Tuesday, July 20th 2021

Intel Core i9-12900K Allegedly Beats AMD Ryzen 9 5950X at Cinebench R20

by
btarunr
 Discuss (155 Comments)
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

Related News

Add your own comment

155 Comments on Intel Core i9-12900K Allegedly Beats AMD Ryzen 9 5950X at Cinebench R20

#26
efikkan
matarSingle-thread 810 wow impressed a big leap from 11gen.
Considering how power limited Rocket Lake is in some applications, we should expect some applications to see huge gains from the newer node.

But don't use Cinebench to extrapolate generic performance, it's very poor at representing that, and it's not even designed for that purpose.
Posted on Reply
#27
phill
I do like the way it says its cooled in a water cooler... Big vat of water was it?! I dunno, I just wait till the TPU reviews hit the sites, all this speculation just drives me nuts and causes so many issues and divide sometimes...
Posted on Reply
#28
Unregistered
Lol at all the massive watt posts. Who fucking cares. If you have the cash to buy one, and the required DDR5 plus a new high end board, you aren't going to be cooling it with a stock cooler or using a 500watt psu. I personally don't give a hoot about power usage any more, it's performance that matters.

You don't buy a ferrari and whine cos it only does 8mpg
#29
GreiverBlade
also another question .... will the "magic trick"/"improvement" bringing those performances up turn out to be a vulnerability and the perf will go down after patching them??? :laugh: (semi serious considering Intel recent behavior :oops: )

i swear, each time i change CPU the other camp get up (are they waiting on me, or what :roll: ) well ... this time it is of no concern ... i like what i got.
Posted on Reply
#30
Melvis
Crackonglol
not even a R20 screenshot
Just rumor
Exactly! any body can do this!

6950X No OC
In water cooler
Cinebench R20
ST: 903
MT: 13 634

BOOM!

Intel 10nm is in laptops, they lost against Zen 3.....

Useless news is useless
Posted on Reply
#31
ExcuseMeWtf
Great news if it's faster and brings competition up, probably won't personally buy it anyways :D
Posted on Reply
#32
Unregistered
Raichu says it might be 200W+ in full turbo mode which aint that bad if it's as quick as it seems.
#33
Richards
dicktracyLost to little cores nonetheless. It’s game over with Raptor Lake.
Thats why they rushed and showed the v-cache because they know they.ll be destroyed performance wize
Posted on Reply
#34
Ferrum Master
Gruffalo.SoldierRaichu says it might be 200W+ in full turbo mode which aint that bad if it's as quick as it seems.
Try to cool that with sustained gaming loads using conventional methods using a mere middle end cooling solution, not a complete enthusiast overkill. 200W is too much...
Posted on Reply
#35
persondb
watzupkenI think there are a few things to consider here when looking at cost. Not only do you need to consider the cost of the processor, but you need a good motherboard and DDR5. All which are not cheap. I've been trying to get a good B560 ITX board, but they cost a lot. In the end, I had to settle for a B460 because it is outdated and thus, cheaper.
londisteIntel's intended prices have been relatively stable across recent generations. x600K costs $260-270, x700K costs $370-380 (although Rocket lake 11700K was $400). 8000-series prices were slightly less. This is a little cheaper than Ryzen 5000 for now. We will have to wait and see.

New socket brings new platform concerns this time around as well. Intel's LGA1700 will not be directly compatible with current coolers. DDR5 should also be coming. Same applies to AM5.
Alder Lake will use both DDR4 and DDR5 and likely only the high end will end up using DDR5, so I wouldn't worry too much about DDR5.
ITX boards also command a premium generally.
Posted on Reply
#36
owen10578
Unless intel did some kind of architectural ass pull magic on 12th gen to gain that much IPC then I don't see this score as real unless they used LN2 or something. 8 big cores + 8 small cores beating full fat 16 Ryzen cores seems like a stretch. But I really hope this is real because AMD really needs some competition to stop them turning into the new Intel.
Posted on Reply
#37
docnorth
Ferrum MasterThe key question. At what power? I suspect 300W+ power hog.

I can get much more out of my 5950 when removing power limits. And I can as I run custom loop.

