Alder Lake-S leaks

[4K8KW10]

Intel Core i9-12900K Alder Lake CPU Benchmark Leaks Out, Allegedly Faster Than AMD Ryzen Threadripper 2990WX 32 Core Chip In Cinebench (wccftech.com)
Intel Core i9-12900K Alder Lake CPU Benchmark Leaks Out, Allegedly Faster Than AMD Ryzen Threadripper 2990WX 32 Core Chip In Cinebench (wccftech.com)

 

[cliffy]

Intel Core i9-12900K Alder Lake CPU Benchmark Leaks Out, Allegedly Faster Than AMD Ryzen Threadripper 2990WX 32 Core Chip In Cinebench (wccftech.com)
Intel Core i9-12900K Alder Lake CPU Benchmark Leaks Out, Allegedly Faster Than AMD Ryzen Threadripper 2990WX 32 Core Chip In Cinebench (wccftech.com)

Admittedly that is fast considering 24 threads vs 32 but my 5950 scores over 30000 and i am running ram at 3200 cl16 and my CPU is clocked no where near 5.3Ghz. I reckon IPC is very similar, would love to see a clock for clock bench

 

[humbug]

Yeah that is a bit odd, 5950X normally scores between 30,000 and 31,000, overclocked as high as 32,500.

My 5800X is 16,300.

Its a good showing tho from the 12900K, but the 5950X is not what its up against.

 

[cliffy]

Yeah that is a bit odd, 5950X normally scores between 30,000 and 31,000, overclocked as high as 32,500.

My 5800X is 16,300.

Its a good showing tho from the 12900K, but the 5950X is not what its up against.

If memory serves min(5950X) was hitting about 4.6 all core to score 30000+ so that is 700Mhz slower per core plus my RAM being 2000Mhz slower although i obviously have an 8 thread advantage

 

[Drollic]

I want to see these ddr5 memory and board prices

 

[humbug]

If memory serves min(5950X) was hitting about 4.6 all core to score 30000+ so that is 700Mhz slower per core plus my RAM being 2000Mhz slower although i obviously have an 8 thread advantage

A lot of Alderlake performance is down to DDR5, not all of it but the largest chunk.

Zen 3D is what its up against but even that is just a short term stop gap, DDR5 Zen 4 probably middle of next year.

 

[Destination]

A lot of Alderlake performance is down to DDR5, not all of it but the largest chunk.

Zen 3D is what its up against but even that is just a short term stop gap, DDR5 Zen 4 probably middle of next year.

It'll be interesting to see if this is actually the case.
Generally memory speeds and timings didn't affect r20, I assume r23 was the same with little to no effect.
So what really matters is if these chips are superclocked or something funny with them, or this is expected results.

AMD 24 threads is the 5900, if this can match the 32 threaded 5950 in multi that will make things very interesting.
It should decimate it in gaming, but again, we'll have to wait and see if the results stack up, if if they somehow play to the benchmark.

 

[Cyber-Mav]

ddr5 latency is high so cant see how its gonna benefit gaming at this stage. once its up to speed it will be good then.

 

[Joxeon]

It'll be interesting to see if this is actually the case.
Generally memory speeds and timings didn't affect r20, I assume r23 was the same with little to no effect.
So what really matters is if these chips are superclocked or something funny with them, or this is expected results.

AMD 24 threads is the 5900, if this can match the 32 threaded 5950 in multi that will make things very interesting.
It should decimate it in gaming, but again, we'll have to wait and see if the results stack up, if if they somehow play to the benchmark.

Also raptorlake will be competing against Zen 4 and with the IPC + doubling of small cores then it should be atleast 30% faster than Alderlake in MT workloads.

 

[BananaDunka]

Well it’s 8x 5ghz cores and 8x 3.7ghz atom cores.

Would interesting to see a 2x8 3.7ghz low power option. That could tempt me to consider a non AM4 platform.

I'm not sure I fully understand, is this a good or bad thing? I can only see a mssive boost in raw performance over my current 6700K?!

 

[jigger]

I'm not sure I fully understand, is this a good or bad thing? I can only see a mssive boost in raw performance over my current 6700K?!

