Intel says a lot of things. Let's see if this one pans out.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Intel Says New Lunar Lake Chips Have 'Fastest Built-In GPU in the Industry'
- Thread starter MacRumors
- Start date
- Sort by reaction score
I was talking about this. They don't mention the AMD iGPU but HX 370 comes with AMD 890M. The base M3 8-core GPU is already about 16% faster than 890M so it's as fast as Intel's top Ultra 9 288V with Arc 140V. For basically the same price as the new Dell XPS 13" you also get M3 with 10-core GPU which is cleary much faster than Core Ultra 7 256V with a downclocked Arc 140V iGPU.
Last edited:
Right now, a highly qualified no thank you. 😒 Let's hope that Intel has properly tested the CPU microcode to avoid the crashing issues that plagued recent generation CPU's.
I know lots of people are being laid off, but not right after their company gets a massive injection of capital from taxpayers. I'm saying that it's suspicious that a company gets a $20 billion loan and subsidies from the government and 4 months later lays off 16,000 of those tax paying workers. What aren't you sure about?
It's interesting to me that all the focus and attention is on the ultrabook and lower-end side of computing. The Snapdragon X, the suffix-less M3 and M4, and now Lunar Lake, are all vying for the same exact types of customers. Meanwhile, H-series Intel and M3 Pro/M3 Max don't seem like they're engaged in as hotly contested of a market.
This all makes sense. I'm not saying I don't understand the phenomenon. I just think it's interesting.
These are not the kinds of computers I use recreationally. A high-end suffix-less Mx or a binned Mx Pro is probably all I need on the Mac side of my tech arsenal. But a U-series Intel chip only really serves the IT consulting side of my computing needs. Recreationally, I generally want something with a bit more heft.
The U-series Intel/Snapdragon X/suffix-less Mx processor/SoC based computers are for the average user both in and out of the enterprise. This is where the action is (and where people are buying computers).
This all makes sense. I'm not saying I don't understand the phenomenon. I just think it's interesting.
These are not the kinds of computers I use recreationally. A high-end suffix-less Mx or a binned Mx Pro is probably all I need on the Mac side of my tech arsenal. But a U-series Intel chip only really serves the IT consulting side of my computing needs. Recreationally, I generally want something with a bit more heft.
The U-series Intel/Snapdragon X/suffix-less Mx processor/SoC based computers are for the average user both in and out of the enterprise. This is where the action is (and where people are buying computers).
I gathered the 30W Lunar Lake-V parts are the H-series replacement, but apparently, that’ll come later with Arrow Lake. (Though I’m reading conflicting information on whether Arrow Lake will have mobile parts.)
As for Snapdragon X, that’s where they’re starting. If successful, surely they’ll add higher-end SKUs in the second generation.
Last edited:
I think the question for the future is whether Intel can find a reason to exist beyond x86. I’m not at all convinced they can. I do believe AMD can.
I also think x86 will be with us for a while just because of the massive amount of legacy built on it. The question for the future is whether x86 needs to come from Intel to satisfy the demand. I don’t think it does. I believe AMD could eventually satisfy that demand just fine.
The point being that I think the world can continue on just fine without Intel and with AMD providing x86 parts out of the back of their catalog.
Oh, great. So Copilot+ doesn't = Snapdragon ARM processors. More confusion for the shopper.
Glad you mentioned that, because I was just about to ask if it was non-upgradable like with Apple.
Are the Intel Lunar Lake likewise without non-integrated GPU options, as Apple's SOC is?
I'm sympathetic to those laid off, but I don't want a company getting an infusion of U.S. tax dollars to use it to continue to run inefficiently by keeping employees it doesn't need.
A common criticism of a larger role in government is that government bureaucracies aren't subject to free market pressures and are thus prone to inefficiency and 'bloat.'
Intel needed to do this, but it may cast enough FUD (fear-uncertainty-doubt) on the Qualcomm ARM chips to stunt their adoption and development for Windows software for ARM.
Another question to consider - is x86 at a point where its prospects for ongoing rapid performance improvement are worse than ARM-based processors? The predicted pending demise of x86 isn't new...IIRC, back when Apple went to Power PC RISC-based processors, x86 was seen as long-in-the-tooth, hanging on for the sake of legacy, tired and surely ready to be put out to pasture before too long. And yet...decades later, it's...still...here.
Put another way, if the Lunar Lake Windows PCs match the Qualcomm Snapdragon Windows PCs, in theory it ought to be a no-brainer to go Intel and dodge the risk of software incompatibilities and possible slowdowns from emulation on ARM. But what about the prospects going forward? Is it practical for Intel to maintain the pace of improvement that Qualcomm and Apple can?
If Lunar Lake performs comparably to Qualcomm's offerings, is there any reason at this time someone ought to choose the Qualcomm over the Lunar Lake?
Last edited:
Intel followed Apple's lead on that for its chips for the Ultrabook market (I think starting with the 13th generation / Meteor Lake) to compete on power efficiency. Not sure about the rest of the mobile product line but I believe they'll continue to design around external RAM for the destkop and server markets.
Intel cancelling the Intel 20A dies likely means that for 2024 and into early 2025 they will just have enough TSMC Arrow Lake dies to do desktop. The fact that the 13/14 Raptor Cove stuff has the overclocking burnout cloud hanging over them is a even bigger forest fire for Intel. They need to shuffle that stuff off the market sooner rather than later. Dragging their feet continuing to try to sell lots of those isn't going to help them long term. Just a bigger liability time bomb with the extended warranties they trotted out. They need to sell better stuff in desktop space sooner.
