Qualcomm responds to benchmark cheating allegations — Snapdragon X Elite/Plus benchmarks claimed to be fraudulent (Updated)
Seriously concerning news for Qualcomm's latest
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 revealed X Elite's benchmark scores
- Thread starter sunny5
- Start date
- Sort by reaction score
Honestly, few details. Qualcomm unveiled a 10-core SoC called X Plus, the specifications of 4 models (three X Elite and one X Plus) and showed a bunch of graphics whose veracity some question.
Seeing the claim of it being “up to 50%” slower than advertised, my knee jerk reaction is to call ********.
I can understand exaggerations of performance based on specific workloads, the whole tech industry does it, right up to the line of “false advertising”.
But 50% is crazy. I think it’s best to wait until it’s in consumers hands before believing performance metrics.
I can understand exaggerations of performance based on specific workloads, the whole tech industry does it, right up to the line of “false advertising”.
But 50% is crazy. I think it’s best to wait until it’s in consumers hands before believing performance metrics.
I sure hope so, sub-50% is going nowhere fast...Intel and AMD will be laughing so hard...
This was the exact case with before and after the release of the surface pro x sq1/2 real deal was around 45% slower
People never learn, they love to be fooled and believe, in this age its better to judge the company by its real product in your hand
The more we approach the release...the more dirt comes out
Last edited:
I think that these chips will be a whole lot faster than the previous chips but I'm also annoyed with them comparing apples to oranges. I would be OK with them comparing the Plus to the M3 and the M3 Pro to the Elite but they're not doing that. I will also hold reservation until pricing is released. That will be the true comparison.
I'm worried about pricing around 32GB and 64Gb models... Let's hope they don't take the Apple way
How is comparing the Plus to the M3 and Elite to M3 Pro not comparing Apples to Orange? The Plus and Elite are the extremely likely the exact same die; just binned with a different label on them. The M3 and M3 Pro are completely different dies. You are trying to compare 1 thing to two things and trying to present that as "apples to apples".
This single die that Qualcomm is between the M3 and M3 Pro in size/functions. So yeah they pick on the M3 to get some multiple core wins. And they also cherry pick some M3 Pro stuff. They didn't make this die to be exactly a plain M3 'apple' like thing (nor did they make a Mn Pro 'killer' ). In all practical terms, there are NOT directly competing with the M-series at all. Qualcomm has to displace x86 SoCs out of Windows laptops to 'win'. Apple was never considering them for a spot in a Mac/iPad at all. It is not the same set of laptop chassis that they can get design wins for.
The single die that Qualcomm picked is mainly skewed to compete with the higher CPU core counts that Intel has been throwing at their laptop SoCs ( but using a style of only "P cores" like AMD. So competing with them also). Same thing with the Graphics. It is mainly designed to cover the current Intel laptop offering.
The M-series stuff is mainly to grab some hype train icing on the cake. If there is a whole bunch of chatter about picking a fight with M-series they'll get more press coverage. Also generates the buzz that maybe the Windows vendors can get more of Apple 'tax' like margins on any systems they create with these (i.e., gets Qualcomm more design wins). The M-series can't be "apples to apples" because the systems don't even run the same operating system. Macs don't natively run Windows and vice versa.
The other reason Qualcomm is pointing at the M-series is that it is almost as 'sparse' in diversity as they are. Apple has three laptop dies and they only have one. Qualcomm probably really doesn't want a "look at the wide variety you can buy' pissing contest with either Intel or AMD (and definitely not both of them). They have only got one die (this generation). Excessively point at M-series is an attention misdirect tactic that happens to have some hype benefits.
The Plus is probably closer to the 'natural' clocks that this die should be run at ( the server clock zone that probably got 'stuck with' with pivoted away from that usage. ) . The top of the Elite is likely a selection of some 'golden sample' dies that can be pushed harder than the others. It isn't a substantively different SoC product. Intel and AMD are the ones who overclock their products to goose them into higher brackets. Qualcomm probably isn't going to be penalized much there because Intel/AMD end up in even worse parts of the diminishing returns zone ( their SoCs in that zone are going to get substantively worse battery life. )
Again likely going to see laptops with lower resolution screens thrown out there that won't match up with the Macbook laptops. Pretty good chance these will have an "additional battery life" premium on the pricing. However, having more than a couple of vendors will also bring a variety of configurations and more competition. Probably will be at least one that this undercut Apple on price. It is just a matter of how they go about pinching the pennies.
Seeing that they only seem to have one core design, they can only make the SoC run hot or hotter.
However, I'd like to see Apple take a similar approach with a 12 or 24-core SoC but with only a pair of efficiency cores. If they added proper cooling and good graphics hardware, they could have a line of gaming laptop computers.
