On May 20, we waited for the release of "TSMC Snapdragon 8 Gen 1" and "SM8475" - Snapdragon 8+. The dragon 7 Gen 1 also came with us. Qualcomm chose 520 to hold the press conference, which was a bit like the name of Snapdragon 888 at that time, obviously hoping to please domestic users.

First of all, the first generation Snapdragon 7, code named SM7450-AB, is the successor of Snapdragon 778G/778G+.

The first generation Snapdragon 7 is a Samsung 4nm process, with a 2.4GHz A710 super core+three 2.36GHz A710 cores+four 1.8G A510 cores. Qualcomm did not mention the extent of CPU performance improvement. GPU performance claims to be 20% higher than Snapdragon 778G. The A710 itself has a weak architecture and low frequency. The CPU performance is expected to be significantly different from that of MediaTek 8100.

The most obvious improvement of the first generation Snapdragon 7 is the periphery It supports a 4K screen and a 200 million pixel camera. The baseband is upgraded to X62. The peripheral chips are also upgraded from the FastConnect 6700 to the FastConnect 6900. The dual frequency concurrent function is added. The theoretical peak value is 3.6Gbps. The press conference said that all major manufacturers' Snapdragon 7 models will be launched in the second quarter, and OPPO Reno8 Pro will launch Snapdragon 7 for the first time.

Today's protagonist, the first generation Snapdragon 8+(Snapdragon 8+Gen 1, hereinafter collectively referred to as Snapdragon 8+), has upgraded its core change to TSMC's 4nm process, claiming that the CPU multi-core performance has been improved by 10% and the GPU frequency has been increased by 10%. The GPU is still Adreno 730. Previously, Snapdragon 8 was 818MHz. If the frequency is increased by 10%, the GPU frequency of Snapdragon 8+should be around 900MHz (899.8).

The CPU of the new generation Snapdragon 8+is still Cortex-X2+3 Cortex-A710+4 Cortex-A510, The frequencies are 3.19GHz+2.75GHz+2.02GHz respectively.

For comparison, the frequency of Samsung's 4nm process Snapdragon 8 Gen 1 is 3.0GHz+2.5GHz+1.8GHz. The frequencies of Snapdragon 8+ Increase 6.7%, 10%, 12.2% ；

However, the frequency of MediaTek's 9000 in TSMC's 4nm process is 3.05GHz+2.85GHz+1.8GHz. The frequencies of Snapdragon 8+ 5%, - 3.5%, 12.2% higher 。

People are most concerned about power consumption. Compared with Snapdragon 8 Gen 1, Snapdragon 8+claims Overall power consumption reduced by 15% (The scene is unknown), and CPU and GPU can reduce power consumption by 30% under the same performance (CPU is SpecINT2K6 test, GPU is GFX Aztec Ruins test). Note that this is not the commonly used data of "peak power consumption reduction XX", but the power consumption change under the same energy, so the power consumption gap should be less than 30% at peak output.

Compared with some score data of Snapdragon 8+released by Qualcomm, stream saving means that the CPU finally catches up with Tianji 9000 and the GPU takes off directly. Regardless of power consumption, Snapdragon 8+has truly returned to the throne of Android performance.

Compared with Snapdragon 8 Gen 1, the CPU of Snapdragon 8+ Single core and multi-core are 8% and 13% stronger, and GPU is about 10% stronger ；

Compared with Tianji 9000, Snapdragon 8+CPU Single core is strong by 3%, multi core is weak by 3%, but GPU leads by more than 20% ；

Compared with Apple's full blood A15, the single core CPU of Snapdragon 8+is 30% weaker and the multi-core CPU is 13% worse. The Aztec Ruins test with the highest GFX pressure was tied to A15, and the GFX test with low pressure was about 10% stronger.

QUALCOMM claimed that the power consumption of the original god was reduced by 30%, and it even released the frame number of the famous "card scoring software" "original god". Under the publicity of 25 degrees, the average frame rate of the game in 60 minutes reached 60.2 frames (stand in place!), and claimed that the power consumption of the original god was reduced by 30%.

The press conference mentioned that Xiaomi will take the lead in carrying the new generation of Snapdragon 8+, while iQOO, True Self GT2 Master Discovery, and One Plus New flagship (the third quarter) all said that they will be the first to carry the new generation of Snapdragon 8+. Moto also has a corresponding poster, but does not mention the first word.

Sina's ROG Snapdragon 8+Engineering Machine Original Test ↑

What's more interesting is that the first Snapdragon 8+was tested by ASUS ROG, whose rabbit running score was 1.118 million, the original spirit was 25 minutes, the average balance mode was 5.4W, 57.7fps, and the body temperature was 44 ℃, Streaming saving is the balance mode of Xiaolong 8+, which has exceeded the performance mode of Tianji 9000, and the power consumption is 0.4W lower (5.4W vs. 5.8W).

Of course, here is the comparison between game phones and "game biased phones". There are obvious differences between the two in terms of fuselage space and trade-offs, which is conducive to Snapdragon 8+. But at least at the power consumption level of Yuanshen, it seems to have a play. I hope it will not be like Tianji 9000 that the mass production machine can not run faster than the engineering machine.

The X2 super core of Snapdragon 8+is 0.2GHz higher than Tianji 9000, but The performance of single core CPU does not open the gap, which may be due to the cache size However, Qualcomm has not announced the cache specification of Snapdragon 8+, which is questionable for the time being. As a reference, Snapdragon 8 Gen 1 is a 6MB L3 cache+4MB SLC (system level cache), while Tianji 9000 is an 8MB L3 and 6MB SLC.

Now Qualcomm Snapdragon 8+and MediaTek Tianji 9000 are all TSMC 4nm processes. Although both are X2, Qualcomm is "small cache"+high frequency, and MediaTek is "low frequency"+large cache. Considering the crotch degree of the X2 architecture itself, it is necessary to match the actual power consumption and heating conditions to determine which side is the more correct.

The frequency of the three core groups of Snapdragon 8+is 5%, - 3.5% and 12.2% higher than that of Tianji 9000 respectively, but the actual CPU single core performance is close. It can be seen that Qualcomm's multi-core CPU scores are mainly supported by four frequency cores. Qualcomm obviously purposely avoids increasing the frequency of the A710 core, which also pulls the hip . How much power consumption effect can this bring? It also needs to be confirmed when the production machine comes out.

Of course, we are more concerned about the power consumption of CPU under low load. Even if the GPU has the highest performance, people will not play games all the time. Qualcomm has changed the process and has the space to raise the frequency, so the power consumption of the same performance can be reduced by 30% and the peak performance can be significantly improved. We have to lament that Qualcomm used Samsung's 4nm technology to pull the legs. The GPUs of Snapdragon 888 and Snapdragon 8 Gen 1 can still be so strong. Adreno GPUs from AMD's team are really rich.

What else can I say? TSMC Powerful! There is not much time left for Fage. Come on, Fage.

Follow our Weibo @ Love Computer

Follow our WeChat official account: playphone

Of course, we also pay attention to our Bilibili account: love computer