Monday, July 1st 2024
Intel "Arrow Lake-S" to See a Rearrangement of P-cores and E-cores Along the Ringbus
Intel's first three generations of client processors implementing hybrid CPU cores, namely "Alder Lake," "Raptor Lake," and "Meteor Lake," have them arranged along a ringbus, sharing an L3 cache. This usually sees the larger P-cores to one region of the die, and the E-core clusters to the other region. From the perspective of the bidirectional ringbus, the ring-stops would follow the order: one half of the P-cores, one half of the E-core clusters, iGPU, the other half of E-cores, the other half of the P-cores, and the Uncore, as shown in the "Raptor Lake" die-shot, below. Intel plans to rearrange the P-cores and E-core clusters in "Arrow Lake-S."
With "Arrow Lake," Intel plans to disperse the E-core clusters between the P-cores. This would see a P-core followed by an E-core cluster, followed by two P-cores, and then another E-core cluster, then a lone P-core, and a repeat of this pattern. Kepler_L2 illustrated what "Raptor Lake" would have looked like, had Intel applied this arrangement on it. Dispersing the E-core clusters among the P-cores has two possible advantages. For one, the average latency between a P-core ring-stop and an E-core cluster ring-stop would reduce; and secondly, there will also be certain thermal advantages, particularly when gaming, as it reduces the concentration of heat in a region of the die.
Every P-core would be no more than one ring-stop away from an E-core cluster, which should benefit migration of threads between the two core types. Thread Director prefers E-cores, and when a workload overwhelms an E-core, it is graduated to a P-core. This E-core to P-core migration should see reduced latencies under the new arrangement.
Source:
Kepler_L2 (Twitter)
With "Arrow Lake," Intel plans to disperse the E-core clusters between the P-cores. This would see a P-core followed by an E-core cluster, followed by two P-cores, and then another E-core cluster, then a lone P-core, and a repeat of this pattern. Kepler_L2 illustrated what "Raptor Lake" would have looked like, had Intel applied this arrangement on it. Dispersing the E-core clusters among the P-cores has two possible advantages. For one, the average latency between a P-core ring-stop and an E-core cluster ring-stop would reduce; and secondly, there will also be certain thermal advantages, particularly when gaming, as it reduces the concentration of heat in a region of the die.
86 Comments on Intel "Arrow Lake-S" to See a Rearrangement of P-cores and E-cores Along the Ringbus
An Intel CPU with a supposedly 35W TDP can easily reach 50-60W under high load. This leads me to question the abnormally low numbers obtained here.
So an Intel cpu with a 35w PL2 limit will stick to 35 watts.
But 2 cpus released at the same time, with the same msrp and the same name (I mean amd deliberately chose the intel naming scheme to make direct comparisons straightforward) are as direct competitors as they could possibly ever be. What much more do you need to directly compare 2 cpus?
And the funny thing is I'm not in disagreement with you. Yes, exactly as you already mentioned, an 8core cpu has no chance in hell even under ln2 to match a 16 core cpu in MT performance. That's PRECISELY the problem, and precisely the reason intel wins in Mt efficiency. Because they simply offer more cores at each given msrp.
It's not unfair comparing these 2 cpus, what's unfair is amd releasing 8core chips and pretending they are i7 competitors and asking i7 prices. They are not. They aren't even i5 competitors. And they don't seem to have learned the lesson, if rumors are true the 9700X is going for 399 again. 399!!! A cpu that most probably will be losing to a freaking 2 generations old 13700k.... Yes, most likely it will be even be losing in efficiency. Is it unfair as well to compare a 2 year old i7 to a brand new R7 in ISO efficiency? I mean come on..Because it runs with unlimited (4096w) power limits. Have you actually tried one or are we going with "trust me bro"?
This is a tech forum man. I get you people hate intel but please, please, can we at least TRY to stick to facts? It's really not that hard. Stop spreading missinformation, please.
The only thing you're right about is that I don't want an intel CPU, not even for free.
