• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Intel Core i9-11900K CPU-Z Benchmark Score Leaks

360 AIO / Conductonaut , stock.
 

Attachments

  • Capture11878.PNG
    Capture11878.PNG
    36.5 KB · Views: 145
Not that impressive, to be honest. Seems like a rushed stop-gap from Intel in an attempt and take the wind out of AMD's sails...

Will hold further judgement until we see the performance per watt(where Intel has historically lagged far behind AMD) and MSRP.
 
Single core performance is the real deal and lets compare 8Core Intel with 8Core AMD. So it seems will be intel faster again, but with more power use...
 
Nothing is sacraficed! Best 8 cores multithread I see.
The 5800x is above the 11900k in the multi-threaded graph.
If the 11900k is priced above the 5800x, at or near the 5900x's price, it won't even be worth considering in most use cases(assuming 5800x & 5900x are in stock).
 
Det0x: has a 5950x 844.3 on single thread in cpu-z on here in the forums.

Haha ... this was made with special watercooling maybe outside. At Stock this 11900k is faster than ryzen 5950x at stock. With OC this 11900k is the fastest SC CPU.

Here is the problem: you have to pay again 550 Bucks for this i9 with only 8 cores. At the end of the year you will see Alderlake (new socket1700, new DDR5, maybe PCI-E 5.0).
 
There are two aspects here:
- As an SKU, performance should end up close to 5800X. With a full node disadvantage, power is a problem and pretty certainly that is why multi-core result falls behind. We will have to wait and see what the power consumption will look like but no doubt it will be higher compared to 5800X. Price is up to Intel, they are in disadvantage and should price it lower. On the other hand it is Intel so...
- Architecturally and from "IPC" perspective 3% ahead on a presumably 6% faster clock speed puts it close enough to Zen3. Rocket Lake's Cypress Cove is basically straight Sunny Cove backport. Willow Cove in Tiger Lake adds more cache and some optimizations around that. This puts Intel's architectural state on par with AMD.
 
There are two aspects here:
- As an SKU, performance should end up close to 5800X. With a full node disadvantage, power is a problem and pretty certainly that is why multi-core result falls behind. We will have to wait and see what the power consumption will look like but no doubt it will be higher compared to 5800X. Price is up to Intel, they are in disadvantage and should price it lower. On the other hand it is Intel so...
- Architecturally and from "IPC" perspective 3% ahead on a presumably 6% faster clock speed puts it close enough to Zen3. Rocket Lake's Cypress Cove is basically straight Sunny Cove backport. Willow Cove in Tiger Lake adds more cache and some optimizations around that. This puts Intel's architectural state on par with AMD.

A 3rd aspect, i think: if this 11900K is a highly binned CPU, then Intel will have problems with availability. No word yet if this is or not the case but, considering it's a in a super refined node, odds are that it is.

One thing the 5800X is not is a highly binned CPU.

We'll just have to wait and see ...
 
For me this is truly representative of superior performance because ST + 8 cores remain the optimal configuration. Most of AMD's success is build upon the 10nm failure and not genuine architectural innovation.
Spoken like a TRUE-BLUE Inhel fanbuoy.

It's good to see competition out there and this is how it should be, one pushing the other in a tight race for the crown instead of a one-sided affair as it had been before.
 
Haha ... this was made with special watercooling maybe outside. At Stock this 11900k is faster than ryzen 5950x at stock. With OC this 11900k is the fastest SC CPU.

Here is the problem: you have to pay again 550 Bucks for this i9 with only 8 cores. At the end of the year you will see Alderlake (new socket1700, new DDR5, maybe PCI-E 5.0).
pcie 5? lol just as a reminder intel barely started to have pcie4 available on its motherboard when its available for over a year on AMD side.

DDR5 will be on both side tho, and so far AMD tend to use ram speed better than Intel does. unless intel changes something significant in their next architecture.
 
Ahhh, the benchmark that was gimped because Ryzen CPUs started beating Intels... Yeah, so this benchmark result must be fact!
 
People should not read too much into any synthetic benchmark scores, and certainly not extrapolate this into generic performance. Wait for proper benchmarks of real world applications.

The mainstream segment will be heating up with good products from both camps, and 6- and 8-core models is where most PC builders should be looking anyway. Those who have real workloads scaling beyond 8 cores generally also benefit from more memory channels, more PCIe lanes, sometimes even ECC, etc. I wish the next round of Threadrippers would start at 12 cores, not 24.
 
intel going back to 8core cpu ???
Yap they need decrease cores becouse nuclear reactor need fc 360AIO ...
 
Nice as this number may look on a sheet not accompanied by any other info, if that 250w number i see floating around is true, then it is not all that impressive.
 
Single-core performance is important but let's not confuse single-thread benchmark runs (1T loaded and all other cores reserved and asleep) with real world applications where 1-2 threads are fully loaded and several others are ticking over with background tasks, probably averaging 5 cores active at all times and 20-30% CPU load, rather than the 6% CPU load of a pure 1T test.

Intel have long been great in single-core benchmarks by simply turning the power consumption up to 11 and flaying the absolute snot out of one core. For a game where typically 2-3 threads are in heavy use, and all cores are awake to run background threads you're going to need an all-core overclock and some insane cooling to get your 5.3GHz.

