The Intel Gen11 GPU is also expected to outperform the Radeon RX Vega 10 by 4 percent and 11 percent in Rocket League and World of Tanks, respectively. Since Ice Lake-U laptops have yet to launch, however, AMD still has the chance to update its drivers to try and close the narrow performance gap.
https://www.notebookcheck.net/Intel-Ice-Lake-U-promises-up-to-16-percent-better-FPS-in-Overwatch-than-the-AMD-Ryzen-7-3700U.423720.0.html
My my, how the tables have turned.
A few years ago, AMD was using the graphics argument against Intel, since on the CPU side, it couldn't hold a candle. Now, Intel is doing it since Ice Lake is so low clocked.
funny thing is that they benchmarked it with LPDDR4x at 3866 MHZ if i am not mistaken, while Ryzen with DDR4 2400. Yet another fail by intel... My guess is Ryzen will be around 30% ahead with improved memory support. It's a matter of time, all the great empires have fallen, I would say the time for intel has pretty much ran out ;)
Quote from: fdsofldmos on June 11, 2019, 08:59:25
My my, how the tables have turned.
A few years ago, AMD was using the graphics argument against Intel, since on the CPU side, it couldn't hold a candle. Now, Intel is doing it since Ice Lake is so low clocked.
oh, let's not forget one thing: ryzen mobile is 12nm, while intel us 10nm. I dont see this benchmark as a win
BTW guys remember that the low power memory intel using have 29.8gbit/s bandwidth while normal dram 2400mhz dual have ~38gbit/s.. So it's amd with advantage.
Quote from: Opelit on June 11, 2019, 09:53:00
BTW guys remember that the low power memory intel using have 29.8gbit/s bandwidth while normal dram 2400mhz dual have ~38gbit/s.. So it's amd with advantage.
LPDDR4 had 29.8 Gb/s, LPDDR4X has 34.1 Gb/s. However, the specs i found for LPDDR4X did not mention the memory speeds. Since the performance bottleneck of Vega iGPU is the memory frequency, I do not see how would this be in AMD advantage.
I would be really interested to see and AMD APU with LPDDR4X 3866 MHz. I wonder why intel would benchmark it with a different type of memory if it would offer inferior performance, since most of the pre-release benchmarks are done in the best-case scenario. Would you intentionally sabotage your product lunch and benchmark it with an inferior memory?
good thing they'd didn't even try bringing up the Fortnite/PUBG/CS:GO argument :D that'd be completely suicidal of Intel's