Tuesday, May 31, 2022
HomeComputer HardwareTransferring to TSMC for Extra Velocity, Decrease Energy

Transferring to TSMC for Extra Velocity, Decrease Energy


Because the darkish of the night time rolls into China this night, Qualcomm is internet hosting a mobile-focused product launch occasion they’re calling “Snapdragon Night time”. Headlining the occasion is the announcement of the corporate’s new flagship SoC, the Snapdragon 8+ Gen 1. A mid-generation replace to their flagship smartphone SoC, the Snapdragon 8 Gen 1, the 8+ Gen 1 follows Qualcomm’s annual custom of releasing a refresh product to spice up efficiency and to offer companions one thing new to work with for the second half of the yr. And for this yr specifically, we’re a really notable change in chips from Qualcomm.

In contrast to earlier generations the place Qualcomm merely launched a quicker pace bin of their present silicon, for 2022 now we have one thing extra substantial to speak about. Qualcomm has switched up foundries fully – transferring from Samsung to TSMC – and in consequence is rolling out a brand new die. Because of this, the Snapdragon 8+ Gen 1 Qualcomm is reaping one thing of a one-off manufacturing achieve, permitting them to each dial up CPU and GPU efficiency whereas concurrently slicing energy consumption.

Qualcomm Snapdragon 8 Gen 1 Flagship SoCs
SoC Snapdragon 8+ Gen 1 Snapdragon 8 Gen 1
CPU 1x Cortex-X2
@ 3.2GHz

3x Cortex-A710
@ 2.8GHz

4x Cortex-A510
@ 2.0GHz

6MB sL3

1x Cortex-X2
@ 3.0GHz

3x Cortex-A710
@ 2.5GHz

4x Cortex-A510
@ 1.8GHz

6MB sL3

GPU Adreno
(10% Increased Clockspeed)
Adreno
DSP / NPU Hexagon Hexagon
Reminiscence
Controller
4x 16-bit CH

@ 3200MHz LPDDR5  /  51.2GB/s

4MB system stage cache

ISP/Digicam Triple 18-bit Spectra ISP

1x 200MP or 108MP with ZSL
or
64+36MP with ZSL
or
3x 36MP with ZSL

8K HDR video & 64MP burst seize

Encode/
Decode
8K30 / 4K120 10-bit H.265

Dolby Imaginative and prescient, HDR10+, HDR10, HLG

720p960 infinite recording

Built-in Modem X65 built-in

(5G NR Sub-6 + mmWave)

DL = 10000 Mbps

UL = 3000 Mbps

Mfc. Course of TSMC 4nm Samsung 4nm

Rapidly diving into the specs, the brand new Snapdragon 8+ Gen 1 is actually the unique Snapdragon 8 Gen 1 ported over from Samsung’s 4nm line over to one among TSMC’s 4nm strains. Underneath extra regular circumstances, this sort of a shift would probably be unremarkable – or at most, an amusing train in on the lookout for edge instances – nonetheless for Qualcomm’s flagship SoC, the matter is extra vital.

Whereas official sources and statements on the standard of Samsung’s 4nm course of are few and much between, unofficially, it’s develop into clear that Samsung’s 4nm course of hasn’t lived as much as expectations. This has triggered a cascading impression on the chips made on the method node, resulting in the unique Snapdragon 8 Gen 1 growing an affinity for energy consumption, and Samsung’s personal Exynos 2200 not faring any higher. Conversely, by all accounts TSMC’s N4 course of is trying nice, with the optically shrunk node constructing off of TSMC’s already profitable and really performant 5nm applied sciences.

