5900X CPU.

AMD’s new Zen 3-based Ryzen 9 5900X actually has more cores but it appears that Intel’s IPC performance is shining through, even on a refined 14nm process Node.
Intel clocked its new Core i9-11900K processor at all the way up to 5.3GHz which is a pretty high clock speed for a CPU, thus helping it beat AMD’s Ryzen CPU in the end.
The company also compared an older generation Core i9-10900k processor (which is still based on the Skylake architecture), with the Core i9-11900k. The former flagship CPU scored 129 frames per second in Hitman 3, with the Core i9-11900k Rocket Lake CPU scoring up to 140 frames per second.

More here:

Intel's new Rocket Lake-S Core i9-11900K CPU benchmarked, beats AMD Ryzen 9 5900X in gaming

According to Intel the 8C/16T Core i9-11900K reclaims the gaming performance crown from AMD's Ryzen 9 5900X at 1080p.

www.dsogaming.com

 

[LazyParrot]

Low single digit performance advantage AND the thing will most likely double as a space heater, saving you money on your heating bill. Team blue is back, baby!

 

[LOLCats]

intel be like:

14nm for life bitches

 

[Celcius]

I’ll be keeping my 10700k. I game at 4K anyway so I’d most likely see zero difference in my usage.
Interesting to see their top cpu go back down to 8 cores though.

 

[DaleinCalgary]

Testing done at 1080p?

what year is it gif

 

[dave_d]

The big question, will they be able to come anywhere close to meeting demand or will it be another paper launch?

 

[Barakov]

I think people with a 5900x don't need to worry that much.

 

[Neo_game]

Testing done at 1080p?

what year is it gif

That is the way to test cpu performance as at higher res with more pixels it is more gpu bound than the cpu.

 

[DaleinCalgary]

That is the way to test cpu performance as at higher res with more pixels it is more gpu bound than the cpu.

Thanks!

 

[LazyParrot]

That is the way to test cpu performance as at higher res with more pixels it is more gpu bound than the cpu.

That's one reason. The other is that if the performance advantage is already this small at 1080p, it's going to be nonexistent at 4K, which wouldn't make for very good promo material.

 

[TyphoonStruggle]

Testing done at 1080p?

what year is it gif

That's usually how CPU's are benchmarked for games. Otherwise most modern CPU's give the same performance at 4k.

 

[truth411]

At CES 2021, Intel allegedly announced its 11th Gen Core Rocket Lake-S series during the CES 2021 press briefing.
Official announcement slides have been leaked as well, which shows almost 14% IPC Boost with 11th Gen CPU lineup. Some screenshots have also been shared by Geeknetic from the said presentation, showcasing Intel’s Rocket Lake-S CPU’s performance against AMD.

As mentioned previously, Intel’s new Rocket Lake-S CPU beats an undisclosed AMD 12-core Ryzen in the Metro Exodus game. Now the company’s flagship Rocket Lake Core i9-11900K processor has been tested in a variety of games in CPU limited situations.
Testing was done at 1080p resolution, and yes it includes the Cyberpunk 2077 game as well.
According to Intel the 8C/16T Core i9-11900K reclaims the gaming performance crown from AMD’s Ryzen 9 5900X at 1080p, albeit even by a small margin. Intel ran these tests with an EVGA RTX 3080 XC3 graphics card, and the company claims the lead ranges from 2% to 8%.

The gaming benchmarks which were tested include Total War: Three Kingdoms, Gears of War 5, Metro Exodus, Cyberpunk 2077, Watchdogs: Legion, Farcry: New Dawn, and Assassin’s Creed Valhalla.
All games were tested on the 1080p graphics settings to truly test the capabilities of the CPU. There appears to be a difference in performance range from 8% in Total War: Three Kingdoms game, which is a very CPU intensive title, to 2% in Assassin’s Creed Valhalla.
On an overall aggregate basis, Intel’s Core i9-11900k is about 4.4% faster than AMD’s Ryzen 5900X CPU.

AMD’s new Zen 3-based Ryzen 9 5900X actually has more cores but it appears that Intel’s IPC performance is shining through, even on a refined 14nm process Node.
Intel clocked its new Core i9-11900K processor at all the way up to 5.3GHz which is a pretty high clock speed for a CPU, thus helping it beat AMD’s Ryzen CPU in the end.
The company also compared an older generation Core i9-10900k processor (which is still based on the Skylake architecture), with the Core i9-11900k. The former flagship CPU scored 129 frames per second in Hitman 3, with the Core i9-11900k Rocket Lake CPU scoring up to 140 frames per second.

