The test data of MediaTek Tianji 9000 was finally lifted on December 20, and we will now take a look at the performance of the first TSMC 4nm process, the more full X2+A710+A510 and the deca-core Mali-G710.
First re-read the specifications of the Tianji 9000, TSMC 4nm, 3.05G X2 (1M L2) + 3x2.85G A710 (512K L2) + 4x1.8G A510 (256K L2), 8M L3 and 6M SLC cache, Mali-G710 MC10, the first 7500Mbps LPDDR5x memory support.
The "enemy of a lifetime" Snapdragon 8 Gen 1 is Samsung 4nm, 3G X2+ 2.5G A710, the L2 cache on both sides is consistent, Qualcomm's A510 small core L2 cache is unknown, using a more area-saving kernel 2+2 structure design, 6M L3+4M SLC, 6400Mbps LPDDR5 memory.
The Tianji 9000 is higher in super core and large core frequencies, L3 and SLC caches are higher, and memory specifications are higher (although there is no mass production machine with LPDDR5x memory), but the biggest difference may be the process difference between TSMC and Samsung.
Tianji 9000 is an engineering machine that does not even have graphite heat dissipation stickers (unfortunately, this brick has a large heat dissipation space), 12GB of LPDDR5 6400 memory + 256GB UFS 3.1 + LCD screen (claimed to consume 0.6W more power than A screen) + native Android 12. GeekBench and Antutu are both offline special versions, but the sub-project and the latest version are not much different, and can be compared horizontally.
[GeekBench 5] Arm's X2 super core this time, is a model of standing still, even if it is TSMC + FaGe, single core has not made much progress, single core 1264 points level, so far can not win the IPhone 11 Pro Max on the A13.
But the multi-core score, the Tianji 9000 has three 2.85GHz high-frequency A710 (higher than most SoCs' super core frequency), and the Android camp's first appearance of the big cache, multi-core score is smaller than A14. While other media teachers have better heat dissipation and physique in their hands, the Tianji 9000 test machine has broken through the 4400 point mark, but the one in our hands only hovers below 4300 points.
GeekBench 5 sub-project, join the Snapdragon 888 and 888+ to accompany the run. Tianji 9000 and Snapdragon 8 Gen 1 are both X2 supercores above and below 3GHz, and the scores are also very close, but in the sub-projects, the navigation, HTML5, image repair and motion structure of the Tianji 9000 are weak, with amplitudes between 1% and 6%, while PDF rendering, Gaussian blur, image compression and other items are 10% or more stronger.
In the multi-core part, Tianji 9000 is 21% stronger in the case of a total score of 14%, and the most exaggerated ray tracing and machine learning projects are more than 30% stronger, and the position of the Android CPU boss is solid. However, looking at the change in scores, the 2.85GHz ultra-high frequency A710 (Qualcomm is 2.5G) of the Tianji 9000 seems to be significantly weaker on HTML5 and Rang.
It should be noted that the early Snapdragon 8 Gen 1, multi-core performance compared to Snapdragon 888+ has no advantage at all, and subsequent manufacturers have to come out to redeem it.
[GFXBench] This part of the introduction of Feiwei teacher's unique Snapdragon 8 Gen 1 QDR data, this year Qualcomm's Adreno GPU to find the glory of the past, and finally can play with Apple's A15.
The GPU of the Tianji 9000 is one gear weaker, not able to do the Snapdragon 8 Gen 1, but far stronger than expected, in the high-pressure Aztec test, even smaller than the iPhone 13's youth version A15, close to the iPad mini 6 Max-Q version A15, and iPad mini 6 difference of 4%, and Snapdragon 8 Gen 1 QRD difference of 11.4%, and A15 difference of 18.1%; and in the low-pressure Manhattan and Manhattan 3.1 is closer.
[An Tutu] Here to borrow THE 8 Gen 1 refrigerator data of WHYLAB, Fa Ge is still passive heat dissipation. The CPU and MEM (memory + flash memory) of the Tianji 9000 are 13% and 12% stronger, respectively, UX (data security/data processing/strategy games/image processing and I/O performance and other projects) are 2% stronger, and the GPU is 12% weaker, but the total score is 102W, and the GPU weight of the entertainment rabbit is wrong (manual funny .jpg).
The rest of the PCMark (18162 points), ETHZ AI benchmark V5 test (1058K) will not be elaborated.
With the AndSPEC06 test that Feiwei teacher stole before, the X2 super nuclear energy efficiency of the Tianji 9000 is 49% ahead, and the A710 nuclear energy efficiency is 40% ahead (or in the case of higher frequency).
A thousand words come together in two sentences: TSMC Yes! MTK Yes！