Snapdragon X Elite Emulation Performance and Battery Life Tested!
video description
Date: 2024-05-30
Related videos
Comments and reviews: 20
MrPerrisC
Thanks for the comparison, but frankly, x is only competing against the very worst CPU Apple makes and the midland grade meteor lake
There's no reason anyone should be looking forward to this CPU unless you're locked into the surface, since other x86 CPUs thrash the model surface uses
X only beats the very worst M3 , and gets thrashed by the newest worst, the M4
X also loses to year old x86 AMD 7940 HS which isn't close to the best x86, and gets thrashed by the yearold HK and 9000 series
While losing to 7940hs, x only gives 10 percent better battery life
On top of any of this, by the time X is released, ZEN 5 is around the corner, since zen4 gives it a beating, we can imagine zen 5 gives it a bloodbath
What's odd is these thrashings are already published, yet for some reason reviewers seem impressed that this chip is so bad, it has to go against cherry picked CPUs where it barely shines at specific tasks but not all tasks.
Off topic, I point out, ARM is smoke in the mirror, it's not the instruction set giving it performance, it's transistor count, needing exponentially more transistors to still lose to x86 at everything but instruction per watt
For instance, the very best DESKTOP Multra (drawing 90 watts) has 100 billion transistors and loses to 7945hx LAPTOP cpu that only uses 10 billion, granted the ultra GPU is on SoC, but adding a better GPU to the HX it still has less than half the transistor count of Ultra.
Apple, ARM and Qualcomm have done a fabulous job marketing an instruction set that needs exponentially more assets to compete against x86
reply
Thanks for the comparison, but frankly, x is only competing against the very worst CPU Apple makes and the midland grade meteor lake
There's no reason anyone should be looking forward to this CPU unless you're locked into the surface, since other x86 CPUs thrash the model surface uses
X only beats the very worst M3 , and gets thrashed by the newest worst, the M4
X also loses to year old x86 AMD 7940 HS which isn't close to the best x86, and gets thrashed by the yearold HK and 9000 series
While losing to 7940hs, x only gives 10 percent better battery life
On top of any of this, by the time X is released, ZEN 5 is around the corner, since zen4 gives it a beating, we can imagine zen 5 gives it a bloodbath
What's odd is these thrashings are already published, yet for some reason reviewers seem impressed that this chip is so bad, it has to go against cherry picked CPUs where it barely shines at specific tasks but not all tasks.
Off topic, I point out, ARM is smoke in the mirror, it's not the instruction set giving it performance, it's transistor count, needing exponentially more transistors to still lose to x86 at everything but instruction per watt
For instance, the very best DESKTOP Multra (drawing 90 watts) has 100 billion transistors and loses to 7945hx LAPTOP cpu that only uses 10 billion, granted the ultra GPU is on SoC, but adding a better GPU to the HX it still has less than half the transistor count of Ultra.
Apple, ARM and Qualcomm have done a fabulous job marketing an instruction set that needs exponentially more assets to compete against x86
reply
vernearase3044
Really, the snappiness of a consumer computer relies mostly on the single core performance as most consumer workloads are not multithreaded - it's when you start getting into video/audio editing, graphics design, or transcoding where you start to make use of those extra cores.
So when you denigrate workloads like Speedometer 3.0 there's nothing in the test that's Apple specific - it simply points out that single core speeds are what the consumer notices. This test is what _I'd_ use to determine a computer's snappiness and speed.
Most consumers owning a computer with a zillion cores are simply running paying for a lot of idle cores sitting there soaking up power - unless the SoC is smart enough to shut down the idle cores (which my core-i9 10910 10c 20t 2020 iMac 5K clearly isn't).
reply
Really, the snappiness of a consumer computer relies mostly on the single core performance as most consumer workloads are not multithreaded - it's when you start getting into video/audio editing, graphics design, or transcoding where you start to make use of those extra cores.
So when you denigrate workloads like Speedometer 3.0 there's nothing in the test that's Apple specific - it simply points out that single core speeds are what the consumer notices. This test is what _I'd_ use to determine a computer's snappiness and speed.
