Monday, February 6th 2023
Intel Meteor Lake to Feature 50% Increase in Efficiency, 2X Faster iGPU
Intel's upcoming Meteor Lake processor family is supposedly looking good with the new performance/efficiency targets. According to the @OneRaichu Twitter account, we have a potential performance estimate for the upcoming SKUs. As the latest information notes, Intel's 14th-generation Meteor Lake will feature around a 50% increase in efficiency compared to the 13th-generation Raptor Lake designs. This means that the processor can use half the power at the same performance target at Raptor Lake, increasing efficiency. Of course, the design also offers some performance improvements besides efficiency that are significant and are yet to be shown. The new Redwood Cove P-cores will be combined with the new Crestmont E-cores for maximum performance inside U/P/H configurations with 15-45 Watt power envelopes.
For integrated graphics, the source notes that Meteor Lake offers twice the performance of iGPU found on Raptor Lake designs. Supposedly, Meteor Lake will feature 128 EUs running 2.0+GHz compared to 96 EUs found inside Raptor Lake. The iGPU architecture will switch from Intel Iris to Xe-LPG 'Xe-MTL' family on the 14th gen models, confirming a giant leap in performance that iGPU is supposed to experience. Using the tile-based design, Intel combines the Intel 4 process for the CPU tile and the TSMC 5 nm process for the GPU tile. Intel handles final packaging for additional tuning, and you can see the separation below.
Sources:
@OneRaichu (Twitter), Thanks P4-630 (TPU Forums) For the Tip
For integrated graphics, the source notes that Meteor Lake offers twice the performance of iGPU found on Raptor Lake designs. Supposedly, Meteor Lake will feature 128 EUs running 2.0+GHz compared to 96 EUs found inside Raptor Lake. The iGPU architecture will switch from Intel Iris to Xe-LPG 'Xe-MTL' family on the 14th gen models, confirming a giant leap in performance that iGPU is supposed to experience. Using the tile-based design, Intel combines the Intel 4 process for the CPU tile and the TSMC 5 nm process for the GPU tile. Intel handles final packaging for additional tuning, and you can see the separation below.
75 Comments on Intel Meteor Lake to Feature 50% Increase in Efficiency, 2X Faster iGPU
28W is absolutely doable in 1.2Kg laptops because I've already owned two of them - a Lenovo S540 13" Ryzen with a 35W cTDP (dual-fan) and an HP Envy with 28W in a very thin chassis. The S540 was exceptionally small, light, and powerful for under £1000. We seem to be taking steps backwards when it comes to adequately-cooled, reasonably priced AMD ultraportables. My hope with these new Intels is that there will be so many design wins (because Intel) that at least some of them won't suck!
The reason I keep mentioning ultraportables with APUs is because anything approaching 1.5kg is going to lose out to the enormous selection of good, readily-available 13-14" gaming laptops. Not just in size/portability - but also cost, because those small gaming laptops are popular and seem to have economy of scale. So giving up a dGPU and going with integrated graphics isn't just for fun; it's a massive performance sacrifice that isn't worth making unless there's a significant benefit in portability.
Now I agree that APUs are still not what we are expecting of them, but maybe we're expecting too much? Or too much of a niche market, or both?
The clue is in the sentence "... Supposedly, Meteor Lake will feature 128 EUs running 2.0+GHz compared to 96 EUs found inside Raptor Lake..." contained in this particular story. First of all, only mobile Raptor Lake CPUs have 96 EUs so we are not talking about the desktop here. Secondly, there are only 6 Raptor Lake mobile CPUs with 96 EUs, they are all 6P + 8E and they form part of the H series. There is an HX Raptor Lake series that includes 8P + 16E, 8P + 12E and 8P + 8E processors but only with 32/16 EU graphic units (because they will ship with NVIDIA mobile graphics chipsets).
So all this seems to add up to is that these particular H series Raptor Lake mobile CPUs when upgraded to Meteor Lake will retain their current 6P + 8E cores, and the graphics unit will be replaced by a 128 EU graphics tile. That's it.
And those pics... Probably from an Intel presentation on the Meteor Lake H series processors. Of the Raptor Lake H series, the top 6 are 6P + 8E there is then a 6P + 4E, some 4P + 8E and a 4P + 4E chip. So in that context the diagram makes sense, and does not have wider implications, I think.
As for the iGPU... well it's tough to say when Intel has a lot to do with their dGPU, not to mention Intel iGPU has nearly never outperformed AMD iGPU. But perhaps with the experience they've gained through developing dGPU, their iGPU has a chance to evolve.
One more thing, I hope one day Intel could make real progress in making a power-efficient enough architecture, in which case they could get rid of so-called E-cores. P+E is never a final method. It does more scores than real-world performance, otherwise they would have already applied E-cores to their busniess processors.
For me, what I don't like is that Intel is selling us cheap E-cores which are Celeron/ Pentium Silver class chips at premium prices. The E-cores may be fast, but still much smaller, less complex and cheaper to produce. And you can clearly tell that Intel is spamming these small cores. And you can also tell that Intel is not really targeting efficiency when you have like 16 E-cores on say the i9 13900. In light loads, do we really need 16 E-cores? I am happy with my 12700K with just 4 E-cores.
Inefficient? No. Using way too much power,? Yes.
I've even seen similar comments about the 4090. Just because it's using a lot of power it doesn't mean it's inefficient.
For the others, that ask for more e or p cores, that's a weird way of formulating the question. What matters is what performance and efficiency you are getting for the money, whether Intel or AMD achieve that with one core, one type of multicore or fifteen types of cores, that's completely irrelevant. I would dare to argue that they know a bit more what they should do than the people in this thread.
A fully gaming capable NUC... *drool*Except CPUs don't work like that. Intel's E cores can't do all of what P cores can. So it definitely does matter how they arrive at performance. Similarly, look at AMD's X3Ds. Great at specific tasks, not so great at others. Efficiency is impacted in much the same way; will vary depending on task and CPU layout.E cores are about saving die space/transistor count and the die space won translates into needing less juice to power it, quite simply. E core is not about flexible cores - throttling a core already happens, but doesn't make it more efficient, its still the same floor plan, so it works on the same curve of power/performance. You can't power half a core - that's the moment you implement another core and divide the load.