Discussion Qualcomm Snapdragon Thread

Page 40 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

SpudLobby

Senior member
May 18, 2022
991
684
106
The 8 core Purwa part sound like it's perfect for Chromebooks, in addition to <$800 Windows Laptops
Yeah agree except don’t expect it in that many like sub-600 laptops, they shouldn’t bother with genuine crap.

But competing with AMD, who’s mainstream Phoenix successor is in a way the 4+4 Zen 5 Kraken (which is around the size of Phoenix — and Phoenix is 178mm^2, while the 4+8 Strix is 220-230mm^2) is then possible on cost.

Getting some $700-1300 laptops with Purwa would be huge.
 

SpudLobby

Senior member
May 18, 2022
991
684
106
If Purwa has only half the iGPU of the X Elite, it will be too weak (vs competition).
View attachment 97330
Half the iGPU of X Elite means 22 FPS, which is even less than the OG Apple M1.

I doubt it will be able to keep up with 8 CU RFNA3.5 of Kraken Point or 8 Xe2 ARC GPU of Lunar Lake.
Hmmm yeah, probably true. They might cut it down in other ways, e.g. transistor layout re: clocks. But I don’t think most people are going to game at 30FPS with this anyways.

They could reduce perf down by 20-30% and be fine probably.
 

SpudLobby

Senior member
May 18, 2022
991
684
106
I dont like this. I want it to have the same fmax as X Elite. That's one good thing about Apple Silicon- whether it's the base M chipvor the top tier M Ultra, the P-core runs at the same clock speed**

That and the 4P+4E, i think they could get away with halving the memory bus.
Dude no

If it's a 4P+4E CPU, I see no reason to cut down the L2 in half. In X Elite, a 4P cluster has 12 MB L2. Keep it as it is. For the 4E cluster, it might be 4 or 6 MB of L2.

**I am aware that there was a slight clock speed difference between the base M chip and M Pro/M Max in the M1 and M2 generations. However, in the M3 generation they are all running at 4.05 GHz
Agree but E Cluster needs 12MB in this case still. Would hurt efficiency otherwise.
I dont like this. I want it to have the same fmax as X Elite. That's one good thing about Apple Silicon- whether it's the base M chipvor the top tier M Ultra, the P-core runs at the same clock speed**
I mean I am pretty sure we already know the X Plus doesn’t have the same fmax at least functionally, like with Adreno for ex.

I agree but within some parameters it’s fine, the issue with Intel and AMD is doing segmentation beyond belief and weird branding for SKUs.
 

SpudLobby

Senior member
May 18, 2022
991
684
106

IF this is true, which I totally doubt, I expect it to be a bit more about Arm’s architecture in the Cortex X5 and A730 than the process node of Samsung’s SF3 (2nd gen 3NM) which probably still struggles by most accounts. Even with that struggle it’ll be better than like their 2nd and 3rd gen 4NM on power/performance I bet and probably closer to N4P if not ahead (possibly closer than expected to N3E)

But the issue is it really should be ahead of TSMC’s N3E, the density is similar but they’re using superior transistors, which alone are worth significant upgrades in terms of leakage, Vmin, and drive current, provided similar yields.

Ideal scenario is that the process ends up being as good as it should be for something as dense as N3 but with GAAFET, and Arm’s new architecture is also good as is Qualcomm’s but Samsung’s node gives them an edge.

I just doubt this is what will happen though
 

SpudLobby

Senior member
May 18, 2022
991
684
106
We know N3E is great.

My bar for the 8 Gen 4 being good is + 20-25% more ST performance than the 8 Gen 3, also at 10-15% lower power. For SpecInt/FP and possibly GB6.

Would like to see them do a 2700-2900 GB6 at ~ like 4-5W, 5W being the top end here and ideally just something that can finally match or beat Everest in the A16 on perf/W.

For the littles, beating those 8 Gen 2 A715 1-1.2W results in Spec by 30+% around the same power draw seems reasonable tbh.
 

gdansk

Diamond Member
Feb 8, 2011
3,276
5,186
136
You want them to tape out 3 different (N3?) chips for a small market?
That is crazy speculation.
 

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,594
106
An 45 TOPS NPU is so big on TSMC N3B I don't think they can scale it so easy into 100-400 TOPS, not realistic at all

A17 Pro - 34 TOPS NPU - 5 mm²

This is the Snapdragon 8 Gen 2. We don't know the TOPS value, but the NPU is certainly tiny.

So, yeah I was conflicted as to whether the all 5 SKUs should have the same NPU performance, or whether NPU performance should scale up from the lowest to highest end SKU.
 

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,594
106
You want them to tape out 3 different (N3?) chips for a small market?
That is crazy speculation.
If Semiaccurate is right, they are already taping out two dies with the first generation- Hamoa and Purwa.

So what I did was add another die at the high end. This is essentially a Strix Halo-esque part, and is intended to compete with Intel/AMD's HX series laptop parts in terms of CPU performance.

If you think about it, the Snapdragon X Elite with it's 12P cores is barely strong enough to do so.
 

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,594
106
I wouldn't trust that site.
And here's the key point:
As is usual for Qualcomm, the company doesn’t actually state the per-block performance increases, but instead opts to showcase an aggregate computational throughput figure shared amongst all of the SoC’s IP blocks, including CPU, GPU, and the new Hexagon accelerator block. This new figure lands in at 26TOPs for the Snapdragon 888, which is 73% higher than the 15TOPs figure of the Snapdragon 865.
 

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,594
106

That 4 MB of graphics memory is interesting. I believe it is 4 MB of GEMM, not 4 MB of L2.

Chips&Cheese found that the Snapdragon 8+ Gen 1's iGPU has 2 MB of GEMM and 256 KB of L2.


1024 ALUs corresponds to 2 MB of GEMM.

Considering that X Elite is using the same Adreno 7xx architecture as 8 Gen 1, we can estimate that it has 2048 ALUs, corresponding to 4 MB of GEMM.

8 Gen 1: Adreno 730 : 1024 ALUs : 2 MB GEMM
8 Gen 2 : Adreno 740 : 1536 ALUs : 3 MB GEMM (?)
8 Gen 3 : Adreno 750 : 1792 ALUs : 3.5 MB GEMM (?)
X Elite : Adreno 7xx : 2048 ALUs (?) : 4 MB GEMM (?)

Of course this estimate could be entirely incorrect, because the 4 MB figure comes from the above mentioned leaker, and I assumed it was for solely the GEMM. (Perhaps it could be the sum of GEMM + L2).
 

StinkyPinky

Diamond Member
Jul 6, 2002
6,900
1,131
126
I really like the Samsung laptops, they are well made and designed and have great displays. Problem is so many of their models top out at 16GB ram. I'll be watching out for that Samsung model with interest.
 

FlameTail

Diamond Member
Dec 15, 2021
4,238
2,594
106
Interesting thing about the 2nd tweet is that Ideapad is Lenovo line for mainstream <$1000 laptops.
Looks like Lenovo is going all in.

No wonder Qualcomm gave Lenovo's CEO a prominent position during the Snapdragon X Elite announcement in last October.
I suppose these will be using Purwa with 8 or 6 cores option
Idk if Purwa will have a cut down 6-core part. May not happen, especially if it is a 4P+4E setup.
 

Ghostsonplanets

Senior member
Mar 1, 2024
701
1,122
96
Idk if Purwa will have a cut down 6-core part. May not happen, especially if it is a 4P+4E setup.
If Purwa is a separate, 4+4 die, I would be very surprised if QCOM achieved perfect yields or didn't had a binned SKU. 6C is still a perfectly valid option for low-end. Specially when competition (Intel) sells 2P + 4E. 6P would be a literal erase on Intel offerings.
 
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/    |