New cpu - i5 vs Ryzen 5

formula1996

Registered
Hi guys!
I am looking for a new pc. I cant decide between the i5-7500 vs ryzen 1500x/1600 (maybe 1600 if the prices stay low in my country). I know Ryzen is better in multithread applications but what about gaming like rF2? Anybody runs rFactor2 on a ryzen build pc? How it is? The gpu will be an Rx570 4gigs.
Pls share your experiences if you have :)
 
I don´t have any experience for myself exactly, but what I can tell you is that singel-core performance
of the i5 is significant better, so when rF2 is the main purpose of your new build the i5 is the better choice.
Also the i5 produces much less heat than the ryzen (if that is somehow important for you).
(TDP´s are missleading here! Seems as AMD is cheating here with their numbers)
 
Both should run rF2 without issues, however, if you are into VR gaming, then the game might become CPU limited in some cases from what users have been reporting, and then you're better of with the i5 with its better single core performance.
 
for some games ryzen 5 is better, for other games i5 is better. rF2 has very limited threading so it falls into the latter category.
 
Not many games can use more than 2 cores,or even two threads,well sims anyway.
Do we know how many threads rf2 uses and what uses them,

Lots of modern games can use far more than 2 threads. Battlefield 1 and Watch_Dogs 2 will both happily use 16 threads. Even F1 2016 shows much better performance with 8 threads compared to 4 (on an i7 with hyperthreading enabled vs disabled, or 6+ core Ryzens compared to i5s).

As for rF2, it only seems to use 2 threads when I've looked at my CPU usage with the game running.
 
Lots of modern games can use far more than 2 threads. Battlefield 1 and Watch_Dogs 2 will both happily use 16 threads. Even F1 2016 shows much better performance with 8 threads compared to 4 (on an i7 with hyperthreading enabled vs disabled, or 6+ core Ryzens compared to i5s).

As for rF2, it only seems to use 2 threads when I've looked at my CPU usage with the game running.

So two games,hardly mainstream is it
 
Lots of modern games can use far more than 2 threads. Battlefield 1 and Watch_Dogs 2 will both happily use 16 threads. Even F1 2016 shows much better performance with 8 threads compared to 4 (on an i7 with hyperthreading enabled vs disabled, or 6+ core Ryzens compared to i5s).

They may somehow crudely divide the work for all 16 threads and the graphic driver may also allocate jobs for idle CPU threads. Still, almost all games released today are split between max three or four main threads. It makes no sense to program a game more multithreaded than that, becase firstly it's very difficult and secondly the games have to work on the common lowest hardware denominator, which today is a quad-core i5 CPU with no HT or i3 with HT (4 threads).
 
So two games,hardly mainstream is it

I was just giving a couple of examples. Games where an i7 (4c8t) crushes i5s (4c4t) simply by virtue of having more threads at the exact same (or even lower) clockspeeds include:

Battlefield 1, Watch_Dogs 2, F1 2016, Hitman, Mafia III, Ashes of the Singularity, Gears of War 4, Overwatch, Total War: Warhammer

That is not an exhaustive list. In most of those games, an older i7 like the i7-4790K is actually faster than the fastest current i5-7600k despite its single threaded performance disadvantage, because of more concurrent threads.
 
It would be interesting to know if S397 team is planning improvements on multithreading.
I'm really interested too in a Ryzen 5 1600X for the 12 threads versus my i5 2500k (4C), and I'd like to know if this is a good investment for rF2. Not only of course, but this will be then main use.
 
It would be interesting to know if S397 team is planning improvements on multithreading.
I'm really interested too in a Ryzen 5 1600X for the 12 threads versus my i5 2500k (4C), and I'd like to know if this is a good investment for rF2. Not only of course, but this will be then main use.

Going from my I7 4770k to a R7 1800X, id say the difference is not noticeable. however you may get in the region of 10 fps higher then your i5 2500k.

I cant speak for S397 but i doubt there will be any major improvements in multi threaded that will allow for any advantages especially given the age of the engine used. simple fact this engine takes advantage of clock rate and looks to use 2 threads, which favors Intel CPU's. So if you just gamed in old engines and older games, Intel will always be the better choice. However when you get to the newer CPUs 7700k / R1800X the difference is not mega in this title IMO. but will lean towards Intel due to the above,

Whether people agree or not, multi threading is the future trend. its the same story always why do we need more threads when we had dual cores. then we had quad cores and the development of game engines finally started taking advantage. then it was the hyper-threading argument. now look at new engines taking advantage of hyper-threading/SMT.

With gaming Console using more and more cores/threads, newer game engines are being build correctly to utilize threads finally. AMD released a very good multi threaded CPU which will take advantage of these newer titles as someone stated above and even intels newer I9 series will aswell.
 
I have a i5 6600k @4.5 and in all Sims it can get upto around 80% @4k using my GTX 1080 Ti and in games like Watchdogs 2 it is constantly at 100%and really holding the game back.
Am now saving for a 7700k. It cant come quick enough.
 
I was just giving a couple of examples. Games where an i7 (4c8t) crushes i5s (4c4t) simply by virtue of having more threads at the exact same (or even lower) clockspeeds include:

Battlefield 1, Watch_Dogs 2, F1 2016, Hitman, Mafia III, Ashes of the Singularity, Gears of War 4, Overwatch, Total War: Warhammer

That is not an exhaustive list. In most of those games, an older i7 like the i7-4790K is actually faster than the fastest current i5-7600k despite its single threaded performance disadvantage, because of more concurrent threads.

