I think I brought most of this up and generally agree. We don’t know what Nuvia is contributing to the rest of the SOC, he did say Nuvia will be an SOC, not just a core. But we don’t really know what that means and what other accelerators (we know Qualcomm has a bunch of its own) it’ll ship with.
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.
Qualcomm Nuvia vs Apple Mx
- Thread starter crazy dave
- Start date
-
- Tags
- amd apple silicon intel nuvia qualcomm
- Sort by reaction score
I think I brought most of this up and generally agree. We don’t know what Nuvia is contributing to the rest of the SOC, he did say Nuvia will be an SOC, not just a core. But we don’t really know what that means and what other accelerators (we know Qualcomm has a bunch of its own) it’ll ship with.
seems doubtful. When Nuvia was an independent start up it would have been extremely doubtful they would have "beat the entire farm" on GAA 3nm. It is a major inflection point on fabrication and if there is a manufacturing hiccup and you have ZERO current revenue to live off of ... that's would typically mean 'death' for a start up.
More than decent chance that Nuvia was just following the herd to TSMC. Probably on a more mature node relatively to their design arrival date just so that there would be minimal chances of a fab production slide or super scarcity of wafer starts available. ( before the pandemic. Now there is an even bigger logjam if you are very small player. )
If Qualcomm wanted to minimize integration risks and had some foundation work on TSMC N5 or N4 already queued up then it wouldn't be surprising to see them minimize integration risk with Nuvia to just join the herd.
Nuvia was early enough in the design then perhaps Qualcomm merged into a SoC they already had queued up for GAA 3nm. With the aggressive timeline they are projecting (sampling in mid 2022 ) that seems less likely.
I'm not sure Qualcomm is 100% "sold" on GAA 3nm. Qualcomm was one of the early folks who signed up with Intel 20A-18A. Qualcomm has lots of products so perhaps they are 'dipping their toe" with a minor future product with that move, but they are trying out options for the long term.
Qualcomm's 8cx Gen 3 is on "Samsung 5nm". The next incremental move would be "Samsung 4LPP" not 3. Circa mid-2021 'public' roadmap.
"...On its FinFET technologies, both 5LPP and 4LPP nodes are new to the roadmap, and set for high-volume manufacturing (HVM) in 2021 and 2022, respectively. ..."
Samsung: Deployment of 3nm GAE Node on Track for 2022
The "arrow transition" from 5LPE to 4LPE in chart above.
The fact that 3GAE "early" dissappeared from the roadmap probably has nobody beating the farm on 3GAA. Samsung is probably going to chew its own dogfood for all of 2023 before anyone with large volumes might jump on it 2024. ( probably will get some smaller volume players though in 2023).
If 4LPP is suppose to be high volume in 2022 then Qualcomm sampling in mid 2022 and going volume in 2023 would make sense. And if they "moved" Nuvia from TSMC to Samsung because the rest of the SoC was already committed. If they basically upgrading the 8cx Gen 3 with Nuvia cores and evolving the DSP/AL-ML/GPU/etc then there could be enough inertia to make that happen.
It seems that Qualcomm will use Samsung's 4nm node for its 8 Gen 1 and Samsung's 5nm node for its 8cx Gen 3.
Snapdragon 8 Gen 1 Mobile Platform | Latest 5G Snapdragon Processor | Qualcomm
Our most advanced 5G platform ever with ground-breaking innovations in AI, photography, gaming, and connectivity.
Snapdragon 8cx Gen 3 Compute Platform | Latest Qualcomm Snapdragon Laptop Processor | Qualcomm
The Snapdragon 8cx Gen 3 Compute Platform combines advanced compute, AI, and 5G technologies to power premium laptops. Explore its features today.
It seems that Samsung's GAA 3nm has delayed until 2023/2024.
He didn't say that . I think you are trying to read what you want to into what was said. First, they don't even have a public marketing/brand name for the core. It will be an SoC with Nviia inspired IP in the CPU cores, but that doesn't make it a 'Nuvia SoC' or that the ex-Nuvia team are the primary drivers of the overall SoC system design.
Second, the article has a snapshot from the Qualcomm Tech presentation weaved into it:
The major SoC component parts are all there. The likelihood that the Nuvia team is contributing a Adreno GPU , a Hexagon Process , Security ( Microsoft Pluton) , Qualcomm ISP , 5G Modem , etc. is really , really small. It is an ex-Apple team. They aren't coming with a stack of Qualcomm stuff.
On the left there is phrase "M-series competitive solution for the PC". So nothing major new in this interview. On the right "set the performance benchmark for Windows PCs". If Qualcomm is really trying to be a well rounded competitive Windows SoC solution they'll need to step-up their GPU cores along with the CPU cores. The ex-Nuvia folks aren't going to contribute much there.
Does the CPU have to share bandwidth the GPU and Hexagon? Yes. Could there be need from more memory channels to handle the increased workload? Yes. Are GPU folks going to allow the memory subsystem they have based their design on be thrown out the window so something new? That's somewhat suspect. [ Apple delegates a fixed subset of bandwidth to their CPU cores as not to impede the GPU . ]
That side deck is suggestive that back in 2019-2020 that "Next Gen CPU" box would have been tagged for "X2 + A710" and that this is acquisition is a substitution in a SoC they already have roadmapped.
Third. The actual question and answer
"...
IC: If you can perhaps clear something up for me: is the Nuvia team making a single core, or both a big core or a little core? Or is it that they’re dealing fully with the SoC structure into which you add in the connectivity and the graphics?
AK: It’s both, all the above. By that I mean that it's impossible for us to put out a chipset solution as sophisticated as this without having the entire system being taken into consideration. Think of it this way: the CPU by itself is part of the ‘one technology roadmap’, but so is graphics, and other things. Then we're really thinking about bringing a complete system solution to the PC and changing it in such a way that you don't go after the traditional designs. ... "
That initial fragment replay of "all of the above" doesn't make it a "Nuvia" SoC. The CPU is "only part of the" roadmap; not the defining element. As a "complete solution" there has to be an effort put into making the CPU well integrated into the overall system solution; but the CPU cores are not the solution.
Qualcomm only announced they were acquiring Nuvia in January 2021. Technically that didn't close until March 2021. This SoC is suppose to sample in mid-2022. So basically had/have a year to make this work. Doing a new SoC takes multiple years to do. So it is pretty likely some compromises are being leverage to wedge Nuvia's stuff into a SoC baseline design that was already in progress when they acquired them. Where the 'old' CPU cores were a modular element of the design, they'll fit this alternative modular set in.
Will it be a bigger die size than they initially targeted if Nuvia (like Apple) is leaning on bigger than Arm "normal" L1/L2/L3 caches? Probably. I don't think that baseline was frozen in stone, but where floorplan , power , and/or thermals could be straightforwardly extended there are likely some shifts being done. more performance and bandwidth isn't going to come for free. That's said, it is doubtful they are throwing it all out and starting over with some "ink not dry on their badges" team driving all the major design calls.
The CPU cores probably will come out to be less than 40% of the overall SoC die space. (decent chance down closer to 30%). It isn't going to be most of what is there. So naming the whole SoC after "Nuvia" is a pretty big stretch.
Is there really a huge gap between. mobile and ultraportable laptop at this point? The iPad Pro is running the exact same chip as the Macbook Air. Microsoft's Surface Pro 8 doesn't come with a keyboard ( it is a "slate" , non-clamshell, just like the iPad is).
Qualcomm's focus in Windows so far has been on "Always Connected Personal Computer". There is a higher percentage of those sold with a cellular modem in them then the iPad Pro or MacBook laptop. Presence of a cellular modem does not "drag in " a mobile adjective? Really? The point of being wireless is to enable the ability to move around at will (move ... mobile ). Disconnecting from Wi-Fi/Bluetooth broadens the range can roam even further.
It would make sense for Qualcomm to do a "Jade versus Jade-chop" variant of their SoC that doesn't have a cellular modem if they want deeper penetration into the Windows PC space. However, the primary design for Qualcomm is going to be the one that has a cellular modem. That is their primary business. Them coming up with a SoC that doesn't have a radio is similar to Apple doing a chip that doesn't provide any GUI. It isn't really the primary business they are in.
If Apple pushes their cellular modem into their iPad SoC that the plain Mx will get them also since Mx is being shared across iPad Pro ( and hence gets iPad Pro design requirements imposed on it. ) Long term Qualcomm is going to be much, much , much happier comparing to the SoC (and its associated costs ) then comparing agains the cellular radio less SoCs that AMD and Intel models will likely offer in the future.
