News Snapdragon X Elite in the wild is allegedly slower than iPhone 12 — first benchmarks of Samsung Book4 Edge disappoint

Are the products readily available for purchase? If so, why do you (the author) think that waiting will change things enough for this to recoup the performance lost? Is there sufficient evidence this is Firmware/BIOS/Driver issues and not an extreme case of Cherry-picking?

I'm not in complete disagreement with the "wait" approach, but I honestly think Qualcomm and whomever chose to build with their new SoC should be responsible for this pathetic outcome; specially for early adopters. It's not like Qualcomm is a start up and has no past experience with building things. They don't need the weird nerds jumping in front of the valid criticism going their way.

Pitchforks and torches at the ready, always.

Regards.
 
Are the products readily available for purchase? If so, why do you (the author) think that waiting will change things enough for this to recoup the performance lost? Is there sufficient evidence this is Firmware/BIOS/Driver issues and not an extreme case of Cherry-picking?
...
The article says the chip didn't boost from 2.52GHz to 4.0GHz during the test like it should. This should be easily fixed with an update.

Edit: Just to add, compare the single core scores. (1841 vs what it should've got, 2977). That's approximately a 60% gap, which just so happens to be the same gap between 2.52GHz to 4.0GHz. This is absolutely a boost issue. AFAIK, Qualcomm's mobile chips do not use turbo boosting, so this is new for them. If the chips are ruined and can't turbo boost, then we can pull the pitchforks out, but I am still expecting a patch.
 
Last edited:
I'm not in complete disagreement with the "wait" approach, but I honestly think Qualcomm and whomever chose to build with their new SoC should be responsible for this pathetic outcome; specially for early adopters. It's not like Qualcomm is a start up and has no past experience with building things. They don't need the weird nerds jumping in front of the valid criticism going their way.
The CPU they are releasing is from a startup they purchased though. That is bound to have some hiccups, especially if QComm is rushing it out the door. Assuming any of this holds once the official reviews come out.
 
Interesting the plugged in and unplugged were basically the same. Maybe it's a power management bug keeping it in low power mode?
That's the benefit of using ARM. You get same performance both plugged and unplugged.

Albeit if these rumors are true, then it's a slap in the face for the whole PC Reborn mantra by Microsoft and Qualcomm.
 
Slow news day? Benchmark amnesia? Doesn’t Tom’s Hardware remember the benchmarks that users literally ran over the last week and at Computex on Snapdragon X Elite disproving? This isn’t hard, folks. See below. Results showing otherwise.

https://browser.geekbench.com/search?q=Snapdragon+x+elite

This one cited score is off because of a firmware or similar device specific issue. In light of the other results, any sane person should be lifting their eyebrow. Do people really honestly think Snapdragon would release a product worse than their Snapdragon 8xc Gen 3 from over two years ago?


Ignore INIYSA in particular. He is an immature bratty teenager or 20-something Apple silicon fanboy who frequently trolls other chip brands and who posts painfully obvious malarkey like this.
 
The CPU they are releasing is from a startup they purchased though. That is bound to have some hiccups, especially if QComm is rushing it out the door. Assuming any of this holds once the official reviews come out.
Sorry, I'm not interested in your excuses for them. If you go back and look at what Qualcomm has said about their timeline for shipping products, these Nuvia/Oryon cores have slipped a couple years, at least!

In terms of performance and efficiency, the one place I'll cut them a little slack vs. Apple is that they're on an older node. These new Snapdragons are on TSMC N4 (not sure which), while Apple's latest iPad (not sure about their phones) is on N3B. Still, that means they have no excuse not to beat at least Apple's M2, which was made on TSMC N5, I'm pretty sure.
 
A side note: If Windows Recall is turned on (which may be default in current test builds), you may have a 10%+ performance hit on Geekbench scores. My X1E78100 went from 22xx to 24xx after I turned off it and related services.
The mighty NPU usage is always 0%.
 
Last edited:
  • Like
Reactions: bit_user
Magical ARM thinking at its best (worst).
Actually, I thought it could be taken the other way - meaning that even if you plug it in, you still don't get any extra performance. Whichever was meant, it's clearly at odds with what Qualcomm has published about the power envelope of these SoCs. The specifics matter a lot more than any generalizations (which don't hold either, BTW).
 
  • Like
Reactions: usertests
Edit: Just to add, compare the single core scores. (1841 vs what it should've got, 2977). That's approximately a 60% gap, which just so happens to be the same gap between 2.52GHz to 4.0GHz. This is absolutely a boost issue. AFAIK, Qualcomm's mobile chips do not use turbo boosting, so this is new for them. If the chips are ruined and can't turbo boost, then we can pull the pitchforks out, but I am still expecting a patch.
That assumes that performance scaling is perfectly linear with frequency which is almost never the case.
 
That assumes that performance scaling is perfectly linear with frequency which is almost never the case.
It's probably closer than you think. Here's what I discovered about the Golden Cove and Gracemont cores, in Alder Lake:

kMltPg1.png


SUdXdDM.png

These are each measuring a cluster of 4 cores, rather than looking just at single-core scaling.
 
  • Like
Reactions: ThomasKinsley