Remember 9% for Lion Cove is combined Int/FP, while Skymont is 32%/72% in Int and FP respectively.
Assuming 65/35% split like in Geekbench, that'll result in an average of 46% gains.
Now tell me in what world would a 32% Int, 72% FP gain even against the 12th Gen Gracemont would result in Lion Cove being 20-30% faster per clock over Skymont?
If you take Raptormont figure, that's 7.2% and 3.5% lead in Int/FP. If it's over Gracemont it's 1% and 3% behind. So with a meagre 9% gain, Lion Cove is as low as 2.9% faster over Skymont.
Lunarlake presentations said Skymont on ring will outperform Raptor Cove by 2% in both SpecInt and FP.
Skymont straight up doubles the number of FP units, so unlike the new instructions which take eons for software to adopt, it'll benefit all the applications dating back to the Pentium 4 era immediately.
In many applications, Skymont has straight up better FP throughput as it has twice the amount of FP units compared to Raptor Cove.
I certainly prefer this approach over the P core teams' almost Elitist approach of shutting down competition by forcing them to adopt new ISA, and for the entire world to cater and recompile for them.
I have gone over the Intel provided CB R24 data in some detail and here is what I have found. Please let me know where I've lost the trail.
Based on my testing of my 14900K in CB R24 MT, Raptor Cove does about 21.2 points/GHz (no HT), 27.8 points/GHz (with HT) and Gracemont about 12.8 in MT.
Using these numbers, a 14900K at stock 5.6/4.4 speeds from Intel Ark would score 2147 points.
Intel claims Arrow Lake is 18% faster or would score about 2525. Stock Arrow Lake speeds are 5.5/4.6 as viewed in Intel Ark.
Since the P cores are similar architectures we'll start with the +9% for Lion Cove over Raptor Cove, using the non HT Raptor 21.2 points/GHz and increase it to 23.1.
So, assuming +9% IPC ST increase for Lion Cove compared to Raptor Cove ST (Raptor Cove ST IPC derived from 8P only run without HT, not ST run, they are different, ST run is generally a bit higher than score obtained by taking MT score most likely due to multithreading overhead), then Skymont would need a 60% IPC increase in CB R23 vs Raptor Lake Gracemont or an increase from 12.8 points/GHz to 20.5 points/GHz, or in other words 3.5% less IPC than Raptor Cove.
Looking at ST scores, Intel claims 8% better for Lion Cove. Now we're looking at 6GHz for Raptor Cove and 5.7GHz for Lion Cove. This would require a 13.5% IPC increase for Lion Cove over Raptor Cove and the Lion Cove score would be about 146.
So, let's review.
According to Intel, looking at Cinebench R24 MT:
Lion Cove will have a 9% increase in MT score when you look at Raptor Cove without HT on vs Lion Cove.
Skymont will have a 60% increase in IPC over Gracemont or be 3.5% below Raptor Cove IPC.
For Cinebench R24 ST:
To achieve the Intel claimed 8% increase in score Lion Cove will need an IPC increase of 13.5% over Raptor Cove.
According to this data, Arrow Lake at 5.5/4.6 should score 146/2525 in Cinebench R24 ST, MT, respectively.
So if these numbers (Intel's claims) turn out to be true then Arrow Lake will be a good CB R24 performer mainly due to the astounding uptick in IPC of Skymont. I did not expect to see +60%. That's Conroe territory
One more thing I forgot to mention. In CB R24 if you compare the MT IPC performance of Lion Cove vs. Raptor Cove without HT, Lion Cove is only better by 13%. If you take the clockspeed deficit into account Lion Cove is better by 35%. Skymont looks to be formidable in FP.
It may be time for Intel to start considering a "workstation" version of Arrow Lake that is 6+24. That would be quite the MT powerhouse especially if they can creep up the clock on Skymont to ~5GHz without dramatically increasing the area.
One final, final thing. With RPL the P/E split for compute with CB 2024 MT was 58%/42%, now with ARL it looks to be 40%/60%. The P's and E's have reversed with Skymont carrying more of the compute load.