So I have been sampled an ES A750 and a retail A770 by Intel and I have promised a few people I would benchmark these for a few weeks now. Since 3.18 is coming up soon I decided to do some runs to gather before and after numbers for Gen 12 vs current render engine. For my testing I have used the same test bench which is equipped with a 5900x and 16GB of 3600c16 DDR4 and have swapped out the GPUs and reset the games USER folder inbetween each GPU test along with a pre run to eliminate first time stuttering. Intel ARC is operating on the latest driver (31.0.101.3959) and Nvidia RTX 3080 is operating on the latest driver (527.56). The OS used is windows 11 Pro with core isolation / VBS turned off. The route and rough timings taken for each run can be seen below in a video i created to allow me to stick to a set route around Loreville to keep the benchmarks as consistent as possible. Now finally on to the results. Firstly the RTX 3080 is the clear winner here with a massively better experience at 1080p compared to both ARC cards. It felt much smoother than them at both 1080p and 4k and practically matching the A750s 1080p performance at 4k.... The A770 appears to have issues with .2% lows as shown by the fact that it has lows half that of the A750. I ran multiple extra runs to determine if this was by chance or not but the A770 consistently had much worse lows than the A770. Both cards are pretty close in 1080p performance when it comes to averages but the A750 feels smoother due to the much better .2% lows. At 4k the A770 pulls ahead of the A750 where it is now noticably smoother than the A750 but both cards feel horrible as an experience because of the overall low framerates and the only card that comes close to a smooth nice experience is the 3080 and even that is below what I would consider an enjoyable experience. Something of note is that CapFrameX indicates that ReBar was not enabled for both the ARC cards (It was enabled in Bios but not in software) showing that it was enabled in hardware but not in software. I am unsure as to the validity of this but it could prove to be an issue. AFAIK after reaching out to some people it might be a bug with the version of CapFrameX that I am using and not an accurate reading. With the results below its also worth remembering that SC is a very CPU limited game when it comes to FPS and the 5900x is not enough to fully remove this bottleneck even at 4k IMO (though I am not 100% sure on this) so take the results I have here with a grain of salt and your experience probably will differ with ARC being as it is at the moment. Finally I want to finish off by adding a little context to why and what im testing. Firstly the A750 is an engineering sample that is near release product so im not entirely sure how indicative its performance is of retail A750s but it should be practically identical AFAIK. The A770 is retail though so no worries there. The RTX 3080 used for testing is an EVGA FTW 3 ULTRA card which I have owned for just over a year and a half now. Great card would recommend to anyone looking for a 3080. ARC is known to have some performance issues with non DX12 & VK titles which means this likely isnt its strongest showing and performance will almost certainly change once Gen 12 is fully out. But id expect this to be reflected more in the 1080p results than 4k since ARC has a massive CPU overhead for drivers. I will come back and retest this at some point but I cant guarentee it will be anytime soon. Another caveat with this is that ARCs drivers are improving rapidly. ARC will be a more capable platform by the time 3.18 hits main PU release. It wont be possible to compare these results with any other results people may get on newer drivers because of this fact. But I hope these results will still prove useful to anyone wondering if its a good idea to buy ARC for SC, my opinion is probably not, but depending on the price your opinion might be different.