Ok but with simulations especially driving sims it's actually harmful to the game to run more threads,latency is key and as of now,no one has came up with a way to offload the work to more threads or even cores,without having the game become a mess due to one thread waiting for the another thread,longer than the allotted time to run a realtime simulation.

parallel programming across cores and threads just isn't there yet,maybe one day,but from what I've gathered it's not down to game devs to come up with a low latency parallel standard,it's up to the coding geniuses the likes of Microsoft have.
 
Ok but with simulations especially driving sims it's actually harmful to the game to run more threads,latency is key and as of now,no one has came up with a way to offload the work to more threads or even cores,without having the game become a mess due to one thread waiting for the another thread,longer than the allotted time to run a realtime simulation.

parallel programming across cores and threads just isn't there yet,maybe one day,but from what I've gathered it's not down to game devs to come up with a low latency parallel standard,it's up to the coding geniuses the likes of Microsoft have.

I suggest you watch Kunos from AC video, there is a video floating around where he talks about mutlithreading in AC and in older games. its quite informative on modern engine design and draw back etc.

Although AC is not the greatest multi threading engine. It seems to have a better implementation then Rfactor 2. It would be nice to see devs explain how multithreading is implemented in Rfactor in what form as the early days of multi threading software development were very crude implementations .

I believe in AC they have a system where you have on one CPU thread feeding graphics and physics working off 2 threads or more feeding to the first thread at 3ms and no more, you can specify to use more physics thread but then you get a point of diminishing returns i assume.

I believe when you go over this 3ms you are no longer in realtime in the engine, and you have the good old processor occupany over 100% like GP4 days.

I do feel Rfactor is battling an old engine when it comes to this side of performance and i don't see that changing, but it prefer to hear it from a Developer.

Cheers
Daniel
 
I suggest you watch Kunos from AC video, there is a video floating around where he talks about mutlithreading in AC and in older games. its quite informative on modern engine design and draw back etc.

Although AC is not the greatest multi threading engine. It seems to have a better implementation then Rfactor 2. It would be nice to see devs explain how multithreading is implemented in Rfactor in what form as the early days of multi threading software development were very crude implementations .

I believe in AC they have a system where you have on one CPU thread feeding graphics and physics working off 2 threads or more feeding to the first thread at 3ms and no more, you can specify to use more physics thread but then you get a point of diminishing returns i assume.

I believe when you go over this 3ms you are no longer in realtime in the engine, and you have the good old processor occupany over 100% like GP4 days.

I do feel Rfactor is battling an old engine when it comes to this side of performance and i don't see that changing, but it prefer to hear it from a Developer.

Cheers
Daniel

If you run simpler physics you have more leeway,the tyre model Allan in rf2 is demonstrably more complex and resource heavy than dare I say it,every other tyre model in use today.
I heard that ac is using a iteration of rf1s tyre model,which would help explain why they can split physics into two threads without negative impacts on latency,because there are so much less calculations going on etc.
If you are correct about the 3ms buffer in ac then they are well ahead of even iracing,which is think is around 16ms tick rate,and they hold this above all else and will not implement anything that runs higher,because they say the end experience isn't very good at all.

For a 3ms tick rate for physics ac are either genius or running a fairly simplistic physics model

Edit iracings physics tick rate I think is about 7ms while the the graphics is 9,totalling 16ms to draw the image and do the physics
 
Last edited:
A little bit OT, but where we are at AC I´ve got a funny story to tell that happend yesterday:
In my "wekend-loft" at my dads house there stands a AM2+ with Phenom II X3, happily overclocked to 2,3 GHz (LOL) + 460GTX.
For the hope of being able to drive rF2 one day on this machine, a friend gifted me a Phenom II X4 965 3,4 GHz.
Build it in, booted up, clocked for first test on 3GHz, instantly noticed in windows that something was wrong.
Watched Spa 24hrs stream, which stuttered more or less often. Windows System was showing the in BIOS adjusted clock/speed.
Thought: lets try a game! Fired up AC as this runs on the X3 without any problems, game runs normal with new CPU.
Not that buttersmooth as before, but physical experience and behaviour was exact the same as at home with my 2500K and with the X3.
Absolut no difference!
Turned out that Motherboard wasn´t really compatibel with CPU and clocked (fix) it in safe-mode at 800MHz!
Sorry, but AC has nothing under the hood! Nothing!
The physics there are the crappiest joke in sim-history!
 
I believe in AC they have a system where you have on one CPU thread feeding graphics and physics working off 2 threads or more feeding to the first thread at 3ms and no more, you can specify to use more physics thread but then you get a point of diminishing returns i assume.

I believe when you go over this 3ms you are no longer in realtime in the engine, and you have the good old processor occupany over 100% like GP4 days.

I do feel Rfactor is battling an old engine when it comes to this side of performance and i don't see that changing, but it prefer to hear it from a Developer.

Cheers
Daniel

rF2 has one thread for physics, one for graphics. It seems though that physics rarely bottleneck a modern CPU in rF2, as you would need to add a huge number of AI's to begin with. So it's questionable if multithreading the physics would do much for rF2, it definitely wouldn't increase the FPS unless the game is out of realtime. The graphics thread is a better candidate for improvement it seems, as some users with latest Nvidia GPU and running in VR seem to be CPU limited, and that's from the graphics thread. But I'm not aware if any sim today does split graphics tasks over many threads on the CPU. rF2 also has huge PCIe bus usage as I pointed out in another thread here, so the whole graphics engine is probably the main bottleneck for the CPU at the moment.
 
Back
Top