This is why I hate these "Arc improves performance by X% for game Y" news posts. It make it look like they achieved some amazing performance improvements, but when the baseline FPS was low, improving by X% becomes a bit misleading. For example: lets say game Y ran at 10 fps before the new driver and now it has 30% improvement. Great, no? But in practice it means that it went from 10 fps to 13 fps, which sounds a lot less impressive.
So 268% improvement sounds spectacular, but for 10 fps before it just means that it now runs at 36.8 fps.
So without some actual FPS numbers these percentages don't mean much.
Going to be really interesting seeing Intel's next gen lineup with all these fixes in the bag.
I'm guessing Battlemage is going to be "just okay". I don't think it will be good enough to cause AMD to lower prices but I don't think it'll be a stupid purchase either. Intel still has quite a lot to catch up on, enough so that I think there is still going to be a substantial amount of room for improvement in Battlemage's architecture (improvements that just didn't yet present themselves or there wasn't time to fix). For the record, I don't necessarily find people who bought Alchemist to be stupid, at least those who knew they were just getting an experimental product. Anyone who bought it thinking it'll be a budget beast in a couple years is likely going to have regrets.
Celestial I believe is where Intel is going to be pretty much caught-up in both architecture and drivers, and where they will be truly competitive.
Crazy Joe:
This is why I hate these "Arc improves performance by X% for game Y" news posts. It make it look like they achieved some amazing performance improvements, but when the baseline FPS was low, improving by X% becomes a bit misleading. For example: lets say game Y ran at 10 fps before the new driver and now it has 30% improvement. Great, no? But in practice it means that it went from 10 fps to 13 fps, which sounds a lot less impressive.
So 268% improvement sounds spectacular, but for 10 fps before it just means that it now runs at 36.8 fps.
So without some actual FPS numbers these percentages don't mean much.
I completely agree - they're effectively just saying "this game is now playable". I get how they wouldn't want to phrase it that way, because they're not trying to say so directly that their product is broken/incomplete, but having astronomical performance improvements like 268% is kind of implying that anyway. When the only thing that changes is your software and you have exponentially better performance, it really highlights how broken things are.
Kind of funny how it's more impressive to hear something like a 10% improvement, because that implies that the drivers were already in acceptable shape but now just further optimized.
I'm guessing Battlemage is going to be "just okay". I don't think it will be good enough to cause AMD to lower prices but I don't think it'll be a stupid purchase either. Intel still has quite a lot to catch up on, enough so that I think there is still going to be a substantial amount of room for improvement in Battlemage's architecture (improvements that just didn't yet present themselves or there wasn't time to fix). For the record, I don't necessarily find people who bought Alchemist to be stupid, at least those who knew they were just getting an experimental product. Anyone who bought it thinking it'll be a budget beast in a couple years is likely going to have regrets.
I don't expect it to be anywhere near the top 30% of performers when it comes out, but less also will do it if they can get stability and compatibility up to par with the competition. REBAR being a requirement also becomes less and less of a "BUT IT NEEDS THIS TO BE GOOD!!!" screech, as it becomes a default option.
There wasn't enough brand on the market (only intel and asrock in here), and so not many sale, but Alchemist were a very good card for the price spended despite the driver were not on top (to not say "crappy") at launch.
Right now with the better driver and that editor support Intel too, people who have bough the 16G have the result of a good AMD or NVidia of the same class but for lot less money.
For a try out, it's a success for Intel, i would not have bet a coin on that before the card was out.
Main question is: What Intel will do now?
Right now with the better driver and that editor support Intel too, people who have bough the 16G have the result of a good AMD or NVidia of the same class but for lot less money.
This is just a myth. 3060 12GB has been 285€ for quite some time now vs. 330€ A770 16GB. And it's still closer to 3060 than 3070 and still often has visual corruption or bad stutter in new games.
ComputerBase tests every new game also on the A770, there are only few positive exceptions.
https://www.computerbase.de/2024-01/intel-arc-grafiktreiber-5186-benchmarks/ And Talos 2 upsampler crashes are also still sitting in the release notes. You still can't really trust Intel imho.