With the release of Alder Lake less than a week away and the "Lakes" thread having turned into a nightmare to navigate I thought it might be a good time to start a discussion thread solely for Alder Lake.
From what I see, the p-core is great, but takes too much power, so a 16 p-core was unfeasable, so they added the e-cores to help with MT performance but keep power manageable. It still takes more than the 5950x, even my lowly 12700F, but keeps up with 1/2 the cores, at half the price, but at more power than all 16 cores of the 5950x. Its a mixed bag type arrangement.Until Intel shows a working E-core only CPU, I have my doubts. Maybe the E-core cluster is a relatively quick hackjob done on the advice of Jim Keller, when it became apparent to them that they would have issues with MT throughput in future and so the CPU cluster cannot communicate to the outside world unless the communication is initiated by a P-core?
I agree that Alder Lake definitely is a mixed bag. You can either get great performance or great efficiency. Unfortunately for Intel, the 5950X can get both great performance and great efficiency.Its a mixed bag type arrangement.
Effective Speed is a gaming focused metric.The effective speed is the average of all benchmarks, not just one. Average speed is yes totally made up, as all benchmarks averages are. But averages are still quite relevant. The average of all benchmarks they use has the G7400 at 73.1% vs the G6400 at 70.0%. That is an unnoticeable difference for most people.
Effective Speed - A measure of CPU speed geared towards typical users. Intel i9-9900K ≈ 100%.
Gaming is by far the most demanding CPU activity most users undertake. CPU Effective Speed (average bench) is calibrated to estimate differences in EFps between PCs.
EFps - A measure of PC gaming performance that includes frame drops.
Effective Fps (EFps) measure gaming experience using both average Fps and frame drops.
From what I see, the p-core is great, but takes too much power, so a 16 p-core was unfeasable, so they added the e-cores to help with MT performance but keep power manageable. It still takes more than the 5950x, even my lowly 12700F, but keeps up with 1/2 the cores, at half the price, but at more power than all 16 cores of the 5950x. Its a mixed bag type arrangement.
Yes, games are part of it. But only a part of their scores. They break out the scores into three categories (all on the original link I provided). If you solely want workstation performance you can look at just the workstation score. Although, I would seriously question your interest in the use of a Pentium chip for that purpose.Effective Speed is a gaming focused metric.
How PC scores are calculated
Gaming
Average fps, and more importantly, 0.1% and 1% lows.
Desktop
Web browsing, office apps, music/video playback.
Workstation
Audio/video encoding, number crunching, virtual machines, databases:
They should put with "Alder Lake -N" since e-cores are not slow and can be perfect for "testing" purpouses on the computer technicians./me wonders about what they do with ADL dies that have zero functional P-cores, but do have some working e-cores. Are those salvagable at all, if the current crop of ADL BIOSes (and I do believe that it is strictly a BIOS limitation) requiring a P-core to be active to boot. (ADL with P-cores disabled, only running on e-cores, wouldn't that be useful for a file-server / NAS role, where the loads aren't "bursty" like PC software?)
From what I see, the p-core is great, but takes too much power, so a 16 p-core was unfeasable, so they added the e-cores to help with MT performance but keep power manageable. It still takes more than the 5950x, even my lowly 12700F, but keeps up with 1/2 the cores, at half the price, but at more power than all 16 cores of the 5950x. Its a mixed bag type arrangement.
The requirement for at least one P-core is, at minimum, baked into the CPU's firmware. BIOS hacks alone would not be sufficient to get that working. However, it's extremely unlikely that there's any hardware dependency. ADL-N will need it to work, after all.Until Intel shows a working E-core only CPU, I have my doubts. Maybe the E-core cluster is a relatively quick hackjob done on the advice of Jim Keller, when it became apparent to them that they would have issues with MT throughput in future and so the CPU cluster cannot communicate to the outside world unless the communication is initiated by a P-core?
From what I see, the p-core is great, but takes too much power, so a 16 p-core was unfeasable
I thought the 12700F was like a 65 watt part or at least not like a 12900k. But mine takes 270 watts from the wall using 16 cores (not sure which ones exactly, but win 11 knows) Stock bios. I could not find pl1 or pl2 or anything like that in bios. (ASUS) So I have no idea what its set to really. But my 5950x takes less for all 16/32 cores to work, and the end result is about the same. (work done)If you look at how well 8 Golden Cove cores can perform when power limited, you may see that that's not necessarily the case. Especially with a power budget of 241W at the top end. A 16c Golden Cove with an all-core turbo of ~4 GHz would not necessarily use that much power (except on AVX-512 workloads, though they could have let clocks slip another 500 MHz and still gotten good results). It's more a question of area efficiency. Such acoredie would have been larger, and with Intel seemingly unwilling or unable to use EMIB on their 10ESF consumer products, going with chiplets ala Sapphire Rapids was also not an option.
Alder Lake as it exists is probably the best they could do given the available die space.
How's a 12700f a 16 core cpu?I thought the 12700F was like a 65 watt part or at least not like a 12900k. But mine takes 270 watts from the wall using 16 cores (not sure which ones exactly, but win 11 knows) Stock bios. I could not find pl1 or pl2 or anything like that in bios. (ASUS) So I have no idea what its set to really. But my 5950x takes less for all 16/32 cores to work, and the end result is about the same. (work done)
The software I am running is picking 16 of the 20 threads. I want to be precise when talking about my power usage.How's a 12700f a 16 core cpu?
I thought the 12700F was like a 65 watt part or at least not like a 12900k. But mine takes 270 watts from the wall using 16 cores
The software I am running is picking 16 of the 20 threads. I want to be precise when talking about my power usage.
Its dual-boot. I am using the windows 11 partition as its fully supported.How to specify a CPU core in Linux with Taskset
Sometimes you need to manually assign CPU cores to programs to get best performance. In Linux, the main tool for this purpose is taskset.tipsmake.com
I think in your case you would have to type:
taskset -c 0-15 [executable_name]
This way I think you can be sure it will run only on P-cores.
From what I see, the p-core is great, but takes too much power, so a 16 p-core was unfeasable, so they added the e-cores to help with MT performance but keep power manageable.
AMD would still have an advantage due to the economics
Has anyone emulated 12900KS performance level by overclocking 12900K to similar clocks yet?
The problem that Alder Lake has is that 4-E cores is possibly the worst possible combination. That number is just too few to really help with many use cases. I think once we get 16 or more E cores that their benefits will be seen. Until then, I'll be avoiding the 4-E core products like the plague.
I've only seen leaked Geekbench scores of a 12900KS and yes it does show just 5% improvement in ST and 3.5% gain in MT score. My point in asking was to see if anyone here turned their 12900K into a KS and using it that way regularly.Surely there are reviews out there with a 12900K overclocked to 5.2GHz all core?
Except if there is something wrong with the Intel Thread Director that can't be fixed with a microcode update. Raptor Lake users will enjoy the improved and refined Intel Thread Director and Alder Lake users with E-cores will get screwed forever when it comes to improper scheduling of processes on E-cores. Not like Intel will give them a refund or anything.Additional cores and most importantly threads have proven to help a lot long term.
Is there any evidence that such a thing exists? It's mostly on the OS anyways.Except if there is something wrong with the Intel Thread Director that can't be fixed with a microcode update. Raptor Lake users will enjoy the improved and refined Intel Thread Director and Alder Lake users with E-cores will get screwed forever when it comes to improper scheduling of processes on E-cores. Not like Intel will give them a refund or anything.