Oh yeah, it will beat the 6700K. The small 8 cores are i7 5775C levels of performance. It’s not been confirmed I have a feeling the small cores are a development of the Broadwell architecture, as an attempt to bring Intel power use inline.

I moved from 6700K to the 1800X and that chip hammered 6700K. The 12900K should be in another performance realm.

 

[jigger]

Have Intel talked about ECC support on Alderlake? That is one area where AMD’s strategy is ambiguous and support is patchy.

 

[KompuKare]

Have Intel talked about ECC support on Alderlake? That is one area where AMD’s strategy is ambiguous and support is patchy.

Haven't heard anything and while I'd always say "never say never" for decades Intel were the segmentation masters.

Blowing random fused became second nature for them, so If they start reversing that and allowing consumer processors to have ECC I will be very surprised. Maybe even conclude they have changed their ways or are worried about something.

 

[Hook]

Haven't heard anything and while I'd always say "never say never" for decades Intel were the segmentation masters.

Blowing random fused became second nature for them, so If they start reversing that and allowing consumer processors to have ECC I will be very surprised. Maybe even conclude they have changed their ways or are worried about something.

Consumer DDR5 won't have the "proper" enterprise EEC.
"What's the difference between DDR5 and DDR4 ECC?
DDR5's on-die ECC doesn't correct for DDR channel errors and enterprise will continue to use the typical side-band ECC alongside DDR5's additional on-die ECC. To further bolster the memory channel, DDR5 DRAMs have additional storage just for the ECC storage. On-die ECC is an advanced RAS feature that the DDR5 system can enable for higher speeds."

 

[KompuKare]

Consumer DDR5 won't have the "proper" enterprise EEC.
"What's the difference between DDR5 and DDR4 ECC?
DDR5's on-die ECC doesn't correct for DDR channel errors and enterprise will continue to use the typical side-band ECC alongside DDR5's additional on-die ECC. To further bolster the memory channel, DDR5 DRAMs have additional storage just for the ECC storage. On-die ECC is an advanced RAS feature that the DDR5 system can enable for higher speeds."

Yes, we were talking about proper real ECC.

The on-die stuff is really to improve error corrections quasi internally. Not the same but a bit like error recognition in NAND or HDD or optical.

Haven't read that much about it, but it sounds like whereas the NAND ones are for making sure that multi-level cells don't have errors, the DDR5 on-die is more in case a signal didn't make it or somehow because of "going to quick" a cell didn't store the correct thing. At least that is what your quote implies. So if it detects an issue it should almost slow down the memory clocks to prevent that. Doesn't GDDR already do something like that, or is the GPU's memory controller which adds wait states if it detects errors?

Anyway, Intel have mostly fused ECC off from most their consumer chip (but witness some i3 being able to run on those HP MicroServer Gen8 which require ECC) whereas AMD have - until now at least - not fused it off, but next to no motherboard manufacturers validate ECC and it is actually hard to force a proper ECC error to test it yourself.

 

[Grim5]

Here is the cpu hierarchy for Cinebench R23 single core score




And now we have the 12900k too

it gets 2050 points in single core

https://videocardz.com/newz/intel-c...g-4-spotted-2k-points-in-cinebench-r23-teased

 

[LePhuronn]

let's hope Alder Lake's single core synthetic actually translates into something usable, unlike Rocket Lake.

 

[Destination]

Indeed, if it truly has a 20% single core improvement over its previous gen, and basically all its rivals, that must translate dramatically.
I just wonder if something is amiss.

 

[humbug]

Indeed, if it truly has a 20% single core improvement over its previous gen, and basically all its rivals, that must translate dramatically.
I just wonder if something is amiss.

20% is pretty good.

TBH Intel have done well over these two last generations, Rocketlake has a 20% IPC gain over Cometlake, that's no small potatoes and another 20% now...

 

[LePhuronn]

I just wonder if something is amiss.

What, like Rocket Lake's IPC uplift never translated into real world usage? Everybody's showing synthetics for Alder Lake right now, but that doesn't mean anything as we've seen many times before.

It's almost October, where are the leaked "12900K is 20fps better in these games than 5800X" or "12900K completes Puget's Premiere Pro test 40 seconds faster than 5950X" slides?

And no, Ashes of the Singularity doesn't count.