I’d take that bet as well. In fact I think if they hit 75% of what Apple has done it will be near miraculous, (but still a big step forward though) not due to lack of talent, but first because of the mounds of heavily protected IP Apple has (and they can’t copy directly) but also just tech history generally.
Much as the term is used and abused, true “game changers” are nearly unicorn in nature.
PS: I have a good friend at Apple and he tells me that Apple is just waiting for QUALCOMM to even toe up to the line in “replicating” what Apple has done with their ARM designs and that the Apple of today is 1000% different than the one that ultimately caved to MS in the 90"s
Ok, but let’s clarify, they only need to surpass any of the M-series, ie. if they clearly beat the M1–ie. >CPU, GPU, and power efficiency–it’s a success.
Hifihedgehog
(Hifihedgehog - Waiting for Surface Pro 10!)
84
I’m no betting man, but if I were, I would.
Let’s put it this way. If Qualcomm’s Snapdragon 8 Gen 3 already beats Apple A16, which is Apple’s smartphone equivalent of M2…
…and Nuvia cores beat both by a significant margin…
…and Snapdragon 8xc Gen 3 is–REPEAT–three generations behind Snapdragon 8 Gen 3 (Cortex-X1 compared to Cortex-X4, meaning (116%*125%*120-130%) Snapdragon 8 Gen 3 is ~75%-100% faster than 8xc Gen 3 when scaled up
…therefore Snapdragon 8xc Gen 4 (which will incorporate Nuvia) will beat Apple M series. Case closed. End of story.
Hifihedgehog
(Hifihedgehog - Waiting for Surface Pro 10!)
85
Qualcomm then has done a masterful job to sandbag then if this is the prevailing perception. That might be all part of their master plan by purposely putting older Cortex-X1 cores into these 8xc Windows-on-ARM processors so the jump is even more startling and unexpected to the tech evangelists and chip workers. I can tell you definitely already that if Qualcomm made 8xc Gen 3 with the same modern Cortex-X3 cores that we see now in Qualcomm 8 Gen 2 instead of the Cortex-X1 cores as found in the 2020 Snapdragon 888, there would be 45% higher single-core performance. That simple math is an assured given just from looking at the current licensed Cortex-X3 designs. The beauty of the ARM licensed designs is you know how predictably they scale in clock speed and IPC from licensee to licensee. I would expect similar gains on the multicore front as well. So I think the surprise is going to catch a lot of people here off guard when Nuvia and Snapdragon 8xc Gen 4 drops.
Honestly, if they can solidly beat M1 on first-gen Nuvia, isn’t that enough?
Aiming too high was the downfall of QC the last time around. Inviting comparisons to the i5-8250U painted a giant target on their back–which every reviewer took a shot at.
I really hope they don’t go around boasting “better than M3”, because getting to where Apple was with M1 Macs, but for WOA, is where they need to be squarely focused.
No. The comment was that QC would surpass Apple in 2024 - your interpretation would only be surpassing the Apple of 2020 - in 2024…
Hifihedgehog
(Hifihedgehog - Waiting for Surface Pro 10!)
89
Wow, what did I just read? I love Apple and they are still managing to cling to the market leader crown, but that gap has narrowed by a lot since 3-4 years ago. Before we go on the part about bets about what hasn’t yet happened, let’s be clear on what has happened because this misconception is blurring perceptions here already about what’s ahead. Apple wants people to think they are 3-4 generations ahead which was the case several years ago, but that is no longer. Why? Again, the double digit year-over-year gains that Qualcomm has been pulling out while Apple rests on their laurels doing meager single-digit gains and calls it a day.
They are at 1-2 generations right now at best. Specifically, the Snapdragon 8 Gen 2 (below) performs like an Apple A14 in single-core and like an Apple A15 in multi-core. That’s exactly 1-2 generations behind by very definition. Let’s not call this a 3-4 generations victory, okay? That is old hat and outdated information from the A12 and A13 glory days when Apple was at their peak dominance. Now? No longer.
It seems like Microsoft are in an abusive relationship with Qualcomm (who are well known to be difficult to work with). And I don’t know why, when Microsoft could buy them if they really wanted to (terrible idea).
They really should have made good with AMD beyond a token Surface Laptop (Xbox is a different matter). Though I suspect Intel have threatened to have a tantrum if they do.
And once again on the Surface anyway (pun intended) Qualcomm seems to be on a similar path with the Nuvia chips with big claims being made that customers will expect them to deliver.
What I still can’t make sense of his how MS has handled this almost from the start. I don’t think it can be overstated how botched the launch of the Pro X was and the lasting damage that WOA incurred fairly deserved or not.
In fact our single largest Surface customer went very big on the launch of the Pro X based on the claimed performance and battery life. And after less 6 months they returned all of them all to MS and in fact MS replaced the entire customer team for them.
So be to clear, I really want WOA to succeed for a variety of reasons and I’ve also been an ARM advocate now for several years.
And lastly, at least in our market customer base, IMHO MS efforts so far to promote even the new Pro 9 5g have been at best, timid, and generally I don’t get the sense that many of MS own employees are on board with WOA especially those in divisions other than the Surface team.
The apathy of the Windows Serrver software folks is especially striking.
Of course these are very much my own observations and conclusions, so take them as such. But again @Marty because of my belief in ARM more generally, and the pedigree of the Nuvia engineers, I remain hopeful about WOA.
PS: I was there at the launch of the original Pro X and to this day it’s the only Surface product that I bought and then returned. OTOH, I’m fairly happy with my work provided Pro 9 5G, though I wish they had opted not to pad on the weight.
I just realized that I forgot to add that it’s doubly impressive because Parallels uses only half of the chip (e.g. 4 cores on the 8-core M1/M2). It’s running Windows 11 with one hand tied behind its back, so to speak.
Unfortunately you still have to get the pro version but it’s “only“ $20 more than the base version, which I see has gone up a lot since a year or so ago when I think it was $70. Of course the pro version is yearly for that price and not one time.
None yet. I guess it’s provision for the future. Unless they’re faking virtual CPUs?
Hifihedgehog
(Hifihedgehog - Waiting for Surface Pro 10!)
100
No, they use the vCPU terminology since hyper-threading/simultaneous multithreading (SMT) exists on chip designs and platforms other than Apple that they cater to in their other products, especially RAS. So far, Apple has not used that approach in their CPU design, so since thread count equals core count on Apple chips, vCPU equals core count here. But in case Apple does implement SMT in the future, know that vCPU equals thread count.