Most consumers owning a computer with a zillion cores are simply running paying for a lot of idle cores sitting there soaking up power - unless the SoC is smart enough to shut down the idle cores (which my core-i9 10910 10c 20t 2020 iMac 5K clearly isn't).
reply
BrockGunterSmith
I’ve been running Adobe Lightroom Classic on all of my M series Macs and even my M series iPads (edit: my mistake, only the full Lightroom runs on iPadOS, I’m wrong about Lightroom Classic running on iPad OS), where it runs ARM native as well, for YEARS. I’m really confused about why they left it out of that test when it is one of the best examples in my mind of professional apps on ARM. I just double checked and Adobe released the native ARM version of Lightroom Classic in 2021. It runs ridiculously fast on ARM which is why it is kind of shocking to not see the M3 MacBook AIR in that test.
reply
I’ve been running Adobe Lightroom Classic on all of my M series Macs and even my M series iPads (edit: my mistake, only the full Lightroom runs on iPadOS, I’m wrong about Lightroom Classic running on iPad OS), where it runs ARM native as well, for YEARS. I’m really confused about why they left it out of that test when it is one of the best examples in my mind of professional apps on ARM. I just double checked and Adobe released the native ARM version of Lightroom Classic in 2021. It runs ridiculously fast on ARM which is why it is kind of shocking to not see the M3 MacBook AIR in that test.
reply
TokyoNeko8
I appreciate the benchmarks but I want to see non-Microsoft Commissioned comparison. Some of the missing data and footpints worries me. battery size, actual Divinci Resolve results. Modern AI test results, is this really using the NPU of M3 or GPU since that's how inferencing is all done on METAL/GUI power. Just saying there are spots of data that is missing. I still think SdragonXElite will kick ass and on par with M3 base. Great achievement from the super crappy ARM device and software support before. The biggest improvement is the EMULATION!! BETTER THAN 12th gen intel cpu. OH YAH!!!
reply
I appreciate the benchmarks but I want to see non-Microsoft Commissioned comparison. Some of the missing data and footpints worries me. battery size, actual Divinci Resolve results. Modern AI test results, is this really using the NPU of M3 or GPU since that's how inferencing is all done on METAL/GUI power. Just saying there are spots of data that is missing. I still think SdragonXElite will kick ass and on par with M3 base. Great achievement from the super crappy ARM device and software support before. The biggest improvement is the EMULATION!! BETTER THAN 12th gen intel cpu. OH YAH!!!
reply
zodwraith5745
So nothing surprising but also kind of concerning that nothing stands out as excellent considering it's a _paid_ advertisement choosing a ton of ARM native apps to show it in the best possible light. Especially when announced pricing isn't cheap. It looks like they _need_ to launch it now to minimize comparison to lunar lake coming out later in the year that's supposed to be a bigger improvement over meteor lake than meteor lake was over alder/raptor lake.
Still it's no slouch and more competition is a good thing.
reply
So nothing surprising but also kind of concerning that nothing stands out as excellent considering it's a _paid_ advertisement choosing a ton of ARM native apps to show it in the best possible light. Especially when announced pricing isn't cheap. It looks like they _need_ to launch it now to minimize comparison to lunar lake coming out later in the year that's supposed to be a bigger improvement over meteor lake than meteor lake was over alder/raptor lake.
Still it's no slouch and more competition is a good thing.
reply
flat6croc
In the end, it's hard to take reports like this seriously regardless of who put it together. It's paid for promotional material and no different from marketing material released by Microsoft itself. It's not that I imagine the numbers are faked, it's just that the scope for cherry picking is endless and Microsoft ultimately gets to. decide what goes out, they're paying for it. It's ultimately a waste of time and money. It doesn't help that they get the name of the X Elite wrong on literally all the graphs.
reply
In the end, it's hard to take reports like this seriously regardless of who put it together. It's paid for promotional material and no different from marketing material released by Microsoft itself. It's not that I imagine the numbers are faked, it's just that the scope for cherry picking is endless and Microsoft ultimately gets to. decide what goes out, they're paying for it. It's ultimately a waste of time and money. It doesn't help that they get the name of the X Elite wrong on literally all the graphs.
