Today, for the first time ever I've switched to the dark side aka AMD. After almost 4 hours of working on the system, 1h was spent on trying to remove the damn 24-pin power cable... After putting everything in, it gave me a panic attack because the PC didn't want to turn in even though it was getting power; turns out I plugged in the front I/O cables wrongly Because of the upgrade, I've went from 4c/4t to 6c/12t & can run 14 cars on American Road with 15fps without having the CPU maxed out, where before just 1 or 2 cars were needed for that... Here's the results: i5-6400 Ryzen 5 2600
When i see the ryzen benchmarks im really thinking of getting one instead of my i5 6600k but this time in waiting on the ryzen 3 series
His is really great for rendering as well as good upgrade from i5-6400, but compare what 1 vehicle result you get to his result, that is what defines how well BeamNG will run and also how many cores you can utilize in 0.16 before hitting limit of that one core. From far overclocked 6600k improvement might not be as big as total banana bench score might let you believe. With 3rd gen Ryzen you should start seeing improvements indeed as long as you pick model that has higher clock speeds than what you are getting now, with same clocks it might not be that much different from your current setup, but that is something I'm looking with great interest, first banana bench results with 3rd gen Ryzen will be really interesting to see as well as how well their FPU manages with several vehicles running on map, etc. According to leaked benchmarks though, 3rd gen Ryzen would be around same performance with same clocks as you have (some models will have higher clocks so performance will be higher on those), except more cores to feed and then unknown is the AVX, FPU and how memory bandwidth works out as that all alos affects how well it will run BeamNG, very interesting times though.
Don't forget that bananabench doesn't really translate to in-game performance past a certain point. Intel is currently better than AMD for BeamNG due to the in-game performance being limited by absolute max single threaded video-rendering related load, and vehicle rendering has received far less focus on optimization than map rendering in recent updates. However, judging by what we saw at CES, Zen 2 should handily outperform intel across the board.
Let's just say that it depends, would not rely too much on anything from CES, everyone these days show their thing in best possible light, real BeamNG performance is known when someone tests the thing. Have seen too many times this product X will be so awesome based on before hands on data, that I rather wait 1 vehicle result and fps at Italy before making any strong opinions to direction or another. What is clock speed at Italy will be interesting number that can't be found before testing thing at Italy. I do kind of need 2nd system, so need to consider that option too, so far looks solid candidate, but until seeing posts of bananabench with 5Ghz model, I will not know. But yeah, total bananabench score is pretty meaningless today, 1 vehicle score is more interesting for keeping fps over 60 with high end GPUs at least. Obviously devs needed to change optimizations in 0.16 so Bananabench numbers with 0.16 are only ones comparable and by Christmas there is still updates to come so they have many chances to mess up our bananabench comparison attempts
Hello, Here are the bench results for my Threadripper 1950X at stock speed. Actual CPU speed was 3697.81 Mhz (from CPU-Z). RAM is DDR4 3200. Tom.
Intel Xeon 2680 V4 tested with a bunch of stuff open, but 237 isnt too bad (I dont need it but I still really want a threadripper...)
You get less than 23Mbeams/s with single vehicle (modern gaming CPU has over 50Mbeams/s for 1 vehicle), that is causing serious bottleneck in many games because of graphics stuff etc. I have been thinking if it would be possible to create new metric that actually could give some more useful data of how many different vehicles one can run in practice, different game versions do change that quite bit though, also difference between normal and high graphics setting is clear, using same copy of car at JRI or grid map is giving huge boost in number of cars one can run, so might not be easy to create such metric. Forget about Threadripper though, soon there are much better ones, if Threadripper gets around 360Mbeams/s with 16 Cores and I get around 314Mbeams/s with 6 cores, just think of what 16 cores at 5Ghz and similar IPC to mine would then do? That is potentially over 700Mbeams/s what Zen2 might do and with around same price as Threadripper, while not having bottleneck in games. It is interesting to see if that will run more vehicles in practice than say 314Mbeams/s depends how game optimizations change the situation, I guess, but there are new dreams to have
Yeah for sure the xeon is pretty bad with single thread speed (it is only 2.2Ghz after all) Though I haven't had many issues with games I don't play an awful lot of games either, most recent is probably FFXV which runs nicely maxed out, I would guess I'd have a hard time with more modern games like RDR2 when it comes to PC. Ryzen is probably where I will end up buying, since I got my current xeon as an engineering sample it was less than half the MSRP of a full chip. I wouldn't really ever spend more than £450 on any CPU so thats my main limiting factor.
My poor old i5 2500 non-K 8 GB 1333MHz GTX 1060 6GB Whatever that means. I've no plans to upgrade any time soon, so it'll have to do.
3900X @fufsgfen if you want to add to https://www.beamng.com/threads/ulti...ttle-has-graphs-and-awards.60389/#post-983713
Wow! whats your full setup, My board supports that Processor i may be getting it a few months down the line.