13/14 Raptop Cove mobile stuff doesn't have the excessive overclocking modality so it is a bigger place to dump the 'extra' desktop dies they need to flush from the pipeline. However, tarnished reputation and competition from AMD/others means that won't last for a very long time either. But it is a incrementally less intense forest fire than desktop. The timeline likely is just in flux. ( If AMD Zen 5 does better than Intel thought it would then maybe Intel has freed up dies for mobile sooner. If the desktop works well then they don't. It just depends .... so why write a date in stone? )
Rather dubious they would have presumed that these were going to be widely successful. They were taking a server core and trying to apply it to the laptop space. In terms of iGPU performance, both AMD and Intel are now kicking their butt. AMD and Intel were going to respond to Apple whether Qualcomm/Nuvia showed up or not. It is not just an "ARM core" problem. They have a learning curve on gaming edge case GPU drivers for Windows to get down. Both Intel and AMD are doing similar efficiency gains by going to embedded LPDDRx so gaps are closing there.
What they deployed is a decent first version , but it also a stop gap until they can get version 2 out the door that is integrated from almost scratch with the rest of the system (i.e., designed by Qualcomm as opposed to gluing together a exteranl CPU core to internal Qualcomm uncore. ). They need a 'plain' X version that is competitive with Intel and AMD offerings in 2025 far more than they need breadth. Version 3 is about where it would be more prudent to branch out. [. Also part of the $1B acquisition overhead justification for Nuvia is also so they can roll out something in mobile phone land. That's a more likely "second die"; a smaller one. Far more synergies when looping in their cellular modem stuff for Qualcomm ... that just isn't primarily the WinPC space. ]
Pretty good chance also that several WinPC systems vendors have been cheerleading for more dGPU support set at higher priority than a larger iGPU. ( notion that Windows customers want to buy Nvidia GPUs. ) . In the bigger die (e.g., much bigger iGPU space) they will be trying to push out Intel , AMD , and Nvidia GPUs. Handle two competitors first before moving to three.
Going too broad too quick is what crippled Intel dGPU effort. "We're rich...just throw money at it" isn't likely going to work.
I found this in the Verge which was interesting:
"It’s a change that reduces the power consumption of moving data through the system by approximately 40 percent"
We've known (unless we're actively refusing to understand) that the integrated RAM is faster and lower power, but this is the first time I've seen anyone put a number to the difference. Apple doesn't give these kinds of stats, but that's a massive improvement for people that probably would never have swapped memory sticks anyway...
Intel is not particularly getting a massive injective of capital. The contractors that build mult-billion dollar fabs don't work for free. ASML doesn't sell tools for free. HVAC vendors ... not free either. The taxpayer money for very specific construction projects only (where what exactly is going to get purchased is explicitly outlined. ). The construction projects are new work (jobs) and the eventual factories are largely new jobs also.
The money was never about saving large number of existing Intel jobs that may or may not been effective. It was about not loosing more manufacturing jobs overseas. Those new factories offshore aren't going to pay US income taxes either.
Intel isn't the only one getting money. Most facilities that want to expand in the chip manufacturing footprint (that would have substantive either national security and/or number of jobs increase impact ) in the USA got money.
If the bus widths are the same width and memory at the same generation iteration it isn't necessarily 'faster'. Apple is wider than that most generic PC that is roughly keeping up with the same DRAM standards updates. Intel has gone wider here also if I recall correctly.
Multiple DIMMs on a single channel is chasing capacity far more than "faster" throughput on parallel loads.
The number is only relative to what Intel was doing. Dumping having to cover multiple protocols probably helps also. ( DDR and LPDDR ).
Arrow. Lake versus. Lunar Lake where the CPU cores are the same but the GPUs substantively different along with differences in memory channel width will be interesting to flush out over time.
The larger outsized performance improvement Intel is making here is in iGPU where they were trailing almost everyone else they are competing with.
The are two chiplets: one for logic (Intel fab) and one for CPU/GPU (TSMC fab) it says here:
Intel's 120 TOPS Lunar Lake AI PC chips have landed
: And all it took was some good old fashioned outsourcing to TSMC
Interesting: RAM is now integrated, just as with M-series at 16 or 32GB. Do it like Apple!
And here I am considering Intel's 12th Gen Chips old. What can I say about 8th gen? Ancient?
The CPU in Latitude 7390 was launched in 2017,
I would say any comparison to Lunar Lake is quite irrelevant.
Anyway Intel has been surprisingly transparent with Lunar Lake launch, tests, showcases and the fact that they are concentrating on efficiency is a very good sign.
There's more competition now between Windows laptops that has ever been in history, and all 3 competitors have objectively good chips. They only have to continue with the same direction.
The fact that Qualcomm is also targeting sub 1000$ laptops now is really great news, I feel like this segment has been neglected.
Last edited:
I doubt Intel considers Apple M chips competition for gaming. Most of the games there either aren't available on MacOS or are emulated so it would have been very very difficult to include the M3 in that chart.
Well, I look forward to the M4 comparison. It's good that everyone is applying pressure to Apple, who didn't significantly move forwards with M2 or M3. M3 Pro was an especially weak and disappointing iteration. Just ramping up clock speeds each generation can't continue much longer, especially without impacting heat and battery constraints.