Discussion Qualcomm Snapdragon Thread

Page 141 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

mvprod123

Member
Jun 22, 2024
186
198
76
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.

Do note that while X Elite is on N4, 8 Gen 4 is N3E.
X Elite - N4P. The difference between N4P and N3E is not significant.
 

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,592
106
X Elite - N4P. The difference between N4P and N3E is not significant.

Versus N5,

N4P
-22% power
+11% performance

N3E
-34% power
+18% performance

I wouldn't say that's a negligible difference, in this day and age where Moore's Law is dead.
 
Reactions: ikjadoon

hemedans

Senior member
Jan 31, 2015
244
125
116
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.
Soc can consume more power but doesn't necessarily means has bad efficiency, plenty of reports that 8 gen 4 will be more efficient than 8 gen 3. Also soc can be efficient but with worse battery life.

The big task for 8 gen 4 is how much power Efficienct core use, if Phoenix-M is improvement over Arm Efficient cores then it doesn't matter how much power those L cores use, Apple soc are superior because of E cores.
 

MarkizSchnitzel

Senior member
Nov 10, 2013
452
74
91
Are they ACTUALLY chasing 1T crown?
Who are they marketing this to, to the OEMs?
Because normal people have not much complaints in regard to performance on flagships in the last few years, battery life is always a pain, or, higher on the list of things to improve.

I guess now with Oryon being in laptops, I kinda get it, but still.
 

mvprod123

Member
Jun 22, 2024
186
198
76
Soc can consume more power but doesn't necessarily means has bad efficiency, plenty of reports that 8 gen 4 will be more efficient than 8 gen 3. Also soc can be efficient but with worse battery life.

The big task for 8 gen 4 is how much power Efficienct core use, if Phoenix-M is improvement over Arm Efficient cores then it doesn't matter how much power those L cores use, Apple soc are superior because of E cores.
8 Gen 4 will consume more power than Gen 3. There are numerous hints and rumours that the next-generation Android flagships will switch to even more capacious batteries (5500+).
 

hemedans

Senior member
Jan 31, 2015
244
125
116
8 Gen 4 will consume more power than Gen 3. There are numerous hints and rumours that the next-generation Android flagships will switch to even more capacious batteries (5500+).
Consuming more power doesn't mean less efficienct as long as perfomance increase more.

About battery life we have to wait.
 

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,592
106
Snapdragon 8 Gen 4 has been spotted in Geekbench Browser, confirming the authenticity of the image shared by DCS;


@ikjadoon Could you do an analysis of the subtest scores, and how it compares to A18 Pro?
 

MS_AT

Senior member
Jul 15, 2024
365
798
96
Snapdragon 8 Gen 4 has been spotted in Geekbench Browser, confirming the authenticity of the image shared by DCS;
It can be there, but the score is not so meaningful if we don't know the power envelope needed to achieve it. OnePlus was known for using special power modes for GeekBench to boost perf in benchmarks and the test platform itself doesn't need to be in sealed phone factor so it might have additional cooling. Just to keep people expectations in check before we go over yet another hype cycle
 

Doug S

Platinum Member
Feb 8, 2020
2,888
4,911
136
8 Gen 4 will consume more power than Gen 3. There are numerous hints and rumours that the next-generation Android flagships will switch to even more capacious batteries (5500+).

If it consumes more power and needs bigger batteries it will also generate more heat. Nowhere for the heat in a phone to go but your hand.
 

ikjadoon

Senior member
Sep 4, 2006
235
513
146
Snapdragon 8 Gen 4 has been spotted in Geekbench Browser, confirming the authenticity of the image shared by DCS;


@ikjadoon Could you do an analysis of the subtest scores, and how it compares to A18 Pro?

That's a fantastic score and, from some GB6 estimates by Arm, X925 SoCs will be notably behind. I won't know much beyond you or others, though.

https://browser.geekbench.com/v6/cpu/compare/7714502?baseline=7720191 (vs the highest A18 score so far)

If we want to isolate without SME (which is another discussion: how frequently is it used in Android in these workloads?), a GB6.2 run on them both would be ideal.