Do I need it? Naah.
The leak says "It may be over 200W in full turbo frequency easily." It's a huge reduction compared to 11900K, but of cource that was somehow expected. Rocket Lake's consumption will be remembered for years, something like Polaris GPUs.
Posted on Reply
#38
Unregistered
Ferrum MasterTry to cool that with sustained gaming loads using conventional methods using a mere middle end cooling solution, not a complete enthusiast overkill. 200W is too much...
As i said, no one will buy a 12900k and use a stock cooler. If you can afford a 12900k you are gonna have a nice high end air cooler or full loop.
#40
Shatun_Bear
Patr!ckAnd as usual AMD fanboys keep talking about the Zen 3D variant like it will be some sort of mass production. That 3D cache variant will only find its way in the Ryzen 9 series. Ryzen 7 and below won't have it. Zen 3 laptops won't have it either. Which means Core i5 and Core i7 Alder Lake will still face the regular Zen 3 Ryzen 5 and Ryzen 7 chips. Ryzen 9 Zen 3D won't change anything because it will be a paper launched halo product with less IPC, no DDR5 and lower clock speeds. There's also the concern of what clocks will the Zen 3D chip be able to maintain with that additional cache and what level of power consumption will be involved.
You're celebrating because an Intel CPU that releases a year later or more than than an AMD CPU outperforms it while drawing a lot more power. Is this how far Intel have fallen now?

It's logical that Ryzen 5000 being relatively old by the time Alder Lake releases means there will be a refresh including V-cache higher end SKUs around the same time.
Gruffalo.SoldierRaichu says it might be 200W+ in full turbo mode which aint that bad if it's as quick as it seems.
What, 200W is way too much? This is not a 32-core.
Posted on Reply
#41
watzupken
Gruffalo.SoldierRaichu says it might be 200W+ in full turbo mode which aint that bad if it's as quick as it seems.
I believe there was a leak earlier that mentioned that the PL2 is 228W and I believe it is possible. In fact that number may be higher in actual testing.
Ferrum MasterTry to cool that with sustained gaming loads using conventional methods using a mere middle end cooling solution, not a complete enthusiast overkill. 200W is too much...
At the top end, i.e. i7 and i9, I feel one will need a high end air cooler or a 240mm AIO. While the CPU can run hot depending on what application you run, but I was surprised that my i7 10700K at stock does not run very hot under gaming load. Games don't usually stress CPUs that much as some apps/ benchmarks to make it hit 200+W. Using HWInfo, I've rarely seen the CPU pulling more than 120W, with temps at around 69 to 74 degree celsius in a room with an ambient temp of 27 degrees and with a 240mm AIO water cooler. This was with boost unlocked, i.e. it does not throttle back to PL1. Was quite a positive surprise coming from a Ryzen 7 5800X user. But do note, I generally don't bother about running benchmarks and such, so the above is purely from the perspective of gaming load. I know this CPU is capable of pulling more power and outputting more heat under all core CPU intensive load.
Posted on Reply
#42
efikkan
Ferrum MasterTry to cool that with sustained gaming loads using conventional methods using a mere middle end cooling solution, not a complete enthusiast overkill. 200W is too much...
Middle end? (It's mid-range ;) )
While I do dislike Intel's PL2, and it does unfortunately confuse a lot of people including you, having a burst power draw of 225W (?) for Alder Lake for a few seconds is not an issue for any half-decent cooler, even the likes of Hyper 212. This burst power draw is only an issue for PSUs, where you have to account for burst power draw.
owen10578Unless intel did some kind of architectural ass pull magic on 12th gen to gain that much IPC then I don't see this score as real unless they used LN2 or something. 8 big cores + 8 small cores beating full fat 16 Ryzen cores seems like a stretch. But I really hope this is real because AMD really needs some competition to stop them turning into the new Intel.
I don't know where you got the IPC figures, but please don't try to extrapolate that from Cinebench.

