News Intel Core Ultra 9 285H 45W mobile CPU falls short of its Lunar Lake brethren in leaked Geekbench 6 single-core benchmark

Status
Not open for further replies.
Another core configuration?
I wonder if the memory controller is on the same die as the cores with this one?
We don't even know what node it is because Intel can just switch nodes and fab companies with their designs now.

Playing with these configurations like they are Legos.
 
what? That's the Ultra 5 245K's config except they're using the MTL SoC tile and either the MTL-H or -U GPU tile
Another core configuration?
I wonder if the memory controller is on the same die as the cores with this one?
We don't even know what node it is because Intel can just switch nodes and fab companies with their designs now.

Playing with these configurations like they are Legos.
 
My GeekBench 6 score for my 258V 17W Lunar Lake are 2806 / 11355 (higher than the referenced 288V 30W processor), meaning that my 17W LL outperforms this 45W H-series processor by more than 5% single core. Admittedly my MultiCore score falls short 25%, but that is with half the number of cores available (4P + 4E vs 6/10). I think having the 32GB memory inside the CPU package is genius (I'm not an Apple fan, but they got it right too). And Pat is saying that CPU packaged RAM of LL is a one-off. I think from both a performance and energy efficiency standpoint it is genius. I bought a LL based laptop over a Strix Point one exactly for these reasons. So if Intel wants to keep there market share up, they should really have a good look at what works or not.
 
Dell jumped on the "Pro Max" bandwagon as well?? Come on folks, stop rubbing Apple's ego.

Big tech marketing: we only need one suffix adjective. There's plenty out there and smashing similar ones together is just silly. For example in this case, "Max" is greater than "Pro," so the latter is superfluous.

Yes, sort of ranting so back on topic: Pat @ Intel did say that Lunar Lake was a one-off; I have to think that stripping back the on-package LPDDR is measurably going to hurt performance, even if power efficiency is possibly the larger advantage of that feature... depends on workload too I suppose.
 
Last edited:
what do you mean by cut down?
I think the point is that the ARL desktop parts all use the same die and in the past the mobile parts have had a separate die rather than a cut down desktop one (except the HX which are literally the desktop chips silicon wise). All of MTL mobile used the same die, and if they were to do the same for ARL mobile using the desktop compute tile would mean wasted silicon because the highest laptop configuration should be 6+8 and if there's an HX it should be using the same die as desktop.
 
  • Like
Reactions: rluker5
Wow another garbage Dell with poor thermal performance and severe throttling. People have been slamming Dell for years over this sort of crap. Also, I would never ever buy a 24 core laptop apu unless it had a true work station chassis.
 
My GeekBench 6 score for my 258V 17W Lunar Lake are 2806 / 11355 (higher than the referenced 288V 30W processor), meaning that my 17W LL outperforms this 45W H-series processor by more than 5% single core. Admittedly my MultiCore score falls short 25%, but that is with half the number of cores available (4P + 4E vs 6/10). I think having the 32GB memory inside the CPU package is genius (I'm not an Apple fan, but they got it right too). And Pat is saying that CPU packaged RAM of LL is a one-off. I think from both a performance and energy efficiency standpoint it is genius. I bought a LL based laptop over a Strix Point one exactly for these reasons. So if Intel wants to keep there market share up, they should really have a good look at what works or not.
This score is just bogus. The on package memory only improves efficiency and gives a negligible improvement to latency.
 
  • Like
Reactions: User of Computers
Status
Not open for further replies.