Straight up, US companies typically give next to zero raises and just work people to burn-out. Your pay erodes to inflation faster than any raises you get. Job hopping is essentially the only way you don't lose money over the long run.
Straight up, US companies typically give next to zero raises and just work people to burn-out. Your pay erodes to inflation faster than any raises you get. Job hopping is essentially the only way you don't lose money over the long run.That's what the Linkedin profile of most top performers looks like in Silicon Valley. European companies are more loyal to employees, so employees are more loyal to companies there. US companies - with few exceptions - have zero loyalty to employees, so people have figured out they need to do what's best for them, not wait for their employer to recognize their contributions and give them a huge raise or big promotion.
I don't see any overblowing. There have been a few waves of departures, first in early to mid-2021, then a year after, and now more recently. You don't have to take my word for it. Look at LinkedIn and search specifically for those who came from Apple. Did thousands of employees leave? Of course not! But how many do they have to begin with? Some reports before the acquisition cited 250 total for NUVIA.The "drain" of Apple's engineering talent to Nuvia was overblown, and now I see people are overblowing the fact that some who left have returned. Losing a top guy like GW III certainly has an impact, but the foot soldiers? There is always turnover, and all the time new people are brought in from elsewhere, or are hired out of grad school (bringing some fresh ideas, or so you hope) Even the loss of your top guys might be a benefit in the long run - it gives an opportunity for others to come up and lead. Maybe they turn out to be better, and had just lacked the opportunity. Even if they aren't necessarily better, they will bring their own ideas and then I guess you find out whether those ideas work out or not when they're put into silicon.
I thought going from ARMv8 (I think this is 8.6?) to v9 would benefit performance. But after seeing this, maybe it’s just not accessible yet? Seems like Nuvia/Qualcomm is doing just fine on the older version!
So Phoenix-L should be close to A17 Pro in "IPC" or am I crazy?
Why would going from ARMv8 to v9 give a performance boost in Geekbench?I thought going from ARMv8 (I think this is 8.6?) to v9 would benefit performance. But after seeing this, maybe it’s just not accessible yet? Seems like Nuvia/Qualcomm is doing just fine on the older version!
Could be that this is a pristine result but tbh I’m skeptical of that.
If so, Qualcomm might be able to claim a Single Core win against Apple. I wasn't expecting this to happen.The more interesting part is that, as per commented on the Apple thread, A18 looks like just a higher clocked A17 Pro
Yea it’ll be a tad less.So Phoenix-L should be close to A17 Pro in "IPC" or am I crazy?
The more interesting part is that, as per commented on the Apple thread, A18 looks like just a higher clocked A17 Pro.
Yea I don’t think it would necessarily?Why would going from ARMv8 to v9 give a performance boost in Geekbench?
That would work just fine in a laptop.
Yeap, 8 Gen 4 would make for some incredible fanless laptops. It would be Qualcomm's Lunar Lake equivalentThat would work just fine in a laptop.
Agreed, and that's why I specified that I work in Europe.That's what the Linkedin profile of most top performers looks like in Silicon Valley. European companies are more loyal to employees, so employees are more loyal to companies there. US companies - with few exceptions - have zero loyalty to employees, so people have figured out they need to do what's best for them, not wait for their employer to recognize their contributions and give them a huge raise or big promotion.
Next gen X Elite should be good.Yeap, 8 Gen 4 would make for some incredible fanless laptops. It would be Qualcomm's Lunar Lake equivalent
Funnily enough, 8 Gen 4 has better CPU Single Core, GPU and NPU performance than X Elite. X Elite's only win is in multi-core performance.
| Geekbench 6.3 Single Core | |
| 8 Gen 4 | 3236 |
| A18 Pro | 3409 |
| Clock speeds | Geekbench6 Multi Core | |
| X1P-46-100 | 8 Phoenix cores @ 3.4 GHz (4.0 GHz Single Core Boost) | 12,000 |
| 8 Gen 4 | 2 Phoenix-L cores @ 4.3 GHz 6 Phoenix-M cores @ 3.5 GHz | 10,000 |
I wouldn’t draw the same conclusion. The gap between the M4 and the M3 in ST is actually more* in GB 5.5 vs 6.3 (Geekerwan slides, Geekbench leaderboard). Based on that I don’t see any reason why that 5% advantage for the A18 doesn’t remain in GB 5.5.
Geekbench 6.3 Single Core 8 Gen 4 3236 A18 Pro 3409
A18 is faster, but it also has SME. So if we remove SME out of the equation by running Geekbench v6.2 or v5... 8G4 and A18 could be essentially tied in single core performance.
I think this might be the method Qualcomm would use to claim victory, and tout that with Oryon, they have finally managed to eliminate Apple's historical advantage in Single Core. Performance leadership has returned to the Android ecosystem!
Another speculation without evidence Is there any real proof of this or maybe you work for the media that has been predicting and predicting a ‘collapse’ for Apple's silicon team? Instead, for some reason, everyone is silent that Apple hired a top architect from Nvidia a few years ago and now he is responsible for GPU architecture, and at least there is evidence for this information.I don't see any overblowing. There have been a few waves of departures, first in early to mid-2021, then a year after, and now more recently. You don't have to take my word for it. Look at LinkedIn and search specifically for those who came from Apple. Did thousands of employees leave? Of course not! But how many do they have to begin with? Some reports before the acquisition cited 250 total for NUVIA.
I would not downplay the role of a skilled team. Sure, having someone who can drive good designs on many fronts is way more important, but remember, somebody has to make those ideas happen. When you lose weeks because your not-so-good RTL engineer can't get to meet specifications, it makes a difference.
Of course, new blood is always welcome. It is just not a good sign if you are onboarding people worse than those who have left.
Considering the frequencies at which 8 gen 4 cores operate, I wouldn't be so excited about it. The power efficiency of the X Elite is already in question. And it's obvious from the scores that Oryon lags behind Apple in IPC.
Geekbench 6.3 Single Core 8 Gen 4 3236 A18 Pro 3409
A18 is faster, but it also has SME. So if we remove SME out of the equation by running Geekbench v6.2 or v5... 8G4 and A18 could be essentially tied in single core performance.
I think this might be the method Qualcomm would use to claim victory, and tout that with Oryon, they have finally managed to eliminate Apple's historical advantage in Single Core. Performance leadership has returned to the Android ecosystem!
This is not the first time I have come to that conclusion. I posted this a while ago;Phoenix-M is a different uarch.
That lines up with the math I did 8 months ago:
I got Phoenix-L = 1950 and Phoenix-M = 850 points, which means again that Phoenix-M has a little less than half the performance of Phoenix-L.
Could Phoenix-M be a different uarch after all?
Core Clock speed GB6
1TPerformance
Per ClockPhoenix-L 4.37 GHz 3250 743.707 Phoenix-M 3.0 GHz 1400 466.666
Indeed, efficiency of 8 Gen 4 remains to be seen. Some people are whispering that 8 Gen 4 could be the return of the Snapdragon 810.Considering the frequencies at which 8 gen 4 cores operate, I wouldn't be so excited about it. The power efficiency of the X Elite is already in question.
Isn't X Elite based on TSMC 4nm while 8 Gen 4 on 3nm? Power efficiency should be improved on some degree.Considering the frequencies at which 8 gen 4 cores operate, I wouldn't be so excited about it. The power efficiency of the X Elite is already in question. And it's obvious from the scores that Oryon lags behind Apple in IPC.