Once get into the class of Windows PC where integrating a cellular modem into the design would be highly abnormal I suspect Qualcomm isn't going to want to go there with deep and heavy effort. Highly immobile , always plugged in PCs are not something they will likely drive these cores to over time. ( they'll leave the door open to looping back to server but unless the radio business craters and everyone else pushing server focused designs fails , that isn't likely. )
Similarly if. Android 12L ( large screen) and follow ons got decent development traction then there would be an iPad OS competitor. And Qualcomms SoC would probably would "fit" an "Android Tablet Pro" product as well as a ultraportable just like Apple's 'plain' M-series is destine to role fill.
If talking iOS then Apple has no significant fraction share with a M-series SoC at all. That's actually the A-series; not the M-series.
Apple does straddling across iPad Pro and Macbook laptops. Qualcomm highly likely wants to do very similar straddling. It has to span operating system providers because Android "Large Screen" hasn't been good. So there is a mix of Android xxL , Windows 11+, and likely Chromebook that Qualcomm will have to juggle through to provide a significant market fraction.
i think you are looking to have Qualcomm skate to where the puck has been as opposed to where the puck is going.
Qualcomm isn't trying to make folks switch off of macOS, Windows, iOS , or Android with a SoC. There is significant software component to peoples decisions that they don't "own". What they can control is making some platform they support go "faster" and/or "user less battery".
The "will my software run" is largely not Qualcomm's job. That is Microsoft's job. ( there is some sharing workload at the cross compiler , legacy "band aids" level, but big picture that is a Microsoft job).
If the Qualcomm SoC offers more performance than the 2-4 year old laptops folks are upgrading from , runs the software , and gets 2-3 hours better battery life, then that is not going to be a hard sell. The native performance levels and battery life are Qualcomm major points.
For the subset of the user base running. multiple platform apps ( e.g., ported to both Windows/macOS , browser apps , Electron apps , video conferencing/streaming , etc. ) it is also a bit of a farce that there is some huge segmentation gap between macOS and Windows. When Qualcomm is likely to compare M-series versus what they got it probably will be with a multiplatform benchmarks. Not some macOS, iOS, iPadOS, Windows, or Android specific metric. If they don't do it then many others in tech press will do it for them.
Most customers aren't running around with Apple envy. The metric in most cases won't be on the MBA but on some functionality that the Air delivers on. Like 10+ hours of moderate work battery like. Can multitask well. etc.
Once get into the zone of "want something that looks and behave like an Air clone" then spans way farther than just the SoC. Case design choices . Operating system level choices , etc. etc. are all largely out of Qualcomms control.
About 3-4 years ago AMD has a "bad" reputation in terms of performance. Can't "talk" their way out of that. Qualcomm has delivered some slow stuff.
If deliver something with a substantive performance uplift then folks tend to notice.
MediaTek’s new Kompanio 1380 ARM processor is really, really fast [VIDEO]
As a part of CES 2022, we have the new Acer Chromebook Spin 513 in the office and inside this handsome Chromebook is an SoC that we've been very, very excited to see in action: the MediaTek Kompanio 1380. To be fair, we thought the Kompanio 1200 was the model we would be looking out...
chromeunboxed.com
Can't find the article at the moment but its seems that they have caught onto the fact that the "always connected" part wasn't as big of a perceived value add as they thought it would be. [ Tethering and addition fees that several USA (and other ) cellular providers tack on means more costs also. The modem focus isn't surprising for Qualcomm but it is also a bit of a boat anchor also on delivering value to a wide segment of the Windows PC market. ]
Not really "half-asses" but really just relabeled phone SoCs for the first 2-3 years of the effort. Some of that was because Qualcomm nuked their internal design teams when Broadcomm started rattling take-over rumblings and they were in bad relationship with Apple.
As ARM standard designs started to look farther out than just doing smartphone cores ( Neoverse , X-series , ...) the possible solutions have been better ( like the 8cx Gen 3 which is really relatively radical new approach than "just a Gen 3" name suggests ). Qualcomm is trying to promote continuity ( and probably save big name change for cores they paid $1B for. )
But they are NOT saying barely anything about the MBA. It is far , far more you subsuming the MBA into the loop because Qualcomm is talking about the M-series. The M-series and MBA are two , very substantively , different things.
Qualcomm is not confused that they are a full system builder and vendor. They aren't really talking much about their completed systems. They are primarily talking about building blocks to build something that is competitive with an Air on some dimensions.
The recompile from x86 binaries to Arm binaries isn't a huge deal if operating off of a base that is faster than the "old" x86 environment. (it is largely how the previous two macOS emulation foundations worked). So if M1 is faster than x86 and Qualcomm sets an objective to try to match from themselves to match that performance level level then the binary conversion stuff will take care of itself.