reply
stevencharette7918
well in that case i do not nor would i ever trust an analysis when the people doing it are getting kickbacks and paid , if someone paid you 10 million dollars would yeah be their friend OF COURSE YOU WOULD Gorden, i know you think most of us are dumb but we know these companies like Intel and Microsoft BUY and pay for overrated statistics if you want an honest non bias review of a product get volunteers to do the analysis there are 1000's of people who would review a new product for nothing
reply
well in that case i do not nor would i ever trust an analysis when the people doing it are getting kickbacks and paid , if someone paid you 10 million dollars would yeah be their friend OF COURSE YOU WOULD Gorden, i know you think most of us are dumb but we know these companies like Intel and Microsoft BUY and pay for overrated statistics if you want an honest non bias review of a product get volunteers to do the analysis there are 1000's of people who would review a new product for nothing
reply
MrPtheMan
I hope the do a rerun with Lunar Lake, I think Qualcomm is doing it's best to have strong numbers, but Meteor Lake is no slouch. Lunar Lake is a major improvement. Hopefully we see best of arm vs best x86, including Intel's Lunar Lake as well as AMDs strix halo.
I don't care how much Microsoft wants to push this. I prefer my native apps run its in native architecture. A lot of proapps won't change suddenly to Arm. And if it's buggy, consumers will take the hit.
reply
I hope the do a rerun with Lunar Lake, I think Qualcomm is doing it's best to have strong numbers, but Meteor Lake is no slouch. Lunar Lake is a major improvement. Hopefully we see best of arm vs best x86, including Intel's Lunar Lake as well as AMDs strix halo.
I don't care how much Microsoft wants to push this. I prefer my native apps run its in native architecture. A lot of proapps won't change suddenly to Arm. And if it's buggy, consumers will take the hit.
reply
Filtersloth
It’s no accident that they compared the most powerful of the new snapdragon chipsets to the absolute lowest in the range of M3 chips in a MacBook Air with no fan.
How about we see it compared to a MacBook Pro M3 max and see what the results are like.
They have an MSI prestige there. The comparison is so unbalanced that it actually makes the snapdragon look worse because of it. Compare the binned snapdragon to this binned M3 in fanless MacBook Air.
reply
It’s no accident that they compared the most powerful of the new snapdragon chipsets to the absolute lowest in the range of M3 chips in a MacBook Air with no fan.
How about we see it compared to a MacBook Pro M3 max and see what the results are like.
They have an MSI prestige there. The comparison is so unbalanced that it actually makes the snapdragon look worse because of it. Compare the binned snapdragon to this binned M3 in fanless MacBook Air.
reply
specialresident22
im hoping Linux is better windows 11 has tested my patience. x means close not in windows 10 then 11, show more options to see previously visible things, im ready to move on, been on windows since 1998 and loved all their os except win 8 on desktop and windows 11, Microsoft must be really special to make some of the choices they did in 11. 150 nits, i dont know anyone watching videoa at 150 nits
reply
im hoping Linux is better windows 11 has tested my patience. x means close not in windows 10 then 11, show more options to see previously visible things, im ready to move on, been on windows since 1998 and loved all their os except win 8 on desktop and windows 11, Microsoft must be really special to make some of the choices they did in 11. 150 nits, i dont know anyone watching videoa at 150 nits
reply
vasudevmenon2496
Not sure if nvidia wants to convert any PC into copilot with higher power usage than npu. They should have tested 13instead of 15 which means Microsoft still hasn't cooled it properly yet again.
Amd and Intel have caught up with next gen due release while oryon first gen which is still shipping in volume and yet to reach mainstream availability. Not even sure even next gen oryon comes up.
reply
Not sure if nvidia wants to convert any PC into copilot with higher power usage than npu. They should have tested 13instead of 15 which means Microsoft still hasn't cooled it properly yet again.
Amd and Intel have caught up with next gen due release while oryon first gen which is still shipping in volume and yet to reach mainstream availability. Not even sure even next gen oryon comes up.
reply
David_L3
Thanks for the video. I have a decision to make. I never really considered ARM. I just picked up a 16gig i7 Surface pro 9 on a closeout sale. Maybe I'll return it for a base Surface Pro 11. Sounds like most of what I need is available native. I have other computers for bigger workloads. I guess I gain mostly battery life. Some performance maybe (web browser and Office 365).
reply
Thanks for the video. I have a decision to make. I never really considered ARM. I just picked up a 16gig i7 Surface pro 9 on a closeout sale. Maybe I'll return it for a base Surface Pro 11. Sounds like most of what I need is available native. I have other computers for bigger workloads. I guess I gain mostly battery life. Some performance maybe (web browser and Office 365).