Because of this efficiency hole between Samsung and TSMC’s 4nm nodes, Qualcomm is taking the weird step of (basically) porting their high-end SoC over to TSMC’s fab. Which, though not strictly obligatory – Qualcomm carries a number of momentum and the 8 Gen 1 has been promoting properly – is actually a prudent transfer for the corporate. Qualcomm is going through particularly stiff competitors this technology from MediaTek, whose flagship-level Dimensity 9000 SoC was the lead product for TSMC’s 4nm node. And that leaves MediaTek with a definite benefit in opposition to the unique 8 Gen 1, one which Qualcomm could be very joyful to nullify.

In the end, the change in fabs is giving Qualcomm an opportunity to enhance upon the unique 8 Gen 1 from each ends of the spectrum, ensuing within the Snapdragon 8+ Gen 1. On the efficiency entrance, TSMC’s node affords them a simple alternative to extend CPU and GPU clockspeeds for extra efficiency. The prime Cortex-X2 core is now clocked 7% increased, at 3.2GHz, and in the meantime the A710 and A510 clusters have seen their clockspeeds dialed up much more considerably, by round 12% every. Now even the slowest A510 cores can run at 2GHz. GPU clockspeeds have additionally been equally elevated, and whereas Qualcomm doesn’t disclose particular clockspeeds there, they’ve confirmed that the 8+ Gen 1’s Adreno GPU block is clocked 10% increased than the unique 8 Gen 1.

However, if something, the majority of Qualcomm’s features from the change in manufacturing nodes are being invested in slicing energy consumption. One thing of a sore level with the 8 Gen 1, TSMC’s higher 4nm course of implies that Qualcomm is seeing a lot decrease energy consumption throughout their SoC at iso-frequency.

Formally, Qualcomm is claiming a 30% enchancment in each GPU and CPU energy effectivity. Although as talked about earlier than, that is at iso-frequency and doesn’t consider the upper peak clockspeeds of the 8+ Gen 1. Consequently, the real-world energy financial savings aren’t going to be fairly as nice on a peak-to-peak foundation, however in accordance with Qualcomm the facility financial savings are nonetheless vital. Total, the corporate is touting a 15% discount in SoC energy utilization underneath “sensible utilization patterns” versus the unique 8 Gen 1, which in flip ought to result in improved battery lifetimes in handsets that undertake the brand new SoC.

Previous that, the official specs for the 8+ Gen 1 don’t replicate any materials modifications to the SoC’s configuration versus the unique chip. So we’re nonetheless trying on the identical built-in X65 5G modem, the identical Spectra ISP, and the identical video encode/decode blocks (sorry, gang, nonetheless no AV1 assist!). So, regardless of assembling a brand new die for his or her mid-generation refresh product this yr, there aren’t any new options to talk of with the 8+ Gen 1.

As for SoC efficiency, formally Qualcomm is claiming 10% improved GPU and CPU efficiency, owing to these aforementioned clockspeed will increase. Whereas we weren’t capable of attend a benchmarking session that Qualcomm held final week, the Efficiency Mode figures being launched by the corporate are roughly in-line with these claims. Qualcomm’s Geekbench 5 outcomes are a number of % increased than what we benchmarked again in December on the 8 Gen 1 launch occasion, although it’s notable that they didn’t obtain meaningfully increased scores in PCMark. GPU efficiency figures are equally combined, with a few of Qualcomm’s official outcomes coming very near our unique 8 Gen 1 outcomes, however I’m reluctant to learn an excessive amount of into them given how a lot if a distinction peak versus sustained testing makes in GPU outcomes. As all the time, the ultimate phrase might want to come all the way down to impartial, third-party testing, although at face worth nothing Qualcomm claims is unreasonable given the clockspeed enhancements and the thermal headroom gained from the change to TSMC.

Lastly, as for customers, they’ll be capable of get their first take a look at Snapdragon 8+ Gen 1 gadgets in Q3 of this yr. Based on Qualcomm, lots of the normal suspects have signed on to launch telephones primarily based on the brand new SoC, together with Asus, Motorola, OnePlus, Honor, and Xiaomi.

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

- Advertisment -
Google search engine

Most Popular

Recent Comments