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.
75 Comments on Intel "Arrow Lake-S" to See a Rearrangement of P-cores and E-cores Along the Ringbus
Intel failed with Netburst; AMD rocked K7/K8 and derivatives
Intel rocked Core architectures; AMD failed with Bulldozer
Currently, AMD is doing better with Zen and Intel is playing catch up with respect to process nodes, core counts and efficiency. Changes such as the ring bus interconnected topography might be steps in the right direction. However, I don't see much changing for gamers and the data center. Intel is losing ground here. People can say you can underclock Intel all you want but you can do the same with AMD. I want out of the box efficiency while staying close to the highest possible absolute performance levels across a variety of applications. This is AMD hands down right now. Zen 5 will only improve things by an estimated 15-20%. It remains to be seen if Intel's new gamble with Arrow Lake will change matters much.
The world has gone crazy.
Is that clear enough?
Maybe you mean something different with efficiency. What I'm talking about is you take 2 cpus, you put them at the same power and measure which one is faster in apps like cbr23, blender etc.. Intel tends to have a big lead there in most segments, as computerbase testing will show you. The 7800x 3d you keep mentioning is vastly losing to my 3 generations old 12900k in that. I have no clue why you keep mentioning it, MT efficiency isn't really it's strong point.
We also don’t need another thread polluted by “moving the goal post” fevgatos. TPU reviews and nearly all other sites conclude and state intels inefficiency problems in the current landscape. Any processor can be tuned to play the cherry picking game he presents, and there’s already a thread he dedicated to doing so.
How many times to we have to beat these very dead horses.
The only fair way to measure effiency is to either lock a target performance and see how much power they consume or lock the power and see the performance. By your logic Zen 3 is more efficient than Zen 4 in a lot of tasks.
Also TPU and all other sites conclude that the non k or T cpus are the most efficient CPUs humanity has ever laid eyes upon.That's simply not true though and I don't get why people keep repeating it. Take an R7 7700x, limit it to whatever power you want, it will lose - by a lot - to an i7 13700k in both performance and efficiency. Computerbase already tested this
www.computerbase.de/2022-10/intel-core-i9-13900k-i7-13700-i5-13600k-test/4/#abschnitt_leistung_in_multicorelasten_klassisch
A 7700x at 142w is slower than a 13700k at 88 watts. At 125w it is tied to a 13700k at 65w. It needs almost twice the power for similar performance. I get that a lot of people don't really care about efficiency when it comes to desktop CPUs, but that's not a reason to spread falsehoods.The only scientifically proper way.
Otherwise you are not testing efficiency but out of the box settings. Which is fine as well of course, but your conclusion can't be about efficiency if you are not testing for efficiency.
Looking forward to arrow lake, it's looking better with every snippet.
A 7700X vs a 13700k is a horrendously disingenuous comparison. On no planet bar LN2 overclocked will a 16 thread processor outscore a 24 thread processor. You’re looking to compare a 7900X/7900X3D.
Stop acting like a clown.
The 7900x / 7900x 3d had a way higher MSRP, being identical to a 13900kf (and the latter is obviously more efficient).Totally not true. The 13700k had an msrp of 399, the exact same as the 7700x. The 7900x had an MSRP of 549$, actually matching the 13900kf.
I get the current pricing argument, but that an efficient CPU makes not. Dropping the price doesn't mean your chips are more efficient. Say Intel drops the 14900 to 150$, will we suddenly be comparing it to a 4core amd part and saying it's more efficient? I mean come on, that's just silly.
Seems like a relatively minor change, but it should be helpful just the same. I'm sure it could've helped some on my 14700K with some of the scenario's I played around with testing out how the hardware behaved and reacted with adjusting P core and E core ratios and/or disabling some of them. From a tuning perspective this just makes things better balanced and easier to fine tune for a particular usage case and make the most of it.
I'd say generally most would be targeting heavier usage of either P cores or E cores for more ST or MT driven scenario's, but there are also mixed usage scenario's and this should aid a good bit in those use cases to extra more performance. Like if you want to tune 3/4 prioritized more for P cores and 1/4 for E cores or in reverse this should behave more nicely than 14th gen at doing so. It remains to be seen in practice how much of a difference that makes, but it really should be a improvement.
A little bit underwhelming, but the low hanging fruit honestly is mostly picked over by this point.
Like, really? And as previously mentioned I’m done entertaining your troll behavior.