But Golden Cove is a major architectural overhaul, so there is certainly the possiblity of some gains.
Posted on Reply
#43
watzupken
Shatun_BearWhat, 200W is way too much? This is not a 32-core.
I won't be surprise it will pull more than 200W. Consider that the Rocket Lake i7 and i9 is capable of pulling close to 300W in Anandtech's review, and that is with just 8 cores. While you can argue that it is due to 14nm, but considering that Intel is pushing both 14nm and 10nm to their limit, I won't be surprise to see crazy power requirement. In fact, it is also possible that AMD may increase their CPU's TDP going forward since their main competitor is pushing power envelop to their advantage.
efikkanMiddle end? (It's mid-range ;) )
While I do dislike Intel's PL2, and it does unfortunately confuse a lot of people including you, having a burst power draw of 225W (?) for Alder Lake for a few seconds is not an issue for any half-decent cooler, even the likes of Hyper 212. This burst power draw is only an issue for PSUs, where you have to account for burst power draw.
Unless Intel strictly stick to the boost timing, it is possible that we see that mobo makers may allow unlimited boost like we see on Comet and Rocket Lake. The problem with sticking to a mid/ low end cooler is that while you are right that it can boost for a bit and then throttle down to maintain temps, but it will also negatively impact performance if the CPU cannot sustain the boost clock. In games, the issue is probably less of a problem from my observation, but it will likely impact performance for some CPU intensive load like actual video encoding/ decoding tasks.
Posted on Reply
#44
efikkan
watzupkenUnless Intel strictly stick to the boost timing, it is possible that we see that mobo makers may allow unlimited boost like we see on Comet and Rocket Lake. The problem with sticking to a mid/ low end cooler is that while you are right that it can boost for a bit and then throttle down to maintain temps, but it will also negatively impact performance if the CPU cannot sustain the boost clock. In games, the issue is probably less of a problem from my observation, but it will likely impact performance for some CPU intensive load like actual video encoding/ decoding tasks.
Intel have made a spec when it comes to power draw. At stock the CPU will throttle after a timeout (up to 56 sec). If you disable this power limit, then you are running the CPU overclocked and will void the warranty and reduce the product lifespan.
Running the CPU stock is no problem for cooling, and is what "all" of you should do (unless you're intentionally overclocking). Unfortunately not all reviewers have enough knowledge to run CPUs completely at stock when comparing.
Posted on Reply
#45
Unregistered
I want a high end 5ghz CPU that's only drawing 20watts........oh wait we're gonna have to wait 20 years for that. /s

High end CPU's will have a high power draw, what do you people want?
#46
rainxh11
those gracemont cores are as fast as skylake cores, in fact they are streamlined skylake cores with hyperthreading removed and smaller decoder, allowing it to nearly be the size of 1/4 th of the golden cove cores
that's impressive having small cores with single core performance of a 10900k is nothing short of amazing
Posted on Reply
#47
Intervention
R-T-BAlder Lake isn't the same node. We finally get to see 10nm.
I know it's 10nm. I mean as far as motherboards go, intel will most likely require a new socket when the next 10nm+ revision is released as it has been ever since I can remember.
Posted on Reply
#48
yeeeeman
Ferrum MasterThe key question. At what power? I suspect 300W+ power hog.

I can get much more out of my 5950 when removing power limits. And I can as I run custom loop.

Do I need it? Naah.
228W
Posted on Reply
#49
R-T-B
InterventionI know it's 10nm. I mean as far as motherboards go, intel will most likely require a new socket when the next 10nm+ revision is released as it has been ever since I can remember.
Oh, if that's what you meant I totally agree.
Posted on Reply
#50
yeeeeman
rainxh11those gracemont cores are as fast as skylake cores, in fact they are streamlined skylake cores with hyperthreading removed and smaller decoder, allowing it to nearly be the size of 1/4 th of the golden cove cores
that's impressive having small cores with single core performance of a 10900k is nothing short of amazing
Yeah, maybe also shorten a bit the pipeline since you won't clock these at 5Ghz and you get even more performance thanks to lower penalty on misprediction.
It all adds up and makes and a lot of sense to use smaller cores also, that is, IF you can manage running all of them (big/small) in tandem with good efficiency.
Posted on Reply
Add your own comment
Dec 18th, 2024 07:54 EST change timezone

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts