Discussion Qualcomm Snapdragon Thread

Page 158 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
Snapdragon 8 Elite
2 × Phoenix-L @ 4.32 GHz [12 MB sL2]
6 × Phoenix-M @ 3.53 GHz [12 MB sL2]

Snapdragon 7 Gen 4 (Speculation)
2 × Phoenix-L @ 3.9 GHz [8 MB sL2]
2 × Phoenix-M @ 3.2 GHz [4 MB sL2]

Will a quad core CPU be unsuitable for Android today?

Even $150 phones have '8-core CPU' nowadays.
 

SteinFG

Senior member
Dec 29, 2021
664
786
106
ARM really went with a nuclear option, huh. I can see their reasoning: Arm is stuck in court for over a year already, and Qualcomm gets good in-house core, intends for this momentum to carry forward. So arm cancells the agreement at the point where oryon future is bright but slightly uncertain, best time to negotiate high % fees. ARM is betting that qualcomm doesn't want their progress to stop and they'll pay extra.
 
Reactions: jdubs03

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
SPEC numbers extracted from Geekerwan's graphs.
A18-PA18-EOryon-LOryon-M
Clock speed4.04 GHz2.2 GHz4.32 GHz3.53 GHz
SPEC INT10.73.38.95.2
SPEC FP16.05.014.08.0
Core size3.0 mm²0.8 mm²2.2 mm²0.9 mm²

Performance wise (rough estimation);

Oryon-L : Oryon-M
100% : 60%

A18-P : A18-E
100% : 33%

Core size wise ;

Oryon-L : Oryon-M
100% : 40%

A18-P : A18-E
100% : 27%
 
Reactions: Raqia

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
It's been an eventful month for ARM and Qualcomm;

- Mediatek Dimensity 9400
- Qualcomm cancels Snapdragon Dev Kit
- Qualcomm Snapdragon 8 Elite, Snapdragon Ride Elite and Cockpit Elite with 2nd gen Oryon cores
- ARM sends notice about revoking Qualcomm's ALA
- Massive Google Tensor G4/G5 leak
- Apple M4 Pro and M4 Max unveil (October 30th?).
 

jdubs03

Senior member
Oct 1, 2013
951
594
136
It's been an eventful month for ARM and Qualcomm;

- Mediatek Dimensity 9400
- Qualcomm cancels Snapdragon Dev Kit
- Qualcomm Snapdragon 8 Elite, Snapdragon Ride Elite and Cockpit Elite with 2nd gen Oryon cores
- ARM sends notice about revoking Qualcomm's ALA
- Massive Google Tensor G4/G5 leak
- Apple M4 Pro and M4 Max unveil (October 30th?).
I’d say the 8 Elite and the Summit as a whole was a bright spot in all of this.
But the rest? Ranges from meh to not good.
 

MS_AT

Senior member
Jul 15, 2024
311
691
96
SPEC numbers extracted from Geekerwan's graphs.
A18-PA18-EOryon-LOryon-M
Clock speed4.04 GHz2.2 GHz4.32 GHz3.53 GHz
SPEC INT10.73.38.95.2
SPEC FP16.05.014.08.0
Core size3.0 mm²0.8 mm²2.2 mm²0.9 mm²

Performance wise (rough estimation);

Oryon-L : Oryon-M
100% : 60%

A18-P : A18-E
100% : 33%

Core size wise ;

Oryon-L : Oryon-M
100% : 40%

A18-P : A18-E
100% : 27%
Do you have compiler settings that Geekerwan used? For quoting SPEC results they are essential, I mean even Qualcomm itself claims this on their slides when pointing fingers at Intel
 

Cardyak

Member
Sep 12, 2018
74
164
106
Comparison of microarchitectures.

A18 P-core and Oryon-L

View attachment 110078

A18 E-core & Oryon-M

View attachment 110080
These diagrams have a few mistakes (but the overall picture is broadly correct). However: The M4 Small core does not feature 192KB L1 Instruction cache, it's 128KB. Also the uops count below the "Map and Rename" sections artificially inflate the scheduling bandwidth. For example the A18/M4 P Core cannot dispatch 10uops to each individual dispatch buffer simultaneously, there are limitations within both the allocation stage, and also the number of uops each scheduler can accept per cycle. Theres also something strange happening with the ST queue on the Oryon M core, Qualcomm themselves have stated that the Oryon L core has a ST Queue length of 56, and I'm skeptical that the smaller core will have a larger buffer. I think there's some interference here from non-scheduling queues and it's affecting the measurements.

For what it's worth my diagrams can be located here: http://bit.ly/32qLLew

There's inevitably some mistakes on these also, but over time as more testing is conducted we can hopefully collate together all of this information and present a more accurate picture.

 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
Snapdragon Summit 2024

Day 1 Keynote (Mobile)


Day 2 Keynote (Automotive)


Day 1 Deep Dive (Mobile)


Day 2 Deep Dive (Automotive)

 
Last edited:
Reactions: poke01

jdubs03

Senior member
Oct 1, 2013
951
594
136
Do you have compiler settings that Geekerwan used? For quoting SPEC results they are essential, I mean even Qualcomm itself claims this on their slides when pointing fingers at Intel
This should help (top right).
 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106

The Hexagon NPU of 8 Elite adds two more Scalar cores and 2 more Vector cores.

They say it's "45% faster". If that means 45% more INT8 TOPS, then we are going from 45 TOPS (8G3) to 65 TOPS (8 Elite).



"6nm single-chip"

@Ghostsonplanets This confirms that the Fastconnect 7900 isn't integrated into the 8 Elite SoC (which is made on 3nm), but that it's a discrete chip.