More here:

Intel's new Rocket Lake-S Core i9-11900K CPU benchmarked, beats AMD Ryzen 9 5900X in gaming

According to Intel the 8C/16T Core i9-11900K reclaims the gaming performance crown from AMD's Ryzen 9 5900X at 1080p.

www.dsogaming.com

Intel: P.S. to achieve this minor advantage the 11900k is consuming 300Watts of Power. Its a cold winter so it doubles as a space heater to keep consumers warm while they game.

 

[Type_Raver]

Thats nice, but at what temperatures and how sustainable are the peak frequencies under load?

I hope its enough for intel to keep up and make it interesting in the CPU market, competition is good.

 

[Aarbron]

intel be like:

14nm for life bitches

The new 14 inches

 

[adamosmaki]

same old same old. negligible difference in select games fewer cores and most likely will serve as a heater for the winter

 

[SantaC]

Intel says the industry should switch from "benchmarks to the benefits" of tech

Oh, and Tiger Lake is still coming later this summer

www.google.com

 

[Kenpachii]

Can't wait for 2 fps more on 100 fps.

 

[small_law]

I'm upgrading for PCIe 4.0 support. From Intel. Good night everybody.

 

[Sean Mirrsen]

I don't get these benchmarks done with modern AAA games. None of them meaningfully load the CPU anyway.

Want to tell me how powerful your gaming CPU is, tell me how big a rocket I can build in KSP with it, or how long it takes to generate a huge thousand-year old world in Dwarf Fortress.

 

[Allandor]

Intel benchmarks.

2 less cores + higher boost = they might get those +5% with a much higher power target. But after all, they always cheated and cherry picked their benches. Even if it wasn't necessary.

 

[FireFly]

This thread is invalid. It wasn't created by Leonidas.

 

[ShirAhava]

1080p? They need to throw in some 720p tests and REALLY show em' who's boss!

 

[alucard0712_rus]

Doesn't 14nm cheaper to produce?

 

[Hudo]

Question is whether it will match the $600 price tag of the Ryzen 5900x or if it, like Intel always do, will be horribly overpriced and costs $800+

 

[Deleted member 17706]

Going down in core count? Interested in seeing their non-gaming productivity benchmarks.

Did they really sacrifice two cores just to ever so slightly beat the R9 5900x in gaming?

 

[Rentahamster]

Single digit improvements while sacrificing two cores is not impressive.

 

[Kadve]

Question is whether it will match the $600 price tag of the Ryzen 5900x or if it, like Intel always do, will be horribly overpriced and costs $800+

Yes, a bit stupid to compare to the normal Ryzen series when it might be more comparable to the Threadrippers.

 

[Low Moral Fibre]

Intel rehashing 14nm like Marge Simpson and that Chanel Suit.

 

[V1LÆM]

i'm not replacing my 9900K until Alder Lake which will be 10nm, DDR5, and PCIE 5.0

 

[Pakoe]

intel be like:

14nm for life bitches

Hey hey hey, this is 14nm++ okay??

 

[The Janitor]

But when are we gonna get intel motherboards with pcie 4.0 support?

 

[Skifi28]

Cutting corners for that 1%, literally.

 

[llien]

Let's be clear on 14nm vs 7nm guys, the left figure is from a company that is not in Fab business, unlike the right one.

Actually measured transistor size (CPU cache):

TSMC 7nm - 22nm x 22nm
Intel 14nm - 24nm x 24nm

Low single digit performance advantage

... at 1080p.... and on Intel's slides...

 

[The Skull]

Done with DDR3200 as well. Ryzen gets a nice boost with higher frequency ram/timings. I hope Gamers nexus do a tuned shootout between the 11900k and the 5000 series.

 

[Trimesh]

The big question, will they be able to come anywhere close to meeting demand or will it be another paper launch?

Presumably this is another hand-picked SKU - so it's going to depend on how the wafers come out of the fab. I can't help thinking that Intel are painting themselves into a corner with this approach, though - we have reached the point where you no longer get automatic clock speed gains from scaling, so if their top SKU is based on the best performance you can get out of a highly optimized and extremely well characterized process it could be a significant challenge to top it with a new design.

 

[x@3f*oo_e!]

I don't get these benchmarks done with modern AAA games. None of them meaningfully load the CPU anyway.

Want to tell me how powerful your gaming CPU is, tell me how big a rocket I can build in KSP with it, or how long it takes to generate a huge thousand-year old world in Dwarf Fortress.

I don't think either of those are using much SIMD, or in the case of Dwarf Fortress - any float math.. Might be wrong.

Also AVX2 needs testing since it seems to thermally throttle some chips.