If the M-series falls behind x86, then yeah they should change the comparison. The baseline need is to set a target level above x86.... M-series is that for the ultraportable and tablet space at the moment.
Qualcomm can't really guarantee that because Windows could screw it up. They could provide the baseline "horsepower" but if the OS scheduler and conversion software piss it into the ground then it won't be delivered to the end user.
AMD has been bitten by exactly this. "We got 16 cores in our SoC ... more power than Intel" ... and then the Windows scheduler pisses that down the drain.
Similar with Android between Qualcomm SoCs and end users. Some stuff gets lost in that Java layer.
Microsoft should be selling Windows primary pitch person. It is their product. Qualcomm sells SoCs so makes lots of sense to compare to another SoC with similar goals.
@deconstruct60 you have a lot here, I read it all, and I’m trying to think how to respond. Unfortunately some of it, to my reading, seems self-contradictory across these posts or written in a way that implies I disagree when I made the same point in an earlier post. Maybe I’ll try again in the morning. 
Cost Microsoft from what start date? I sounds like you are forgetting that Microsoft had Windows RT (and Windows 10 mobile ) . That is a start date back in probably 2010-11 for 2012 release (and 2015 for Windows 10 mobile ) . So 10 years at a $10-40M/yr year isn't a whole lot. [ From 2013-2016 Microsoft owned the Nokia smartphone business and tried to stuff "windows" into that. ]
Windows 10 on Qualcomm ARM start date was around 2016. But that wasn't like Microsoft was starting off in some blank, "greenfield" port.
Windows RT -> Windows 10 on Arm -> Windows 11 on Arm ... so basically at "version 3" which is about par the course of Microsoft getting anything substantively correct. Decade long evolution would add up. But isn't like they haven't shipped any product to pay for that also. Maintaining and evolving macOS solely on x86 over a decade cost a boatload of money also.
"Blown money wise" though, they blew more on the foray into the smartphones. However, that probably did assist in leading into the special latch up with Qualcomm they have been in for last several years.
The Qualcomm link up was probably a convenient interlude. Look at the major ARM implementors. Apple wasn't viable (they don't share). Samsung largely has always been just a arm IP value-add touch ups. Mediatek? Nah. Huawei/HiSlicon ( have to high to bet the farm on the Chinese Party good graces ).
If Microsoft needed a proxy to lean on Arm to make baseline designs better for Windows then Qualcomm was a better proxy than all of the others. Over time X1 , X2 , and now X3 have appeared. Likewise Qualcomm also ( if not stressed by greenmail stock vultures ) to go back into doing their own custom cores . ( irony Microsoft picked up the custom core folks that Qualcomm dumped around 2018 https://www.zdnet.com/article/micro...ineers-to-work-on-its-quantum-computing-team/ ). If Qualcomm had gutted their internal development teams in 2014-2016 then perhaps Microsoft would have gone a different way, but in 2015-16 they were a more robust choice.
Also by 2015-16 if looked closely it was probably easy for Microsoft to tell that Intel was making bonehead moves. Intel bobbled the 14nm intro around that time. And AMD was in the "we got something just wait" mode.
Microsoft already had a 32-bit core of Windows running on Arm. It is a drop dead easy "Plan B" to pursue. At worst it would be a tool to keep Intel (and AMD ) in line to making progress.
Microsoft is also leaning on Qualcomm for their Hololens product.
In 2022 though there isn't as great of a need to depend upon just Qualcomm. Samsung and Mediatek are going to get access to X2 (and X3 ) cores. Arm is far more focused on making bigger , more higher performing cores as they walk up to the server class SoCs. Microsoft themselves are jumping into doing custom server processor. (so in much less of a need of a proxy to talk to Arm. )
Does Microsoft benefit from an exclusion lock out at this point? No. Could Qualcomm ride on semi-standard ARM cores and hold a differentiation lead over the other possible Windows on Arm providers . No. ( even more so with Samsung partnering up with AMD on GPU which a better natural fit with Windows. )
P.S. Windows 11 on Arm is helpful not only because it brings x86_64 compatibility but that Windows 11 in general dumps a bunch of crufty 32-bit and older stuff. ( the folks grumbling about how can't run DOS apps are getting of at Windows 10 anyway. x86_64 or Arm doesn't really make a difference ).