// looking at 1T

The best I can note is that the Pts / GHz is much closer to what Qualcomm initially claimed (3227 / 4.2 GHz = ~768 Pts / GHz in for the X Elite under Linux; this score is 3216 / 4.3 GHz = ~748 Pts / GHz; X Elite on Windows is closer to ~711 Pts / GHz) from Oryon. I assume there's something unique in the Oryon uArch that lets it run noticeably better under Linux systems (e.g., some of the same branch prediction stuff we saw with AMD?).

Maybe Arm X925 smartphones will be only ones below 4 GHz next year. A18 / M4 is likely to keep the overall Pts / GHz "crown" here, though the X Elite as the clocked +6% higher.

In 1-2 subtests, though, Oryon has a slightly higher Pts / GHz (!): see PDF renderer & HDR.

A18: 3261 pts @ 4.04 GHz → 807.2 Pts / GHz
Oryon: 3644 pts @ 4.3 GHz → 847.4 Pts / GHz (+5%)

HDR is 1% advantage for Oryon, as well. Notably, neither test uses SME.

// looking at nT

This is more a PL1 / burst power limit test, but Oryon is notably faster over the A18 (though it's +20% with ~33% more cores).

//

It can be there, but the score is not so meaningful if we don't know the power envelope needed to achieve it. OnePlus was known for using special power modes for GeekBench to boost perf in benchmarks and the test platform itself doesn't need to be in sealed phone factor so it might have additional cooling. Just to keep people expectations in check before we go over yet another hype cycle

That is possible. IIRC, the OnePlus & Samsung schemes were to throttle non-benchmark apps; technically, the benchmarks showed the SoCs at their best: they just refused to let users actually use that power in anything else. So benchmarks were ran faster than real apps (thus cheating), but by slowing down real apps, versus boosting benchmarks.

Tactic 1: boost benchmarks, normal perf in real apps
Tactic 2: normal perf in benchmarks, throttle real apps

I believe it was Tactic 2.
 

Gerard Williams

Junior Member
Apr 1, 2014
10
70
91
That's a fantastic score and, from some GB6 estimates by Arm, X925 SoCs will be notably behind. I won't know much beyond you or others, though.

https://browser.geekbench.com/v6/cpu/compare/7714502?baseline=7720191 (vs the highest A18 score so far)

If we want to isolate without SME (which is another discussion: how frequently is it used in Android in these workloads?), a GB6.2 run on them both would be ideal.

// looking at 1T

The best I can note is that the Pts / GHz is much closer to what Qualcomm initially claimed (3227 / 4.2 GHz = ~768 Pts / GHz in for the X Elite under Linux; this score is 3216 / 4.3 GHz = ~748 Pts / GHz; X Elite on Windows is closer to ~711 Pts / GHz) from Oryon. I assume there's something unique in the Oryon uArch that lets it run noticeably better under Linux systems (e.g., some of the same branch prediction stuff we saw with AMD?).

Maybe Arm X925 smartphones will be only ones below 4 GHz next year. A18 / M4 is likely to keep the overall Pts / GHz "crown" here, though the X Elite as the clocked +6% higher.

In 1-2 subtests, though, Oryon has a slightly higher Pts / GHz (!): see PDF renderer & HDR.

A18: 3261 pts @ 4.04 GHz → 807.2 Pts / GHz
Oryon: 3644 pts @ 4.3 GHz → 847.4 Pts / GHz (+5%)

HDR is 1% advantage for Oryon, as well. Notably, neither test uses SME.

// looking at nT

This is more a PL1 / burst power limit test, but Oryon is notably faster over the A18 (though it's +20% with ~33% more cores).

