.I quickly at various OEM sites and couldn't find any laptop with both X1E80/X1E84 and 32 GB RAM. Did I miss something?
yesApple's been using on-package memory since the M1. Intel is using on-package memory for Lunar Lake. Nvidia ARM SoC is rumoured to feature on-package memory.
Does it make sense for Qualcomm to also pursue on package memory in the future?
I'd rather see non-soldered memory than on-package memory in normal laptops. Ultra low power stuff is another matter.Apple's been using on-package memory since the M1. Intel is using on-package memory for Lunar Lake. Nvidia ARM SoC is rumoured to feature on-package memory.
Does it make sense for Qualcomm to also pursue on package memory in the future?
I can't give up hope. A fanless laptop is coming this time, I'm sure of it.
Apple doesn't release a 40 hour laptop because that would eat into their profits (why release one device when you can release a cheaper to make device with half battery life and sell twice the units for same BOM?).
Non-Apple users just buy an extra batteryDo you think people who have laptops with batteries that don't last a day buy TWO laptops to get through the day?
So do Apple users, you gotta stop generalising all Apple users as some rich as trust fund group.Non-Apple users just buy an extra battery
yes that’s likely right. It’s a different die, so.Interesting that we have Purwa between us but no announcement so far from QCOM
I wonder what they're waiting for give that they need to announce it as the silicon is shipping currently.
What the s#%*![Snap 8 Gen 5]
The software side of 8850 has already started, the frequency of nuvia is 5ghz, and it is estimated that the next generation will start at 4.5ghz.
Is that sarcasm.If you only look at the power consumption of dou, it is excellent.
Yea but Qualcomm is soldering RAM to the mainboard. At least on-package ram has advantages. Soldering RAM to the mainboard not so much.I'd rather see non-soldered memory than on-package memory in normal laptops. Ultra low power stuff is another matter.
With LPDDR, there's 3 ways to go:Yea but Qualcomm is soldering RAM to the mainboard. At least on-package ram has advantages. Soldering RAM to the mainboard not so much.
Battery life for laptops only matters to a point, and that point is "when you go to sleep" - and even there it is only relevant for extreme cases where people really are on their laptop every waking hour. So I'd argue that while the M1 battery life was impressive, especially when it was put up against Apple's Intel laptops (which Apple already cherry picked from Intel's most efficient SKUs) the battery life it provided it was overkill for most.
Sure there may be a handful of people who were demanding enough in their use and using it long enough in a day that they'd run down the battery and need to charge, but that's a niche market. I think its overblown as something that will create a big wave of people towards ARM PCs (even if you accept as fact that x86 can never be as efficient as ARM, which I don't)
The "mobile heritage" argument only worked so long as Intel/AMD designed a single core that had to scale from laptops to servers (modulo different amounts of L3 etc) Once you design a second core, all that's out the window, because they can target laptops specifically with the small core. And if Intel's claims are proven correct when people get their hands on their new "small" core, I think that will easily fall into Apple's "overkill" territory for battery life - especially if you consider Apple didn't exactly put the largest "legal" battery in the M1 Air. If they had the battery life would have been TWICE as long. Can you imagine the hand wringing from some if Apple sold the Macbook Air as a slightly larger/heavier model but with forty something hours of battery life? They didn't because they knew that was ridiculous, there's no target market for that aside from severe insomniacs.
So it doesn't matter if LL catches up with Apple/Qualcomm (and of course Apple is irrelevant as far as crazy wet dreams of ARM PCs being 50% of the Windows market) because it doesn't have to. It just needs to get from current x86 territory of "not enough battery life" to cross the Apple M1 line of "overkill". There's a lot of inertia to staying with x86, especially in the business world. Why would enterprise buyer choose ARM, given that the prices are not going to be any lower than x86? There's no upside for them choosing ARM, making this one of those "you'll never get fired for buying IBM" type of situations. Consumers may buy ARM, but they won't CHOOSE ARM. They'll just buy a PC at Walmart or off Amazon and not really realize what they're getting. Just like the average consumer doesn't know/care about the difference between Intel and AMD, they're just buying whatever Dell or HP is selling them.
I was watching Lunar Lake presentation, and came scross this slide about "Render Slice".Those are like GPU cores. Slices (defined in that post as physical instances) would be like multiple GPUs in a GPU. That sounds like a can of worms to me.
LPDDR6 is actually the biggest single Q. There’s really no other question of that caliber. We expect a big IPC upgrade and/or mild clock boost — I would say 25% total performance upgrade is the low ball, 40% the most I can see, Adreno being better is a given, NPU upgrade yeah though I don’t know about 100 TOPs? We’ll see. N3E/N3P also a given I think.X Elite G2 will be very good
• 2nd gen Oryon CPU
• Adreno 800 series GPU architecture
• 3nm node
• 2nd generation (~100 TOPS) NPU
• LPDDR6 (possibly?)
Doing it with LP5 and LP6 is even harder.If not, then one issue is then they’d have to ship a memory controller compatible with LPDDR5 and LPDDR6. With LPDDR4 and LPDDR5 this was a PITA apparently.
Wouldn’t take Kepler’s word for it but intuitively it makes complete sense because the architecture divergence is massive. So they’ll have to have two controllers if they want LPDDR5 and LPDDR6 support.Doing it with LP5 and LP6 is even harder.
And all of that is going to be Qualcomm Snapdragon X devices, since no other ARM vendors will be releasing chips this year.Rough projections suggest Microsoft expects to take roughly 5% of the market with the Arm-based laptops by the end of the year, selling about 1 million to 2 million units.
indeed.LPDDR6 is actually the biggest single Q. There’s really no other question of that caliber.
Microsoft will double the NPU TOPS requirment for next generation Copilot PCs accoriding to rumours...NPU upgrade yeah though I don’t know about 100 TOPs? We’ll see. N3E/N3P also a given I think.
And it will be a bigger PITA with LPDDR5/6.But LPDDR6? Big Q. Does it arrive on time, and do they want to rely exclusively on that? If not, then one issue is then they’d have to ship a memory controller compatible with LPDDR5 and LPDDR6. With LPDDR4 and LPDDR5 this was a PITA apparently.
Yeah, they need the bandwidth to feed the huge NPUs.If they do LPDDR5-only, then it kind of hurts for future proofing and it’d hurt more than doing that in phones because a PC can really benefit from LPDDR6 assuming they’d also widen the bus which I suspect they would to 192B.
Are you talking about Custom Oryon derived E-cores or ARM Cortex A5xx? The former is likely, while the latter is never going to happen.Also, I expect to see mobile E Cores from phones thrown in as long as scheduling can make good use of them.
People always talk about P/E core scheduling and stuff when it comes to Intel, and AMD to some extent (becuase of their Zen C cores). But Qualcomm was the first to bring a hybrid CPU architecture to Windows PCs with their Snapdragon 8cx. Doesn't Windows-on-ARM have a good in-built P/E core scheduler?Wouldn’t work the same way as macOS with APIs but they could just migrate from E to P for increased user interaction and stuff with kernel scheduling. It’s too tempting not to if only for area gains and mild battery gains.
I don't think it makes sense to go beyond 8P+4E for consumer/prosumer devices for now. Speaking of which, it's time to reveal this:We could probably see either 8 + 4 or a 12+4 setup for the big one, and I expect to then see a smaller die like Purwa doing 4 + 4 or 4+6 — could also use it in Android tablets or handhelds probably. (Or they do 8+4, who knows)