Last edited:
I wouldn’t even rule out Microsoft working on a chip design with Qualcomm where Microsoft owns a portion of the IP (the parts that more efficiently handle code output by Visual Studio) such that they can then co-license the solution so that folks outside of Qualcomm could make compatible chips. Since Samsung’s been just producing other’s reference solutions for awhile, why can’t we expect that they’d license a WinARM solution design and make the version of the chip that goes into their ultrabooks?
No platform? What is 'below' or 'underneath' the Qualcomm hardware?
Qualcomm builds "engines and drivetrain" those are foundational elements.
Qualcomm doesn't sell completed systems or a full operating system. However, that is going up from the platform (foundation).
Complete to replace as the macOS foundation "engine" ? No. But that is like saying Honda motors can't compete with Ferrari engines in Ferrari forumla 1 cars.
Qualcomm can't hyper specializes what they engine is used for. Windows , ChromeOS , Linux, and Android are all options though.
Visual Studio is quite capable of producing native ARM binary code. There is little unique that Microsoft would bring to the party.
I suspect trying to say that Microsoft could help Qualcomm add some "x86 emulation " features to the CPU cores. Frankly, that probably isn't necessary. First, native Arm performance would be a bigger "pull" to get folks to move over to native code than trying to push optimizations into the hardware so that software developers could be even more lazy. Second, it probably would help as much as folks think it will. Incrementally, perhaps but huge changes ? Probably not.
In the evolution cycle of a architecture license implementation of an Arm core the first order of business should be to get the standard Arm stuff right. Once that is solid then could add flourishes, but need a solid foundation first.
Could Microsoft help in collecting up the most common blocks of instruction traces . The types and frequency of object method invocations and/or 'plain' function calls. Collect up the atomic locking idiums. Yes. Is all of that unique IP .... not really. It is more metrics of what is already there. It is lots of work , but isn't "invention".
There is really no good reason for Microsoft to fork central and foundation Windows code off into proprietary instruction types and/or opcodes off from standard Arm.
Better effort for Microsoft would be to try to expand their security processor into the zone of authenticating the firmware rather than going to gyrations to claim have booted on trusted environment when actually didn't even look. But Pluton is incorparted into both Qualcomm and AMD designs at this point. It is not a vendor lock in tech. It is a IP design module that others can put into their silicon dies if they want to.
Rumors have it is that there is/was some Qualcomm-Microsoft lockout contract for Windows on Arm. Long term, I doubt Microsoft would cling to something like that. (if it exists ) Short term it was probably better not to get into a "crabs in a barrel" situation where 'too many' Arm implementers were chasing 'too few' customers. Also most phone SoC vendors seem to have short attention spans on long term OS support. If have some soft of definitive longer term time horizon for them to shoot for then probably easier to get them to take a more long term view on support.
But yes.... Windows on Arm isn't licensed for but it runs on Apple's Arm implementation. It works on Ampere's server SoCs. Microsoft hasn't done much lockin with that port. They have some specific boot environment checks they want done, but as stated above they are open to licensing those .
I'm not sure there was a real hard core "lock out" contract. To date Samsung really hasn't had anything substantively better. Qualcomm has mostly been offering up warmed over smartphone chips. That is all Samsung had also. ( Samsung has been off in the weeds trying to do their own modified cores and GPUs and not really producing differentiating results . )
Samsung has been one of the vendors actively exploring with building systems with Qualcomm SoCs those. That has sometimes been a precursor to them jumping in with deeper penetration with their own in-house/vertical solution . Samsung is working with Google on the Tensor SoC in the Pixel 6. Pretty good chance that will help them evolve to a state where don't have to use the Qualcomm solutions for their USA targeted phones. Will it happen in 2022. Probably not. by 2024 or so ... has good odds. Exynos 2200 probably has some "version 1' issues to work out with shift to AMD GPU architecture baseline and hiccups on the Samsung fab.
Similar with doing an "iPad Pro" and/or ultrabook specific SoC. Probably 2023 or 2024 before they have something that is "decent enough" to use in their own products. If the ex-Nuvia core stumbles for Qualcomm, Samsung will be ready to pounce. I have trouble imagining Microsoft would provide "air cover" to Qualcomm hold Samsung back.
If the ARm implementors make "tablet and up" SoCs then Microsoft probably won't keep them out starting 2022 and going forward. Windows 11 is a real competitive starting point.
Qualcomm Snapdragon 8cx Gen 3 (ST 989/ MT 5533) ?