Also "AI Enhanced Wifi". ROFLed at that one.
 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
He used to add this information in the past. But glancing on his 8 Elite video I haven't seen it so the question is it still the same. He might have changed things in between, he might have not.
Andrei's comment about this matter;
he's using NDK binaries on SPEC for Android which will have an inherent handicap vs iOS shared runtime libraries
it's fine given that this represents the userspace experiences between the OS', however if you would really want to look at just µarch you'd deploy glibc+jemalloc binaries on Android get get somewhat of a similar allocator behavior to what iOS does in which case the competitive performance differences here are going to be fundamentally smaller

geekbench difference is smaller because of the way it's built counteracts some of these differences at the moment, and actually if you run 6.2 (non-SME) that's probably as close as you can reasonable get for a 1:1 µarch comparison
Posted in Chips&Chees Discord.
 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
But I guess no one in android space, due to marketing reasons, will dare to release a SoC with less than 8 CPU cores. Even Qualcomm had to spam M cores to keep the magical 8 number going
I wouldn't call it spamming. 2L+6M is a perfectly fine configuration for 8 Elite.

That's why I am wondering what they will do for lower tier chips?

If they are going Oryon L/M, then they'll have to put <8 cores.
Snapdragon 8 Elite
2 × Phoenix-L @ 4.32 GHz [12 MB sL2]
6 × Phoenix-M @ 3.53 GHz [12 MB sL2]

Snapdragon 7 Gen 4 (Speculation)
2 × Phoenix-L @ 3.9 GHz [8 MB sL2]
2 × Phoenix-M @ 3.2 GHz [4 MB sL2]

Will a quad core CPU be unsuitable for Android today?

Even $150 phones have '8-core CPU' nowadays.
A wild bit of speculation is that Qualcomm has an Oryon-S core in the works.

If L = Large and M = Medium, doesn't that suggest that there will be a "Small" core?

Cortex X ≈ Oryon-L
Cortex A7xx ≈ Oryon-M
Cortex A5xx ≈ Oryon-S

Oryon-S would allow Qualcomm to scale the Oryon CPU all the way down to low end Snapdragon 4 series chips and Snapdragon Wearable chips.
 

MS_AT

Senior member
Jul 15, 2024
311
691
96
Andrei's comment about this matter;

Posted in Chips&Chees Discord.
What Andrei said is raising an interesting dillema does the end user care about uarch comparison or the actual performance in context of Geekbench version (6.2 vs 6.3). I mean I understand that introducing SME will have less impact in android ecosystem due to fragmentation, but I guess that vertically integrated Apple is already making use of it due to having much better control over software stack.

At least this gives some context to Qualcomm slides where 6.2 was used, as from uarch vs uarch it's more fair seeing SME unlike Neon is not part of the core.

On the other hand if Andrei wanted to do accurate uarch vs uarch comparisons he should have enabled AVX for x64 when doing comparisons on Qualcomm slides as the AVX units belong to the core the same Neon units do (I guess he is somewhat involved with those if the rumours are accurate). But yea, that is a digression
 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
A wild bit of speculation is that Qualcomm has an Oryon-S core in the works.

If L = Large and M = Medium, doesn't that suggest that there will be a "Small" core?

Cortex X ≈ Oryon-L
Cortex A7xx ≈ Oryon-M
Cortex A5xx ≈ Oryon-S

Oryon-S would allow Qualcomm to scale the Oryon CPU all the way down to low end Snapdragon 4 series chips and Snapdragon Wearable chips.
Oryon-S doesn't even need to be an entirely new microarchitecture. To create Oryon-S, they can just take Oryon-M, tweak it a bit and massively reduce the clock speed to about 2 GHz (Oryon-M in 8 Elite runs at a crazy 3.53 GHz).
 

jdubs03

Senior member
Oct 1, 2013
951
594
136
He used to add this information in the past. But glancing on his 8 Elite video I haven't seen it so the question is it still the same. He might have changed things in between, he might have not.
Yea I hear ya. The A18 Pro numbers match so my assumption is it’s still the same.
 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
There was also this interesting slide from the Mobile Deep Dive;

128 KB L1 = 1 ns
12 MB L2 = 5ns

Can someone convert nanoseconds to clock cycles?
 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
4 cycle L1
~21 cycle L2
So L1 latency is same as Oryon core in X Elite, whereas L2 latency regressed by a few cycles.

Almost directly explained by the fact that Oryon -> Oryon-L, L1 capacity regressed from 192 KB -> 128 KB, but L2 capacity stayed the same at 12 MB.
 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
It would be a great feat if they can hit 5 GHz next year with Oryon Gen 3, without regressing the latency.
 

FlameTail

Diamond Member
Dec 15, 2021
4,097
2,469
106
View attachment 110062
It's interesting that Qualcomm has changed the rendering technique from Tile-Based Rendering (TBR) to something more akin to the Immediate Mode Rendering (IMR) used in desktop class GPUs. It's fed by the huge 12 MB of GPU cache. This signals that Qualcomm is evolving Adreno to become a desktop class GPU architecture.

But Adreno 830 wouldn't be the first smartphone GPU without TBR. Exynos 2200 with it's AMD mRDNA2 GPU demonstrated that it's possible for a smartphone to have a non-TBR GPU.
Does Adreno 830 exclusively use Binned Direct mode?

Vince suggests that Adreno 830 could be using TBIM (Tile Based Immediate Mode Rendering), which is a hybrid of the tradional mobile TBDR (Tile Based Deferred Rendering) and traditional desktop IMR (Immediate Mode Rendering).

 
Reactions: Ghostsonplanets
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/    |