Tuesday, July 20th 2021
Intel Core i9-12900K Allegedly Beats AMD Ryzen 9 5950X at Cinebench R20
With qualification samples of the upcoming Intel Core i9-12900K "Alder Lake-S" processors and companion Socket LGA1700 motherboards hitting the black-market, expect a deluge of benchmarks on social media. One such that stands out makes a fascinating claim that the i9-12900K beats AMD's current flagship Ryzen 9 5950X processor at Cinebench R20, which has been AMD's favorite multi-threaded benchmark. At stock speeds, with liquid cooling, the i9-12900K allegedly scores 810 points in the single-threaded test, and 11600 points in multi-threaded.
To put these numbers into perspective, a retail Ryzen 9 5950X scores 641 points in the single-threaded test, and 10234 points in multi-threaded, in our own testing. The i9-12900K is technically a 16-core processor, just like the 5950X, but half its cores are low-power "Gracemont." The "Alder Lake-S" chip appears to be making up ground on the single-threaded performance of the "Golden Cove" P-core, that's a whopping 25% higher than the "Zen 3" core on the 5950X. This is aided not just by higher IPC, but also the max boost frequency of 5.30 GHz for 1~2 cores, and 5.00 GHz "all-core" boost (for the P-cores).
Given the multi-threaded scores, it's safe to assume that either Intel or Microsoft has figured out a way to leverage the P-cores and E-cores simultaneously in peak multi-threaded workloads. This is possible when both the "Golden Cove" and "Gracemont" cores have the ISA capability needed by the workload, which in case of Cinebench R20, is AVX. "Gracemont" is Intel's first low-power core to support AVX, AVX2, and AVX-VNNI instruction sets. "Golden Cove" features a more lavish ISA that includes AVX-512 (select client-relevant instructions).
Sources:
OneRaichu (Twitter), VideoCardz
To put these numbers into perspective, a retail Ryzen 9 5950X scores 641 points in the single-threaded test, and 10234 points in multi-threaded, in our own testing. The i9-12900K is technically a 16-core processor, just like the 5950X, but half its cores are low-power "Gracemont." The "Alder Lake-S" chip appears to be making up ground on the single-threaded performance of the "Golden Cove" P-core, that's a whopping 25% higher than the "Zen 3" core on the 5950X. This is aided not just by higher IPC, but also the max boost frequency of 5.30 GHz for 1~2 cores, and 5.00 GHz "all-core" boost (for the P-cores).
155 Comments on Intel Core i9-12900K Allegedly Beats AMD Ryzen 9 5950X at Cinebench R20
Catch you boys at the lake.. Alder Lake :cool:
Also, I don't think that reviewers are not knowledgeable enough to run the CPUs at stock. This Intel chips' behaviour is nothing new at this point in time. So any reviewer that does not know this, is really odd. The thing is you need to determine what is the reviewer trying to achieve here. Does he/she want to show their viewers, the full capability by means of unlocking the sustained clock speed, or does he/she want to show you performance strictly at stock?While this is factual, I think it is more of a power vs performance, rather than just looking at power in silo. Looking forward to see what Alder Lake can do as there is no point comparing Intel's 14nm products with AMD's Zen 3 since the latter is using a superior node.
Let's not forget that Intel's R&D budget is literally over 684% larger than AMD's and therefore Intel SHOULD be crushing AMD. If Alder lake is better than Zen4, which based on preliminary reports, I don't think it will be, that would be the LEAST Intel should be accomplishing considering their $13.56 Billion R&D budget vs AMD's $1.98 Billion budget.
If that is true, there are 2 scenario, and for some reason, i think the worst one will be the one that happen.
Scenario 1. Intel release it, price it like the 5950x forcing AMD to reduce it's price
Scenario 2. Intel release it, price it 20-30% more expensive than the 5950x so people pay for the extra performance
My hope aren't super high that it's not Scenario 2. Then AMD could release Zen 4 with similiar or better performance at similiar or higher price... It look like since few years, the only way to get more performance is to pay more.
However, its just speculation at this point without seeing some scores out in the wild. Especially seeing that this is that massive a jump over their own 11th gen I will be a little skeptical till some leaked benchmarks start coming out.
There are tons of potential power/clock/TVB/AVX reasons why these numbers might be misleading and not representative of a real CB20 score, but with nothing other than rumour to go on, we can at least hope.
I know people want AMD to win because they're the underdog with the tiny budget but as consumers all that should really matter is fierce competition that makes both companies try harder for us.
That being said I could see intel maybe matching the 5900x in performance. Beating a 16 core 5950x when 8 of intels 16 cores are sub skylake in performance seems most unlikely.
K-series used to determine whether the CPU was unlocked or not, but today it usually also affects sustained power draw and throttling. With the Skylake family and beyond, the non-K models above 4 cores are practically a waste of money, as they throttle too much. This wasn't the case back in the Haswell or Sandy Bridge days, where the real-world differences between K and non-K models were minor.Whether it's due to knowledge, or if it's just a conscious choice to not benchmark stock, is up for debate. Just judging by how little most non-TPU reviews knows about memory speeds I'm leaing towards the first option.
But regardless, any reference comparison is useless unless it's stock. It's fine to have a separate comparison of OC vs. OC, but unfortunately most CPU reviews today are actually portraying their mild OC as "stock" and then have subsection of OC in addition. Very commonly we see overclocked IF/memory on AMD samples and removed power limits on Intel systems, but neither of these are actually stock, and is useless for a fair comparison.
That said, for once, I am a little worried about the power draw. 200W+ was never an issue for CML because of thicker IHS/die thinning/just a very good thermal design. 5GHz all-core on a midrange U12S isn't far fetched at all. RKL took a step backwards in a number of ways, so it ran like an inferno.
But Alder Lake is on 10ESF, and will mark the first time that Intel experiences N7FF-esque thermal/power density on the desktop.
Under an air cooler and at stock power limits, a 7nm 5900X/5950X runs hottest with high single core (actually 2-core) boost where it scales to 4.9-5.0GHz @ 1.4V+ and 15-20W power per-core - not during MT. However, once you get to about 180-200W power draw, MT temps begin to overtake ST temps as the freq, volts and power per-core comes up.
So more than anything, it's still the thermal density as Zen 3 cores only begin spiking/acting erratically on temps when power per-core exceeds about 13W. Same deal with the 5800X, only reason the 5900X/5950X avoid that fate is by keeping per-core power down in MT (below 10W at stock), whereas 5800X is somewhere around 14W per core, above the 13W threshold.
Will be interesting to see how Intel tackles this problem, since most of the power should be going to the Golden Cove cores, of which there are only 8 to share the power budget. Being monolithic is a boon, but it changes little at 7nm/ESF density.