This is some userbenchmark level arguments. It's just not the case. Intel cpus respect their Pl limits, in fact a lot more than amd cpus do, cause amd chips don't have an actual controller on the die, it's on the motherboard. If anandtech showed a 35w pl2 intel cpu pulling 400w then they are full of bs, but I don't think they did. I think you are simply making it up. Please provide the link.These are not stock settings. It even says so right there in the charts. W1z removed the temperature limit manually because he didn't want to thermal throttle. He even says so in the freaking review man.
Again, I'll ask you as well, very politely. You can hate intel all you want, but can you stop hating facts and reality? Please?
I don't believe any motherboard is setting 4096 W with ICCMax unlimited bit as default regardless of SKU, but now that Intel has published recommended guidelines and BIOS updates are underway to ensure compliance, I think at this point the misinformation can pretty much go.
As i have said before, so much anti Intel crap on this forum. Looking forward to Arrow lake proving them all wrong.............................again
The P cores have always been warmer than the E cores (out the box) on my samples of 12,13 & 14th gen but in saying that they do clock a lot higher.
I wonder if that's because the E cores on the new gen are getting a 50% increase in IPC, which although, remains to be seen.
"Intel unwraps Lunar Lake architecture: Up to 68% IPC gain for E-cores, 14% IPC gain for P-Cores
The resulting Lunar Lake mobile chips employ an entirely new design methodology that focuses on ensuring power efficiency as a first-order priority, and this base architecture will be used as the building block for Intel’s future products, like Arrow Lake and Panther Lake"Anandtech also said that the firmware of those boards seems to be problematic. The 14600k behave in a radically different way by pulling less power than the rated PL2 (181w)
Cooler parts (cache, rarely-used logic, etc. etc.) are supposed to be well-mixed with hot-parts (ex: multipliers, AVX-parts, etc. etc.) because the cool parts help cool off the hot-parts.
Its really obvious when its stated like that but... I mean... look at the original design! All the hot-P cores are sitting next to each other, heating each other up.
www.techpowerup.com/forums/threads/are-you-using-an-amd-ryzen-x3d-cpu-with-3d-v-cache.312452/
I also talked about TPU reviews being the basis for its readership purchasing decisions. Reviews that you are paid to proofread for god’s sake!No wonder Intel’s revenue, margins, stock and marketshare is falling. Their supporters and brand loyalists aren’t challenging the company so Intel thinks it has no reason to compete. Hopefully for competition sake, Intel never reads these posts and realized what extreme trouble they are in.
Again, Intel architectures are in last place behind all others (ARM, AMD, GPGPU) in IPC, efficiency and value. AMD has gone from 2% data center market share to 33% in five years and Intel supporters are talking about ISO. Lololololol!!!!!
The lowest latency dram we've ever had for a "word" didn't have integrated memory controller.
Intel Core 2 FSB platform, it's down to 25ns, that's extremely fast.
from cpu, out to chipset, then out to memory modules (lga 775,the top chipset is where memory controller and pci-e is)
What drives latency is interconnect protocols, buffering, error correction and translating, abstractions making development faster, less buggy etc.
TPU polls are more of a how do you feel? what do you want (to be true)? type of question than anything that anyone should hang their hat on beyond those notions.
AMD is very popular among a relatively small population of highly engaged and vocal enthusiasts, who clearly (especially according to what they readily admit), let personal biases and factors beyond the actual specs, nuances and facts of the products themselves, affect why they buy them.
I just don't let my feelings affect how I perceive reality. Intel is the king in desktops. Low idle - low load power draw, top ST and MT performance and great efficiency either out of the box (with the non k and t lineup) or after power limiting with the K cpus. AMD on desktops is what Intel was back in 2012. Stale. Stuck in 6 cores, reduced the amount of cores at each segment. People like you and denver, who himself admitted no way in hell is getting an Intel CPU is the whole reason amd is stale. You are buying their bs products no matter what, so here we are.
I buy what is best, I'm not a slave to any brand.
With that said I agree with you - intel should support more sockets, but PROPERLY. Not the way AMD is doing it.