Innosilicon announced Fenghua II (aka Fantasy II), a second generation standalone graphics processor. The new GPUs consume between 4W and 15W, so they don’t require an auxiliary PCIe power supply or active cooling. As such, it can target entry-level desktops and notebooks.
Contrary to expectations, Innosilicon’s 2nd generation discrete GPUs do not offer higher performance than the previous generation. In fact, with a raw compute performance of 1.5 FP32 TFLOPS/12 INT8 TOPS, the Fantasy II is actually 3 more than the single-chip Fantasy I graphics card (delivering up to 5 FP32 TFLOPS/up to 25 INT8 TOPS performance) more than twice as slow. , according to 163.com (via @Loeschzwerg_3DC) .The obvious advantage of the new graphics card is its power consumption, which ranges from 4W to 15W, as opposed to 20W to 50W for the single-chip Fantasy I. big brother.
Fantasy I and Fantasy II graphics cards from Innosilicon
| fantasy II | Fantasy I Type A | Fantasy I Type B | |
| Number of GPUs | 1 | 1 | 2 |
| FP32 performance | 1.5 FP32 TFLOPS | 5 FP32 TFLOPS | 10 FP32 TFLOPS |
| INT8 performance | 12 Tops | 25 Tops | 50 tops |
| pixel rate | 48 Gpixels/sec | 160G pixels/sec | 320 gigapixels per second |
| video decoding | ? | 4x4Kp60, 16x1080p60, 32x720p30 | 8x4Kp60, 32x1080p60, 64x720p30 |
| number of users | ? | 16 1080p users | 32 1080p users |
| total graphics power | 4W to 15W | 20W to 50W | ? |
Considering the Fantasy II GPU’s compute power, it competes with Nvidia’s GeForce GTX 1630 graphics board, which uses the TU117 GPU, which was first introduced in April 2019. (And let me tell you, this will never be on the list of the best graphics cards available today.) Meanwhile, Innosilicon hasn’t published any real-world in-game performance results for its new graphics chips. . The only thing it says is that it reached 6500 on the old GLMark2 benchmark. IT home.
Innosilicon has not disclosed which architecture is used in its Fantasy II GPUs, but the new graphics chip is said to use the same ImgTec’s PowerVR architecture used in its Fantasy I GPUs. You can guess. However, newer graphics processors support DirectX, Vulkan, OpenGL, OpenCL, and even OpenGL ES application programming interfaces (APIs). Newer graphics processors, on the other hand, incorporate a RISC-V security core with a Physical Uncloneable Function (PUF).
Fantasy II GPUs can support 2GB, 4GB, 8GB LPDDR4/4X/5/5X memory configurations with data transfer speeds up to 10 GT/s. The chip supports DisplayPort/eDP 1.4, two HDMI 2.0, LVDS, and D-Sub/VGA outputs, but it’s unclear how many display pipelines it has (i.e. how many monitors it can support simultaneously). . As for the host interface, the chip uses the PCIe 3.0 x8 bus.
Innosilicon’s Fantasy II is definitely not the best entry-level GPU, but it has another advantage (besides lower power consumption) over AMD, Intel and Nvidia’s discrete graphics offerings. CentOS, Kirin, KylinOS, Tongxin, UOS, Ubuntu, etc., used only in China). In addition, it supports all Chinese CPU platforms (those major vendors’ GPUs don’t support), including Loongson, Huawei Kunpeng, Hygon Dhyana, Shenwei (Sunway), and Zhaoxin Feiteng. Additionally, Innosilicon will release Windows 10 drivers for its Fantasy series GPUs.