Rocket lake finally brings the 'Cove' architecture gains to desktop but still doesn't fix the boost/base clock disparity nor the massive power consumption spikes that go hand in hand with a 250W PL2 limit to achieve such madness.

11400F should be the best budget gaming option on the market IF priced below $200. AMD blew it with 5600X pricing.
AMD blew it with 5600X pricing is past tense, and you're comparing it to a future tense hypothetical processor that doesn't yet have a launch date, final specs, or price?

How did you work that one out? As far as I see it, the 5600X is priced against intel 10th Gen and it's sweeping the floor with them.

If you somehow believe that there won't be an AMD answer to Intel's 11400F when it hits the market, then you need to take off your Intel-branded spectacles. AMD and Intel price things as high as they can get away with because they're in the business of making profit first and foremost. If the 11400F is a great gaming CPU for budget gamers, then that's great news but don't expect it to sit far outside the price/performance curve that both AMD and Intel use to extract the maximum profit from each model. Chances are high that AMD will release a vanilla R5 5600 at that point, probably close to the $200 price point, but subject to change based on Intel's pricing of the 11th-gen i5 models.
 
No one can buy it.
 
For me this is truly representative of superior performance because ST + 8 cores remain the optimal configuration. Most of AMD's success is build upon the 10nm failure and not genuine architectural innovation.
Um, did you forget about Bulldozer and Excavator architecture before the move to Zen architecture? There is innovation there, but I guess Intel fanboys are blind to that... When a 6 core 5600X can meet and beat Intel's top of the line 8 core 10900K, it takes innovation. Drop the fanboyism and get your facts straight. Did I mention how much more efficient AMD chips are right now compared to Intel chips that get so hot I could cook dinner with one?
 
This chip is probably massive in comparison with a 5800X not just because of the uarch and the ancient 14nm density but also that iGPU that they insist to put in every single CPU no matter it's target market. Even now when they are literally running out of transistors they refuse to let go of that dead slab of silicon.
 
I have a feeling Intel is still working on a 10 core 20 thread successor but it won't be ready for awhile, they'll probably call it the i9 11950K or some BS, I just cannot fathom a core count decrease.
 
Intel have long been great in single-core benchmarks by simply turning the power consumption up to 11 and flaying the absolute snot out of one core.
Not really. At least since Zen2, AMD is doing that to a relatively higher degree. Zen2 and Zen3 push a good 20+W into one core to get the boost speeds. These 1.4-1.5V voltages are doing their job :)
What Intel keeps doing is mess with power limits to not throttle multi-core performance too much.
 
People should not read too much into any synthetic benchmark scores, and certainly not extrapolate this into generic performance. Wait for proper benchmarks of real world applications.

The mainstream segment will be heating up with good products from both camps, and 6- and 8-core models is where most PC builders should be looking anyway. Those who have real workloads scaling beyond 8 cores generally also benefit from more memory channels, more PCIe lanes, sometimes even ECC, etc. I wish the next round of Threadrippers would start at 12 cores, not 24.

I'd like a single CCX 9 core myself. Basically the 3300X Threadripper equivalent would be great. More affordable and performs optimal. Now if they could do a 12c single CCX part that would also be quite interesting, but I think the pricing wouldn't be as a optimal and peak frequency scaling might not be as ideal for a lone wolf CCX.
 
Not really. At least since Zen2, AMD is doing that to a relatively higher degree. Zen2 and Zen3 push a good 20+W into one core to get the boost speeds. These 1.4-1.5V voltages are doing their job :)
What Intel keeps doing is mess with power limits to not throttle multi-core performance too much.
Fair enough, but doesn't the stock PL1 limit even lightly-threaded workloads like gaming after about two minutes? I haven't actually built and tuned an Intel gaming rig since the 9900K which was already so roasty-toasty that I decided a 360 AIO was mandatory because I wasn't willing to ship a system with an air cooler over 1KG hanging off the socket. The only way I could keep performance reasonable was by editing the PL1 limit to 207W to match PL2. At the stock setting clocks nosedived to the point that he could have just saved €300 and bought a 3700X, it was that bad....

I have a feeling Intel is still working on a 10 core 20 thread successor but it won't be ready for awhile, they'll probably call it the i9 11950K or some BS, I just cannot fathom a core count decrease.
The 10-core was a stunt anyway. Skylake's topology was never intended for more than 6 cores and 8-core chips like the 9900K were already pushing the limits of cache and transport across a monolithic die using that architecture.

Rocket Lake is limited by 14nm+++ and won't go any bigger than 8C for manufacturing (and power consumption) reasons, whilst the successor will be Alder Lake which I believe stops at 8 "cove" cores and fills up the multi-threaded performance with Atom cores instead.
 
AMD blew it with 5600X pricing.
If AMD priced 5600x anywhere close to 3600xt let alone 3600 (non X) it would sell more than all the sales of 5xxx combined atm, not to mention their margins would suffer. They didn't blew it as much as they had little choice, AMD's probably selling more (volume) chips than Intel+Nvidia combined right now in the DIY space.
 
Single core performance is the real deal and lets compare 8Core Intel with 8Core AMD. So it seems will be intel faster again, but with more power use...

We should compare stuff at the same price level as that is the competition.
 
Back
Top