Alder Lake-S leaks

[Destination]

Because it has 16 cores, not 8. I dont' know why you thought it only has 8

No, I get that, but only some do HT, and the others are little piles of poo, its more the progression.
810*16 is 12960, suggesting the weak cores can manage almost the equivalent of the output of the main cores.
If this is the case I'll be incredibly surprised, unless intel have done something funny to optimise purely for cinebench.
I'd like to see an r23 comparison.

 

[alec]

benchmark makes no sense at all, even if the single is as powerful as it is, how on earth does it stack up that only 8 cores can pull that amount of multi

little cores are helping too
plus, CB run is so short, spends a lot of time in 228W PL2 mode

 

[Destination]

ah true, the r23 might looks a little different, esp on single core

 

[Grim5]

No, I get that, but only some do HT, and the others are little piles of poo, its more the progression.
810*16 is 12960, suggesting the weak cores can manage almost the equivalent of the output of the main cores.
If this is the case I'll be incredibly surprised, unless intel have done something funny to optimise purely for cinebench.
I'd like to see an r23 comparison.

We know the 8 little cores are not as fast as the big 8 cores. Leakers say the little cores are on the same level as a 8 core 6700k would be. So take a look at the 6700k's R20 multi, double it and that's how much the little cores add to the 12900k's score with the rest coming from the big cores.

Now off the top of my head I think the 6700k gets about 2300 so the little 8 cores on the 12900k contribute 4600 points with the remaining 7000 coming from the 8 big cores and for reference the 5800x gets 6000 points

 

[Grim5]

little cores are helping too
plus, CB run is so short, spends a lot of time in 228W PL2 mode

Yep with 16 cores the R20 test completes very fast - probably short enough to fit in the PL2 window. The 12900k benchmark above was tested with a 360mm AIO cooler

 

[muon]

No matter how you spin it, that single core score is stupidly fast.

 

[jigger]

No matter how you spin it, that single core score is stupidly fast.

It’s just a shame it’s attached to 15 others that are garbage

 

[Joxeon]

I hope it's true as it will force AMD to increase performance by more % than prices next time around.

 

[SmokSmog]

It’s just a shame it’s attached to 15 others that are garbage

How ? all 8 big cores are super fast, way faster than Zen3 cores. Someone calculated 8250 points in cinebench r20 for big cores at 5ghz. 5800X scores 6000 at stock.

 

[Perfect_Chaos]

I hope it's true as it will force AMD to increase performance by more % than prices next time around.

Yer, but I'd only jump in at Zen 4, or Meteor Lake. Earlier might be tempting with Intel but not if it means slower more expensive DDR5, and I'd like to be able to upgrade the CPU at least once if it means an extra 15-20% boost per gen now.

Be interesting to see if this single thread performance uplift is accurate.

 

[LePhuronn]

Be interesting to see if this single thread performance uplift is accurate.

I hope so, otherwise Intel will continue to look silly when Zen 3D shows up early next year.

 

[SKILL]

No, I get that, but only some do HT, and the others are little piles of poo, its more the progression.
810*16 is 12960, suggesting the weak cores can manage almost the equivalent of the output of the main cores.
If this is the case I'll be incredibly surprised, unless intel have done something funny to optimise purely for cinebench.
I'd like to see an r23 comparison.

Would it not be 810*24, as in 8 big cores, 8 big 'cores' HT and 8 small cores. Obviously neither small cores nor HT 'cores' are going to be as fast as the single thread performance, plus add in the thermal limits etc.

How ? all 8 big cores are super fast, way faster than Zen3 cores. Someone calculated 8250 points in cinebench r20 for big cores at 5ghz. 5800X scores 6000 at stock.

lolwut, so a single big core at 5.3GHz scores 810 so 8 of them at a slower speed score 10.18x as much?

(810 / 5.3) * 5 * 8 = 6113 But even that is likely incorrect as it doesn't account for HT (which the 5800X score presumably does) or anything else.

 

[tommoT.Striker]

Alder Lake Looks to be a promising release finally from Intel.
If the leaks are true then ST is superb, and MT is great. But we'll have to see how it behaves across different workloads (Rocket Lake was great on Cinebench too).

I'm sure AMD will have a strong response.

 

[jigger]

How ?

Because the people running Intel don’t know what they’re doing.

 

[SmokSmog]

Would it not be 810*24, as in 8 big cores, 8 big 'cores' HT and 8 small cores. Obviously neither small cores nor HT 'cores' are going to be as fast as the single thread performance, plus add in the thermal limits etc.



lolwut, so a single big core at 5.3GHz scores 810 so 8 of them at a slower speed score 10.18x as much?

(810 / 5.3) * 5 * 8 = 6113 But even that is likely incorrect as it doesn't account for HT (which the 5800X score presumably does) or anything else.

I was including Hyper threading 8/16T around 8250 for 12900K at 5ghz all core and 6100 for 5800X stock

 

[SKILL]

I was including Hyper threading 8/16T around 8250 for 12900K at 5ghz all core and 6100 for 5800X stock

Ah, my bad... If anything that might be an underestimate then? Surely HT gives more than a ~33% improvement?

 

[alec]

Ah, my bad... If anything that might be an underestimate then? Surely HT gives more than a ~33% improvement?

Nah, in real computation heavy tasks like Cinebench, it has been 20-25% for Intel HT. Historically slightly more for AMD SMT.

 

[Destination]

I was including Hyper threading 8/16T around 8250 for 12900K at 5ghz all core and 6100 for 5800X stock

This is interesting, and I ponder if correct, as given the alleged released bench of over 11000, one wonders if the main cores are giving the 8k, and the weak addon cores managing 3000, interesting, wonder will it equate to anything useful in real world, as the 11th gen certainly didn't.

 

[Perfect_Chaos]

https://www.theverge.com/2021/7/22/...gs-chips-semiconductor-shortage-manufacturing
Intel webcast tomorrow.

IMO Ocean Cove (Meteor Lake) looks to be a big change over everything else, guess we'll have to see how it pans out, but that's what i have my eye on.

Too bad Intel won't allow more upgrades on their boards.

 

[Grim5]

This is interesting, and I ponder if correct, as given the alleged released bench of over 11000, one wonders if the main cores are giving the 8k, and the weak addon cores managing 3000, interesting, wonder will it equate to anything useful in real world, as the 11th gen certainly didn't.

Single thread numbers translate pretty well into games though and 810 ST is massive