Thursday, November 18th 2021
Intel Core i7-12700H Beats Ryzen 9 5900HX by 47% In Leaked Cinebench Scores
Intel is expected to announce their Alder Lake mobile processors in Q1 2022 with rumors and leaks for several of these chips already surfacing. The i7-12700H has recently been spotted running Cinebench on what could possibly be an MSI GE76 Raider 12UH with the results showing impressive performance gains. The processor features six Golden Cove (P) cores and eight Gracemont (E) cores for a total of 14 cores and 20 threads.
The i7-12700H includes a configurable base TDP of 35 W - 45 W and this specific sample was running at a reported base frequency of 2.9 GHz during testing. The processor scored 689 points in Cinebench R20 single-core which places it 12% faster than the Core i9-11950H and 21% faster than AMD's flagship Ryzen 9 5900HX. The processor widens this gap in Cinebench R20 multi-core with a score of 7158 points placing it 47% above the 8-core, 16-thread Ryzen 9 5900HX. We can also see that multi-core performance is 49% faster than the recently released Apple M1 Max in Cinebench R23. AMD is preparing their Ryzen 6000 processors for an early 2022 launch which will be competing with these Alder Lake chips but we have yet to see many performance leaks for them to compare with.
Source:
Notebook Check
The i7-12700H includes a configurable base TDP of 35 W - 45 W and this specific sample was running at a reported base frequency of 2.9 GHz during testing. The processor scored 689 points in Cinebench R20 single-core which places it 12% faster than the Core i9-11950H and 21% faster than AMD's flagship Ryzen 9 5900HX. The processor widens this gap in Cinebench R20 multi-core with a score of 7158 points placing it 47% above the 8-core, 16-thread Ryzen 9 5900HX. We can also see that multi-core performance is 49% faster than the recently released Apple M1 Max in Cinebench R23. AMD is preparing their Ryzen 6000 processors for an early 2022 launch which will be competing with these Alder Lake chips but we have yet to see many performance leaks for them to compare with.


70 Comments on Intel Core i7-12700H Beats Ryzen 9 5900HX by 47% In Leaked Cinebench Scores
A 15W CPU can turbo up to 50W.
Each GHz results in 105p for Skylake/CometLake, 123pt for Rocket Lake, 129p for Zen3 (maybe slightly less for Cezanne but the difference is very minor), 150p for Alder Lake P-cores (and 105p for E-cores).
Zen3 SMT efficiency is 1.27x, Intel's HT is a couple percent lower at around 1.25x.
CB R20 ST at 689 points is that core running at 4.5 or 4.6GHz, Alder lake at that clock should be ~20W on one core.
For comparison, Zen3 core at same clock consumes 15-17W. Although to be fair, this is really a different comparison because Zen3 core needs ~5.3GHz for the same 689 points result (and goes over 20W already at 5GHz).
Multicore is difficult to analyze because 12700H is apparently a 6P+8E configuration which we have not really seen thoroughly tested. 12900K is 8P+8E, 12700K is 8P+4E and 12600k is 6P+4E.
Rough numbers though:
- If 12700H is running the base 2.8GHz on E-cores which would leave 4.2GHz for P-cores to get that score. 4.2GHz P-core uses 15-16W, so 90-96W plus whatever E-cores consume. So you may be right about 125W :)
- Alternatively if 12700H is running boost 3,7GHz on E-cores this would need only 3.6GHz from P-cores to get the score. 3.6GHz P-core uses 10W, so 60W plus around 30W for 8 E-cores at these clocks. around 90W.
Keep in mind though that power consumption comes down a lot faster than performance when clocks are decreased.Looking closer at the results - this seems to be a 5900HX with stock power limit applied. Looking at a couple other reviews this means 45W TDP and power limit in 50-60W range.
TDP of "35 W - 45 W" with this performance ? Not in a million years, this thing probably boosts over 100W during the entire benchmark, we've seen the atrocious power efficiency these chips have on desktop, the cat is out of the bag. I mean sure, power and clocks don't scale linearly but there is no way there can be such a chasm between these operating frequencies in terms of efficiency.
He will be abused with posts like undervolting doesn't work, my crap laptop is burning 100C... something doesn't work, throttle this, throttle that.
xD Laptop gamers are the most hilarious bunch. Like wanting to do drag race with a Dacia.
After all that I can play comfortably on a silent laptop with great performance, pretty sure I extended the life out of the laptop too.
But it's true, because the lack of a standard design it's hard to do an apples-to-apples when it comes to mobile CPUs.
But i am not that much impress that a design that come from mobile might be good in laptop. They just screwed up on PC as they pushed way too much the power to get few % of CPU to ensure they win as hard as possible against AMD. But on many test, when the CPU isn't boosting like crazy or at medium load, the power consumption is great. So in Laptop, it's probably going to look very good.
I don't think AMD being good on Desktop CPU and HEDT made intel scared that much. But loosing in laptop and server were something Intel couldn't allow. It's where their money is and why ADL is Big/Little in the first place.
There was nothing wrong with my laptop at stock, I wanted to make it better, and I did :D.
I prefer laptops that I can open up, If nothing else, being able to clean the dust every now and then makes then run cooler and more silent.
Got one AMD laptop (Ideapad older) and it's good :) Bought 2 months ago IdeaPad Flex 5 14ALC05 14" FHD. Cool thing :)
I'm thinking of using it as a Win11 test machine, if I get over the MS account requirement.
Sure you can open anything up but I prefer to get my money's worth out of the box.
Oh boy Intel mobile CPU sure runs hot :roll: