Tuesday, January 12th 2021
Samsung x AMD: South Korean Giant Announces RDNA Integration in Next-Gen Exynos
Samsung today at its Exynos 2100 launch event announced that its labor with AMD to integrate the company's RDNA graphics architecture onto Exynos chips has born fruit. It's unclear today on which set of technology this integration is bound to - whether RDNA, RDNA 2, or a combination of both - and actual products will only hit shelves by the end of 2021 and beginning of 2022.
Samsung has announced that the design-in for AMD's RDNA platform into the company's flagship Exynos products for the 2021-2022 timeframe have been successful, and that the first iteration of the design will see the light of day on the upcoming Exynos 2100. The collaboration has reportedly resulted in very good performance values obtained from their IP merger in May 2020. It seems we have a few months to look towards to before we see a Galaxy phone with an RDNA-powered engraving, though.
Samsung has announced that the design-in for AMD's RDNA platform into the company's flagship Exynos products for the 2021-2022 timeframe have been successful, and that the first iteration of the design will see the light of day on the upcoming Exynos 2100. The collaboration has reportedly resulted in very good performance values obtained from their IP merger in May 2020. It seems we have a few months to look towards to before we see a Galaxy phone with an RDNA-powered engraving, though.
7 Comments on Samsung x AMD: South Korean Giant Announces RDNA Integration in Next-Gen Exynos
I tip my hat for that.
qualcomm>>>>>>
The SHIELD's GPU is relatively fine, but those ARM Cores date to like 2012.
At the very least, Samsung may have the GPU performance crown down - Nvidia only updates their tablet power consumption level SOC every other year when they get tired of counting money, and no one else should be able to claim that title, unlike a certain Qualcomm who just overclocks their GPU (literally).
Samsung will also be able to optimize the SOC with their devices. Speaking of optimizing for your own devices, Apple SOC might still just be faster than all the other ARM SOC anyways.
What I'll be looking out for is RDNA integration performance. As this is a flagship SOC, it should be on parity with SD888 memory controller using LPDDR5 (totaling 50GiB/s, according to Wikipedia), and we can see what kind of performance RDNA can bring to APU's on D4-3200(same memory bandwidth), as well as speculating on memory scaling into D5-6400(assuming 128bit channel, I have no idea how bandwidth works with consumer D5 platforms) with current RDNA1 and RDNA2 GPUs.