GPUsNews

AMD Rembrandt iGPU Faster than NVIDIA’s MX550 Mobile dGPU [Rumor]

A while back, NVIDIA announced its GeForce RTX 2050, MX550 and MX570 mobile GPUs. Two of these SKUs are based on the Ampere microarchitecture, with the RTX 2050 and the MX570 leveraging the GA107 core, and the MX550 being a derivative of the Turing-class TU117 core. The MX550 features around 1,024 shaders paired with 4GB of GDDR6 memory, while the MX570 will likely increase that figure to 1,280-1,536 (Ampere) FP32 cores.

 MX570MX550
CUDA Cores1024+1024?
ROPs32?32?
Memory Clock12Gbps GDDR6?12Gbps GDDR6?
Memory Bus Width64-bit64-bit
TDP Range30W?30W?
GPUGA107TU117
ArchitectureAmpereTuring
Launch DateSpring 2022Spring 2022
 RTX 3060
Laptop GPU
RTX 3050 Ti
Laptop GPU
RTX 3050
Laptop GPU
RTX 2050
Laptop GPU
CUDA Cores3840256020482048
ROPs48323232?
Boost Clock1283 – 1703MHz1035 – 1695MHz1057 – 1740MHz1155 – 1477MHz
Memory Clock14Gbps GDDR614Gbps GDDR614Gbps GDDR614Gbps GDDR6
Memory Bus Width192-bit128-bit128-bit64-bit
VRAM6GB4GB4GB4GB
TDP Range60 – 115W35 – 80W35 – 80W30 – 45W
GPUGA106GA107GA107GA107
ArchitectureAmpereAmpereAmpereAmpere
Manufacturing ProcessSamsung 8nmSamsung 8nmSamsung 8nmSamsung 8nm
Launch Date01/26/202105/11/202105/11/2021Spring 2022

The RTX 2050, despite using the Turing nomenclature is based on the Ampere design, packing 2,048 FP32 cores and 4GB of GDDR6 memory via a 64-bit bus. Thus, we’re essentially looking at the 3050 with a slimmer bus. Two 3DMark Time Spy results of the RTX 2050 and MX550 have surfaced. The RTX 2050 scores 3,369 points in TimeSpy while the MX550 is limited to just 2,510.

In comparison purposes, a recently leaked Rembrandt score shows AMD’s next-gen RDNA 2 integrated graphics scoring over 2,700 points, easily outmaneuvering the MX550. This is rather disappointing as this SKU will launch alongside Rembrandt in early 2022, and being a discrete GPU should be at least 15-20% faster than AMD’s integrated graphics solutions. We’ll probably see situations where OEMs pair these two in one design, making the dGPU redundant, and a sheer waste of silicon.

Via: Hardware

Areej

Entrepreneur, tech-analyst, PC gamer, and hardware enthusiast. Former co-founder of Techquila (2017-2019), a fairly successful tech outlet. Been working on Hardware Times since then, an outlet dedicated to computer hardware and its applications.

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button