Cinebench 2026

igor_kavinski said:

Custom cooled 285K is able to barely get over 11000 too.

Sucks how thermally limited halo chips are :(
Click to expand...

I get over 10000 with nothing other than a 360mm AIO, I also don't find it hard to cool at all.

thumbnail_20251231_225812.jpg
Didn't run with my 5090 because I sold it a few weeks ago for more than I paid due to prices going stupid here in Canada for almost everything again.

2P 5.7 E-Cores 4.9ghz 7200C34 Ram, Asus Strix Z890MB, basically minimal tweaking,
igor_kavinski said:
www.maxon.net

Cinebench Downloads

Maxon Cinebench is an industry-standard benchmarking software based on the cutting-edge technology.
www.maxon.net www.maxon.net

Anyone want to volunteer?

I don't want to risk burning out my 9950X3D on the ASROCK mobo. The 245KF score will be underwhelming so not worth trying for me.

EDIT: Much faster download mirror: https://www.techspot.com/downloads/7820-cinebench-r26.html

In case of "broken files" error, kill the Cinebench process and download/import certificate from here: https://support.maxon.net/hc/en-us/...when-I-launch-or-install-Cinema-4D-on-Windows
Click to expand...

Imagine having to worry about a chip burning up and using that as an excuse to not run a benchmark, all while throwing shade at intel.....
 
Tyler-98-W68 said:
Imagine having to worry about a chip burning up and using that as an excuse to not run a benchmark, all while throwing shade at intel.....
Click to expand...
Wasn't doing that. I said "halo" chips. 9950X3D isn't easy to cool either.

And it's the mobo's fault, not the chip that I have to worry about the possibility of burning up. Thanks to ASROCK.

The comment "Custom cooled 285K is able to barely get over 11000 too." was meant to convey that Intel has lost the optimization advantage it had with CB 2024.
 
285K R26 sgl .png
Reran the test, I guess i'm only 600 points off the guy with the "extreme cooling" 285K is, just a simple 360mm AIO, does that qualify as "custom"
 
Good score though my MT scaling of 11x (out of 14 cores) is fantastic compared to your 17.75x :P

I'm joking! I'm joking!

That Centaur CPU is the boss of MT scaling. 7.25x out of 8 cores!
 
That's a trashy arrow lake cpu that no one likes......its no different than a 285k with reduced power limits. Ive tested my 285k at a bunch of different PL1 and PL2 settings and how it affects gaming performance. The official intel TDP limits state up to 160w of turbo and a base of 55w.

It's up to laptop oems to choose the final Power limits. Based on the score that laptop likely has pretty high sustained PL1 limits to score like that
 
LazyGamer said:
a desktop version would be even higher with higher power limits.
Click to expand...
Not that I've seen. It needs serious cooling to get to 10K or higher. 275HX is more impressive in that it is getting that score with the limited cooling that can be built into a laptop chassis.

My guess is that the good bins are going into laptops while the leakier ones are ending up as 285K. Once the Aliexpress 275HX frankenmobos start appearing, maybe someone will show how good a bin it really is.
 
Last edited:
What is your definition of serious cooling?

So I'm getting 10419 points with nothing more than a 360mm AIO (Nautilus 360) is that serious?

My system isn't even fully tweaked either, I don't quite understand the logic that arrowlake is hard to cool and requires exotic cooling, it's so far ahead of raptor lake in power efficiency and ease of cooling.
 
Tyler-98-W68 said:
So I'm getting 10419 points with nothing more than a 360mm AIO (Nautilus 360) is that serious?
Click to expand...
You have a good sample. Other people are not so lucky with their 285K CPUs.

From Computerbase:

1768082894392.png

I'm assuming that they used an air cooler for both.

I'm not sure if that same sample will gain 600 points with a 360mm AIO.

This guy is also using a 360mm AIO:

1768083084396.png

Tyler-98-W68 said:
What is your definition of serious cooling?
Click to expand...
Custom sub ambient cooling.
 
Last edited:
What happens if you make a 285K a 12 Core CPU 8P and 4E Cores (E-Cores get more Cache)

Not a bad result at all
285K R26 multi 8P 4E 50 ecore .png

@igor_kavinski you were saying intel lost it's optimizations for Cinebench for R26, so what are comparable lower core count AMD chips getting?

I've also heard the argument that E-Cores are just fake.

So in theory if I have a 12 core CPU......an AMD 6 core (with SMT) 12 total threads would be a good comparison?
cinebench cpu monkey .png
 
Tyler-98-W68 said:
What happens if you make a 285K a 12 Core CPU 8P and 4E Cores (E-Cores get more Cache)

Not a bad result at all
View attachment 4251
Click to expand...
That's a crazy GOOD result for your sample. By comparison, I need to run my 245KF's 14 cores at P53E49 in Intel XTU to manage 6011 MT score.

Seems your two extra Lion Coves are doing almost the same work as my 2nd E-core quad cluster.

Tyler-98-W68 said:
So in theory if I have a 12 core CPU......an AMD 6 core (with SMT) 12 total threads would be a good comparison?
Click to expand...
That would be an absurd comparison because no matter how much SMT improves MT performance, it cannot replace real cores.

SMT is supposed to be a "cherry on top" sort of thing for a physical core, not a substitute for it.

An Intel CPU with 12 physical cores needs to be compared with 12 AMD physical cores. Yes, the AMD one would have double the threads due to SMT but it's Intel's architect(s) who decided SMT is worthless for consumers (probably covering up for the fact that their implementation is flawed and not as robust) and went ahead and disabled it for Lion Cove and for the Mont cores.
 
Tyler-98-W68 said:
@igor_kavinski you were saying intel lost it's optimizations for Cinebench for R26, so what are comparable lower core count AMD chips getting?
Click to expand...
I was referring to this: https://www.pugetsystems.com/labs/articles/intel-core-ultra-200s-content-creation-review/

1768107829850.png

In CB 2024, 24 Intel physical cores were able to trounce 16 AMD cores but in CB 2026, Intel's 24 physical cores are almost performing the same as 16 AMD cores. It's almost like Maxon decided that it wouldn't look "kosher" for them to keep supporting Intel when AMD looks to be doing good in the market so they went back to the drawing board and tweaked the code to prefer threads over physical cores. Either that or AVX-512 is playing a bigger part this time around (would be confirmed once Chips and Cheese review the new benchmark).
 
You must log in or register to reply here.
Become a Patron!
Back
Top