Move to TSMC for more speed and less power

As the darkness of night approaches China this evening, Qualcomm is hosting an event to launch a mobile-focused product they’re calling “Snapdragon Night.” Leading the event is the announcement of their new flagship SoC, Snapdragon 8+ Gen 1. A mid-generation update for their flagship smartphone SoC, Snapdragon 8 Gen 1, 8+ Gen 1 follows Qualcomm’s annual tradition of releasing a product update to boost performance and give partners something new to work with. In the second half of the year. And for this year in particular, we’re looking forward to a very noticeable chip change from Qualcomm.

Unlike previous generations where Qualcomm just launched a faster speed enclosure than the silicon it has, in 2022 we have something more substantial to talk about. Qualcomm has completely transformed foundries – moving from Samsung to TSMC – and as a result a new mold is being introduced. Thanks to this, Qualcomm’s Snapdragon 8+ Gen 1 reaps something of a one-off manufacturing gain, allowing them to demand CPU and GPU performance while lowering power consumption at the same time.

Diving quickly into the specs, the new Snapdragon 8+ Gen 1 is essentially the original Snapdragon 8 Gen 1 ported from Samsung’s 4nm line to one of TSMC’s 4nm line. Under more normal conditions, this kind of transformation would probably be unnoticeable – or at most, a fun exercise in searching for evolving states – but for Qualcomm’s flagship SoC, it’s even more important.

While official sources and statements about the quality of Samsung’s 4nm process are few and far between, unofficially, it has become clear that Samsung’s 4nm process has not lived up to expectations. This has caused a cascading effect on chips made on the process node, resulting in the original Snapdragon 8 Gen 1 developing a convergence of power consumption, and the performance of Samsung’s Exynos 2200 was no better. On the contrary, by all accounts TSMC N4 process It looks great, with the optically deflated node constructed from TSMC’s successful and very efficient 5nm technologies.

As a result of this performance gap between Samsung and TSMC’s 4nm nodes, Qualcomm is taking the unusual step of (essentially) moving its high-end SoC to TSMC’s fab. Which, while not strictly necessary – Qualcomm has a lot of momentum and the 8 Gen 1 has sold well – is certainly a wise move for the company. Qualcomm is facing particularly tough competition this generation of MediaTek, which has a leading level Dimensions 9000 SoC It was the main product of the TSMC 4nm node. This leaves MediaTek with a distinct advantage versus the original 8 Gen 1, one that Qualcomm is happy to undo.

Ultimately, switching in fabs gives Qualcomm a chance to improve on the original 8 Gen 1 version from both ends of the spectrum, leading to the Snapdragon 8+ Gen 1. On the performance front, the TSMC node gives them an easy opportunity to ramp up the CPU and GPU Clock speed for even more performance. . The Cortex-X2 main core is now clocked 7% higher, at 3.2GHz, meanwhile, the A710 and A510 combinations saw significantly higher connection speed, at about 12% each. Now even the slowest A510 cores can run at 2GHz. The GPU’s overclocking has also been similarly overclocked, and while Qualcomm doesn’t reveal specific clock speeds there, they confirmed that the 8+ Gen 1’s Adreno GPU block was clocked up 10% higher than the original 8 Gen 1.

But, if anything, the bulk of Qualcomm’s gain from switching in contract manufacturing is being invested in lowering power consumption. Something of a sore point with the 8th generation, TSMC’s better 4nm process means Qualcomm is seeing significantly lower power consumption via its iso-frequency SoC.

Officially, Qualcomm is claiming a 30% improvement in both GPU and CPU power efficiency. Although as mentioned before, this is done at iso frequency and does not take into account the higher clock speeds of the 8+ Gen 1. Thus, real-world power savings will not be quite as great at peak-to-peak, but according to Qualcomm Energy savings are still significant. Overall, the company is touting a 15% reduction in SoC power usage under “Practical Usage Patterns” versus the original 8 Gen 1, which in turn should lead to improved battery life in phones that adopt the new SoC.

After that, the official specs for the 8+ Gen 1 do not reflect any physical changes to the SoC configuration versus the original chip. So we’re still looking at the same integrated X65 5G modem, the same Spectra ISP, and the same video encoder/decoding blocks (sorry, gang, no AV1 support yet!). So, while a new mid-generation refresh product has been molded this year, there aren’t any new features to talk about with the 8+ Gen 1.

As for SoC performance, Qualcomm officially claims to improve GPU and CPU performance by 10%, due to the aforementioned increased clock speed. While we weren’t able to attend Qualcomm’s benchmarking session last week, the performance mode numbers released by the company roughly align with those claims. Qualcomm’s Geekbench 5 scores are several percent higher than what we evaluated in December in 8 First generation launch event, although it is noticeable that they did not score purposefully higher on PCMark. GPU performance numbers are similarly mixed, with some official Qualcomm results getting close to our original 8th-gen results, but I’m hesitant to read too much about it given how much of a difference Peak difference versus ongoing testing is in GPU results. As always, the final word should come down to independent third-party testing, although there’s nothing Qualcomm claims doesn’t make sense given the improvements in clock speed and heat vacuum gained from switching to TSMC.

Finally, for consumers, they will be able to get their first look at the Snapdragon 8+ Gen 1 devices in the third quarter of this year. According to Qualcomm, several usual suspects have signed on to release phones based on the new SoC, including Asus, Motorola, OnePlus, Honor and Xiaomi.