 

[Chiggs]

Yes! 1080p gaming! Just liked I always dreamed of!

 

[Insane Metal]

14nm still, and 8 x 12 cores so that's actually a good result. Can't wait to see what they'll come up with their 10nm, it'll be a beast imo.

 

[x@3f*oo_e!]

Let's be clear on 14nm vs 7nm guys, the left figure is from a company that is not in Fab business, unlike the right one.

Actually measured transistor size (CPU cache):

TSMC 7nm - 22nm x 22nm
Intel 14nm - 24nm x 24nm

I think you're looking at distorted/non-universal results (from where btw?) because it's SRAM .. "10nm" TSMC/Samsung is as good as 14nm Intel, 7/8 nm TSMC/Samsung is better, in terms of density, and much much better for power efficiency

 

[x@3f*oo_e!]

[edit] Not Samsung 8nm (it's just 10nm++++) ignore that

 

[notseqi]

1080p? They need to throw in some 720p tests and REALLY show em' who's boss!

Software mode games like the good old HL-Engine, no graphics card at all, see who's ballin' then.

 

[ethomaz]

Testing done at 1080p?

what year is it gif

CPU gaming benchmarks are done in lower resolutions to make sure GPU is not a bottleneck.

 

[Mister Wolf]

Its pointless. 1440p and up the GPU becomes the bottleneck. Dont spend alot of money on your CPU. That money should always go towards a better GPU.

 

[llien]

I think you're looking at distorted/non-universal results

I'm looking at results of a German dude actually examining the chips.

7/8 nm TSMC/Samsung is better, in terms of density

That... does not contradict what I've stated above.

 

[x@3f*oo_e!]

In other news (did nobody mention this yet?)

Intel Talks With TSMC, Samsung to Outsource Some Chip Production

Intel Corp. has talked with Taiwan Semiconductor Manufacturing Co. and Samsung Electronics Co. about the Asian companies making some of its best chips, but the Silicon Valley pioneer is still holding out hope for last-minute improvements in its own production capabilities.

www.bloomberg.com

It's a rumour, Intel on 4nm TSMC in 2022/3 .., testing chips on 5nm (4nm is "5nm+") - it's Bloombreg though so no immediate discredit

Intel Corp. has talked with Taiwan Semiconductor Manufacturing Co. and Samsung Electronics Co. about the Asian companies making some of its best chips, but the Silicon Valley pioneer is still holding out hope for last-minute improvements in its own production capabilities.

After successive delays in its chip fabrication processes, Santa Clara, California-based Intel has yet to make a final decision less than two weeks ahead of a scheduled announcement of its plans, according to people familiar with the deliberations. Any components that Intel might source from Taiwan wouldn’t come to market until 2023 at the earliest and would be based on established manufacturing processes already in use by other TSMC customers, said the people, asking not to be identified because the plans are private.

[..]

TSMC, the largest maker of semiconductors for other companies, is preparing to offer Intel chips manufactured using a 4-nanometer process, with initial testing using an older 5-nanometer process, according to the people. The company has said it will make test production of 4-nanometer chips available in the fourth quarter of 2021 and volume shipments the following year.

Get exited about late 2022 I suppose

 

[x@3f*oo_e!]

I'm looking at results of a German dude actually examining the chips.


That... does not contradict what I've stated above.

Yeah the guy was looking at the CPU cache (SRAM)

..

22x22mm vs 24x24mm is +20% density - the actual difference in whole chip density is way greater I think

[edit] are you sure about 22x22 - I think he only got width .. FinFETs are narrow and wide - so it might not be the same in the other direction

 

[Elektro Demon]

14nm4life

 

[LordOfChaos]

By Intels own in house benchmarks, a marginal difference in gaming, meanwhile you're giving up 4 more cores on the AMD part for it.


Meh. Has to be priced aggressively for any appeal. Otherwise it's almost the same gaming performance + more multicore performance going AMD, same situation.

Testing done at 1080p?

what year is it gif

If you push a higher load to the GPU, it would show even less difference. These are the "theoretically, it's this much faster if the GPU isn't a bottleneck" tests, the best possible light to paint a marginal CPU gaming performance difference.

 

[Kuranghi]

I have a 3770K thats barely above 60% in any game I play, but I play everything at native 4K if I can.

Are there other factors that would lower your performance in a game when you are GPU locked?

Or should I not care about upgrading my CPU (for gaming) until it actually starts getting above 80% or whenever usage it starts causing throttling issues?

 

[Leonidas]

Good to see a new high performing gaming CPU from Intel finally on a new architecture. Here's hoping there is good supply.