However, I'd like to see Apple take a similar approach with a 12 or 24-core SoC but with only a pair of efficiency cores. If they added proper cooling and good graphics hardware, they could have a line of gaming laptop computers.
Thats what M3 Max is. But the high manufacturing price of this SoC makes it non-viable as a mainstream platform.
According to Android Authority, it seems so.
| X1 Elite (X1E84100) | X1 Elite (X1E80100) | X1 Plus (X1P64100) | |
| Total package power (95% parts) | 98.50W | 52.92W | 42.52W |
| Total package power (50% parts) | 82.33W | 43.40W | 42.52W |
Exclusive: Here's what Qualcomm didn't tell you about the Snapdragon X series
The upcoming Snapdragon X1 will be available in five different models: see their specs and performance here.
www.androidauthority.com
Last edited:
Up until the Plus came out Qualcomm was comparing the 12 core Elite to the 10 core M3 which is the Apples to Oranges comparison I was referring. Now that they have a 10 core to compare it to then I would agree it's a better comparison. I'm also waiting for independent tests before passing judgement. If the plus does edge out the M3 I think it is only good for all of us as competition keeps everyone innovating.
You cannot trust benchmarking results, and there is no rational to compare the capabilities of Qualcomm XElite performance and efficiency cores to the capabilities of the corresponding M3 cores. They could be equivalent, better or worse. At this stage no one knows for sure. So core count tells you nothing with respect to Apple, AMD or Intel processors. It only provides insight in comparison between other Qualcomm XElite configurations.
Need to see the performance of actual machines in the hands of real users to know anything for certain.
Need to see the performance of actual machines in the hands of real users to know anything for certain.
There are no efficiency cores in the QC Soc... They're all performance cores.
Even more reasons why counting the numbers of cores tells you nothing about performance compared to Apple, AMD and Intel. Especially since you cannot trust benchmarking results, especially benchmarks provided by the manufacturers.
What is the comparison basis? Are you comparing the technology platform or the product?
If you are comparing the technology platform, the closest Apple equivalent to Snapdragon X is the M3 Max, since both have 12 performance cores (M3 also has 4 efficiency cores, they only add about 1.5 P-core worth of sustained performance).
If you are comparing the product, it makes most sense to compare by price or by target power usage. I'd say by price, because that's what is relevant. If Qualcomm can sell laptops with a 12P-core CPU for the same price as a 4P+4E M3 MacBook Air, that's a fair performance comparison in my book.
A bit more detail about the allegations at ExtremeTech… apparently the allegations come out of the OEM Partners, who have been unable to get silicon to run as fast as Qualcomm’s own benchmarks.
Not that shocking when the product have to go into an actual product with form factor limitations. Marketing have no such limits 🙈
I guess it also doesn't help that Qualcomm had to lower their target clock speed by 100Mhz as well as lower the LPDDR5X memory speed by a smidge.
It would be interesting to see if it could go in a similar form factor as the MacBook Air (fanless) and what price it could hit.
I am very curious about single-core power draw of Oryon running at over 4Ghz. I can understand OEMs not being able to hit multi-core performance targets, it was fairly obvious from the start that the Snapdragon X platform has some massive issues with multi-core scaling, but the apparent inability to hit the advertised single-core frequencies is worrying. Some knowledgeable people here and elsewhere have pointed out that Oryon was designed as a server core, it is very much possible that 3.4-3.5 Ghz is it's sweet spot and going beyond that requires significantly higher voltage. I hope they are not running those cores at 15-20 watts to reach 4Ghz — that's Intel's territory.
So it seems. According to Android Authority's report on power consumption and performance, the most balanced SoCs are the X1E78100 (12 cores at 3.4 GHz) and the X1P64100 (10 cores at 3.4 GHz) with the other two dropping in efficiency, especially the X1E84100.
Still, if the quoted 50% speed is all they can achieve, that would be shocking indeed. Qualcomm had better send around some engineers to see exactly what is going wrong, otherwise a lot of people looking forward to these products will end up disappointed.
and some were replying to me 1 month ago that this cannot happen, this team is from Apples former SoC team(like it should matter for the end user?!) and users dont understand how marketing works and how open SoC charts vs actual product acts but even if you dont know, youve seen in the past with other QC devices that marketing charts vs real charts were around 45% difference. Some people never learn unfortunately and sorry for those who had high hopes...now that this is settle lets see now the other big thing, compatibility and how will work with our own applications if this will be almost the same as it was in the past, then this is another big failure and this should stop...this thing started with "winRT" and still continue because people are feeding this with their own wallets