reply
tomoprime217
My only takeaway is the price. The X Elite chip is supposed to max at a $150 cost to make, which should equal a cheaper price overall. I can't imagine a surface laptop having a lower price than before regardless so maybe we should just need to shop the non-microsoft brand which is offering the better bargain for price to performance. Lunar lake won't be this cheap.
reply
My only takeaway is the price. The X Elite chip is supposed to max at a $150 cost to make, which should equal a cheaper price overall. I can't imagine a surface laptop having a lower price than before regardless so maybe we should just need to shop the non-microsoft brand which is offering the better bargain for price to performance. Lunar lake won't be this cheap.
reply
Techno_Rama
Gordon, I hope these paid numbers from signal65 are correct and wish snapdragon x elite success, but what is up with the cloak and dagger style marketing for over 6 months Nobody objective has been given full access to the product yet, even now. If the product is that good just give it to Gordon to let him test it and let's get on with life.
reply
Gordon, I hope these paid numbers from signal65 are correct and wish snapdragon x elite success, but what is up with the cloak and dagger style marketing for over 6 months Nobody objective has been given full access to the product yet, even now. If the product is that good just give it to Gordon to let him test it and let's get on with life.
reply
zotriczaoh7098
My reading of the emulated Blender 4.1.1 render time results is the other way around from Gordon's description. The chart says 'lower is better' which is what I would expect. So in emulation mode the Surface Pro 9 completes the task in 36% of the time it takes the Snapdragon in emulation mode. Is that correct
reply
My reading of the emulated Blender 4.1.1 render time results is the other way around from Gordon's description. The chart says 'lower is better' which is what I would expect. So in emulation mode the Surface Pro 9 completes the task in 36% of the time it takes the Snapdragon in emulation mode. Is that correct
reply
pcworld
I think in addition to saying this was a paid report, it would have fair to introduce who wrote the report and his history and experience over the years in reviewing computer hardware for around 15 years as well as his time working at Intel. I have great faith that Ryan Shrout is fair and unbiased in his findings.
reply
I think in addition to saying this was a paid report, it would have fair to introduce who wrote the report and his history and experience over the years in reviewing computer hardware for around 15 years as well as his time working at Intel. I have great faith that Ryan Shrout is fair and unbiased in his findings.
reply
ShawnLangford
Comparing against the Mac is pointless because regardless of what your charts say Windows users are unlikely to switch to Mac and Mac users are unlikely to switch to Windows. Not to mention the M4 has just been launched. Also, Gordon has a hate on for Apple I can't take anything he says at face value.
reply
Comparing against the Mac is pointless because regardless of what your charts say Windows users are unlikely to switch to Mac and Mac users are unlikely to switch to Windows. Not to mention the M4 has just been launched. Also, Gordon has a hate on for Apple I can't take anything he says at face value.
reply
pcworld
Thanks for the video, but the simple fact that even right before release we have to rely on carefully curated information tells me all I need to know about the platform. If it really were such a step-up, legitimate reviewers would be drowning in pre-production samples to help toot their horn.
reply
Thanks for the video, but the simple fact that even right before release we have to rely on carefully curated information tells me all I need to know about the platform. If it really were such a step-up, legitimate reviewers would be drowning in pre-production samples to help toot their horn.
reply
emrexis
That 3DMark graphic benchmarks were best case scenario with native arm versions right If that is the case then everyday gaming with translation layer would be worse.. I wouldn’t hold my breath for these chips beating AMD and Intel current integrated graphics on AAA and esports games then..
reply
That 3DMark graphic benchmarks were best case scenario with native arm versions right If that is the case then everyday gaming with translation layer would be worse.. I wouldn’t hold my breath for these chips beating AMD and Intel current integrated graphics on AAA and esports games then..
reply
pweddy1
Office 365 support is huge. That 90% of office PCs primary use case. The only problem is IT departments tend to stick with if it ain’t broke don’t fix it, because you have to test every new platform you deploy.
Will look interesting to see the AMD Strix APUs compared to this.
reply
Office 365 support is huge. That 90% of office PCs primary use case. The only problem is IT departments tend to stick with if it ain’t broke don’t fix it, because you have to test every new platform you deploy.
Will look interesting to see the AMD Strix APUs compared to this.
reply
Add a review, comment
Other channel videos