OEMVL OEMVL Product Name EV2 - Geekbench
Benchmark results for an OEMVL OEMVL Product Name EV2 with a Snapdragon 8cx Gen 3 processor.
browser.geekbench.com
Multicore is excellent, ST should be in the 1100+ range, given that an X1 is used. On the other hand Windows scores are typically lower than Linux/Android/MacOs scores for Geekbench. As example my Surface Pro X's Geekbench sores are roughly 10% higher under Linux.
Thats of course assuming that we are talking less than 10W power (8CX Gen1 and Gen2 are 7W TDP SoCs).
Last edited:
It’s only about 10% better in MT than A15, and nearly half the score on ST. 8 cores to achieve that?
some context
8cx is about. 723 / 2878
8cx gen 2 is about 788 / 3125
So single thread 25% jump over gen2 and 37% over gen 1
MT. 77% jump over gen 2 and 92% over gen 1
that is progress . About MBP 13” 2019 level
If the previous Window/Arm notebooks were too slow then this addreses most of the issue .
but not ready to compete in the > $ 1,200 sytem level in 2022.
Multicore is excellent, ST should be in the 1100+ range, given that an X1 is used.
X1 cores are smaller than A15 cores - hence lower single core score and more cores needed for same MT performance - no real surprise here.
So lets quickly go back to MT score - If we are taking the Surface Pro X as reference, the Geekbench score is essentially sustainable for long durations, because it does not have to reduce frequency. Most other designs consume much higher power in the short bursts Geekbench is running and hence need to throttle for longer loads. Good example is the Macbook Air M1, while in Geekbench showing similar scores as the Macbook pro M1 it does throttle quite a bit on longer workloads. Same you can observe for the A15 in the iPhone.
As far as M1 comparison are concerned, the M1 might still be ahead, no doubt. However if you compare this to contemporary 7W Intel designs like the Lakefield - this is a huge lead.
Of course, but it is very relevant when you compare something.
Besides this has been pointed out by other forum members several times, when a high Geekbench score of an Intel machine was compared to the lower Geekbench score of an M1 machine.
Not sure why such arguments seem to be only valid when the M1 is behind...
Keep in mind, that these are Windows scores, which should be only compared to other machines running Windows. When comparing against Linux or MaxOS machines you better take the Linux score:
I just ran this few minutes ago on my 8cx gen 1 under WSL2. It handily beats the Gen 2 under Windows
It is ready to compete, if you want to have closed and slim passively cooled tablet like the Surface Pro X...because there is simply no alternative independent of price.
If you put the same chip into a chassis, which could dissipate much more power, the vaiue proposition at 1200$ is not there anymore - agreed. ( I am ignoring the possible positive effect on battery in a larger chassis here - which could also be a good argument for some potential buyers in favor of the 8CX even in larger chassis)
Last edited:
For this A15 system
iPhone 13 - Geekbench
Benchmark results for an iPhone 13 with an Apple A15 Bionic processor.
browser.geekbench.com
the SQLite score is about 22% lower the the full ST score .
for the linked 8cx gen 3 entry the SQLite score is about 10% lower . It isn’t just about “8 cores” . It is also about how much die space throw into the cache hierarch ( and/or how big grow the die ) . The 8cx is dragging around the digital cellular also . If traded that for bigger L3 ( and L2’s ) or grew then closer to being just about cores .
Windows Scheduler is going to do better with 4+4 cores that are closer together than 2+6 that are very far apart ( because grew the P cores and soaked up so much die space had to substantively shrink E cores )
Come on. That ST score isn’t even good for a 2022 phone much less a notebook. The MT score is average for a modern phone but against the latest AMD or Intel generations it isn’t great.Keep in mind, that these are Windows scores, which should be only compared to other machines running Windows. When comparing against Linux or MaxOS machines you better take the Linux score:
I just ran this few minutes ago on my 8cx gen 1 under WSL2. It handily beats the Gen 2 under Windows
It is ready to compete, if you want to have closed and slim passively cooled tablet like the Surface Pro X...because there is simply no alternative independent of price.
If you put the same chip into a chassis, which could dissipate much more power, the vaiue proposition at 1200$ is not there anymore - agreed. ( I am ignoring the possible positive effect on battery in a larger chassis here - which could also be a good argument for some potential buyers in favor of the 8CX even in larger chassis)
It looks like they make this marketing crap to miners so good luck getting ahold of this anytime soon!
Well, we all know that if Qualcomm were really serious, they'd hire Jim Keller.