//



That is possible. IIRC, the OnePlus & Samsung schemes were to throttle non-benchmark apps; technically, the benchmarks showed the SoCs at their best: they just refused to let users actually use that power in anything else. So benchmarks were ran faster than real apps (thus cheating), but by slowing down real apps, versus boosting benchmarks.

Tactic 1: boost benchmarks, normal perf in real apps
Tactic 2: normal perf in benchmarks, throttle real apps

I believe it was Tactic 2.
Fun times ahead !!!
 

POWER4

Member
May 25, 2024
54
28
51
Another speculation without evidence
Well, I specifically gave the source of the information. I will also say that the 250 figure came from an article in the Mayfield's Fund website news section. Is it evidence enough?

You don't have to take my word, but I am also not obligated to do your "homework".
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
I don't know anything about that. You probably did not notice, but I was arguing on the same side as yours, i.e., Apple reclaimed talent they once had lost.

Go easy with the "I must defend Apple" attitude. Sometimes, people are not criticizing Apple.
and at least there is evidence for this information.
Funnily, you (like me) did not care about bringing it here, but unlike me, you haven't even cited the source.

Anyway, this is going too off-topic. I'm stopping here.
 
Reactions: mvprod123

Tup3x

Golden Member
Dec 31, 2016
1,117
1,146
136
That is possible. IIRC, the OnePlus & Samsung schemes were to throttle non-benchmark apps; technically, the benchmarks showed the SoCs at their best: they just refused to let users actually use that power in anything else. So benchmarks were ran faster than real apps (thus cheating), but by slowing down real apps, versus boosting benchmarks.

Tactic 1: boost benchmarks, normal perf in real apps
Tactic 2: normal perf in benchmarks, throttle real apps

I believe it was Tactic 2.
Normally OnePlus have had piss poor Geekbench scores, especially for single threaded results.
 

MarkizSchnitzel

Senior member
Nov 10, 2013
452
74
91
OnePlus uses tactic 2.
They throttle Chrome to oblivion. Not Chrome beta though.

So it's more to give better battery life.

Terrible practice.
 

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,592
106
A redditor's experience regarding how their business has switched to Snapdragon laptops;
Very substantial quantities in case of my one big org I can speak to. We have always used Dell's x86 business laptops, and as of a couple of months ago all new work laptops are ARM-based unless you have a specific need for an x86 model and request a replacement. You then get an older but chunkier higher-end model.

No big deal was made about the shift. It was very matter of fact "the sequels to the Dell business laptops we always used now use ARM chips, which have a much better battery life but may still have some software compatibility issues" kind of nonchalant thing.

My Team uses these for standard remote office work, and those I talked to feel they're cooler, quieter, and last far longer on battery for them. And that all you need to power them is a standard iPhone charger. Which are exclusively perks as far as they're concerned. Otherwise, the experience is seamless.

Thus my experience of a shift. I wouldn't use these as my main personal daily driver yet due to the software incompatibility, as you start running into issues the moment you start looking into games and less common software. But they actually now seem to be the better Outlook+Office+Zoom+Browser standard office work devices. And for the first time I can imagine corporate shifting to these just as their usual vendors (Dell/HP/Lenovo) are.
Anecdotal evidence though, not statistical.
 
Reactions: rosetta and Racan

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,592
106
Negativeonehero:
One year into 3nm and every mobile chip maker is already testing the limits lol (I heard that 8G4 clock may still not be finalized, what the heck)
Now power efficiency is more relevant than ever
Truly.

Dimensity 9400 : 3.63 GHz
A18 Pro : 4.04 GHz
8 Gen 4 : 4.32 GHz

Never did I dream there would phone CPUs running at 4 GHz, and so soon too.

The 8 Gen 4 in particular seems to be a speed demon. It's "E-cores" run at 3.53 GHz, almost the same as the 3.63 GHz peak clock of the Dimensity 9400.

Efficiency is the big question now.

All 3 chips are on TSMC N3E node, so it should allow for some nice iso node comparisons.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |