How is simspeed lag in 2018?

[nine2]

what sort of optimisation would break the mods?

[____]

That is a pro level Seton's game at min 36 with (compared to other maps) insane amounts of units and -2 is the lowest we can see

if only

I was referring to Empty's screenshot with that statement, -2 was the lowest there at that point in time (which does not mean it won't get even slower) xD

[Endranii]

No doubt any desktop Ryzen is faster late-game than an Intel-U. Would be interesting to know actual CPU's of other players.

No Turin wasn't playing this time. So no U cpu.

[E8400-CV]

[...offtopic trolling cut out...]

That is a pro level Seton's game at min 36 with (compared to other maps) insane amounts of units and -2 is the lowest we can see, most other players are -1 and it was already stated that most had a 150 rated CPU. I think we can all agree, that most if not all of these guys have a good CPU, probably from Intel.
And aaaagain, the CPU is not the only variable that determines slowdown, some might have chrome open and such...

No we can't, because we don't know at all. I've see some pretty bizarre claims of Ryzen's supposed greatness, and this one of the more absurd ones. You are comparing against... well, an assumption.

In every other single threaded application Ryzen consistently loses vs the Intel's of the same price range, so unless proven otherwise, I'll take that the same holds true for Ryzen's performance in FAF.

[____]

[...offtopic trolling cut out...]

Oh, so your posts were not off topic but mine are. Running out of arguments again?

No we can't, because we don't know at all.

Who is we? I know it, and I am guessing most others here know it as well. You are pretty much assuming that we are stupid, and incapable of just asking those VERY well-known players what CPU they have. Just because you don't know it, doesn't mean we don't.

I've see some pretty bizarre claims of Ryzen's supposed greatness, and this one of the more absurd ones.

It is not a claim to anything, it was just an educated guess based on Empty's screenshot.

You are comparing against... well, an assumption.

No, I am not, and Empty isn't either, since he literally plays with those guys on the same map rather often (note: I do not know like every CPU they have, but I certainly know the general idea of their performance). All you do here is "assuming" about our knowledge of the CPU performance of other players. You are literally doing what you are critizicing me about. N1.

But no, you know better than everyone else, again.

In every other single threaded application Ryzen consistently loses vs the Intel's of the same price range, so unless proven otherwise, I'll take that the same holds true for Ryzen's performance in FAF.

Yea let's ignore EVERY variable exept for clock speed and base our opinion on it. N1 again.

[Endranii]

And aaaagain, the CPU is not the only variable that determines slowdown, some might have chrome open and such...
No we can't, because we don't know at all. I've see some pretty bizarre claims of Ryzen's supposed greatness, and this one of the more absurd ones. You are comparing against... well, an assumption.

In every other single threaded application Ryzen consistently loses vs the Intel's of the same price range, so unless proven otherwise, I'll take that the same holds true for Ryzen's performance in FAF.

Why do you think I posted it here? Maybe beacause I was sure that my ryzen will also be slower than the intel counterparts here? And guess what, I also was baffled when I took a look at sim speed in game just to see that the CPU is holding up better than intel cpu's scoring from 150 to 200 points in cpu bench when my ryzen is 202-205 points? And once again, there was no U CPU in the game, hell as far as I'm aware no one was even using a laptop in this certain match, although feel free to prove me wrong.

As for the external variables, at least when it comes to me I had discord/slack/TS and chrome running in background.

Anyway I'm gonna install FAF on my laptop with 4900MQ and run a faf benchmark and then seton replay at the same time as on my desktop and record time it takes to finish it at +10.


Everyone knows that intel reigns supreme in IPC(not so much at the moment)/Clocks as such it is expected to have faster single core performance, but who knows if faf actually doesn't somehow benefit from ryzen Uarch.

[E8400-CV]

[...offtopic trolling cut out...]

[...offtopic trolling cut out...]

No we can't, because we don't know at all.

Who is we? I know it, and I am guessing most others here know it as well. You are pretty much assuming that we are stupid, and incapable of just asking those VERY well-known players what CPU they have. Just because you don't know it, doesn't mean we don't.

Well, share the info...

I've see some pretty bizarre claims of Ryzen's supposed greatness, and this one of the more absurd ones.

It is not a claim to anything, it was just an educated guess based on Empty's screenshot.

You are comparing against... well, an assumption.

No, I am not, and Empty isn't either, since he literally plays with those guys on the same map rather often (note: I do not know like every CPU they have, but I certainly know the general idea of their performance). All you do here is "assuming" about our knowledge of the CPU performance of other players. You are literally doing what you are critizicing me about. N1.

But no, you know better than everyone else, again.

In every other single threaded application Ryzen consistently loses vs the Intel's of the same price range, so unless proven otherwise, I'll take that the same holds true for Ryzen's performance in FAF.

Yea let's ignore EVERY variable exept for clock speed and base our opinion on it. N1 again.

I never said anything about clockspeed, but single threaded performance. Different architectures do different things. Want a Pentium 4 at 4 GHz? Nope.

And aaaagain, the CPU is not the only variable that determines slowdown, some might have chrome open and such...
No we can't, because we don't know at all. I've see some pretty bizarre claims of Ryzen's supposed greatness, and this one of the more absurd ones. You are comparing against... well, an assumption.

In every other single threaded application Ryzen consistently loses vs the Intel's of the same price range, so unless proven otherwise, I'll take that the same holds true for Ryzen's performance in FAF.

Why do you think I posted it here? Maybe beacause I was sure that my ryzen will also be slower than the intel counterparts here? And guess what, I also was baffled when I took a look at sim speed in game just to see that the CPU is holding up better than intel cpu's scoring from 150 to 200 points in cpu bench when my ryzen is 202-205 points? And once again, there was no U CPU in the game, hell as far as I'm aware no one was even using a laptop in this certain match, although feel free to prove me wrong.

As for the external variables, at least when it comes to me I had discord/slack/TS and chrome running in background.

Anyway I'm gonna install FAF on my laptop with 4900MQ and run a faf benchmark and then seton replay at the same time as on my desktop and record time it takes to finish it at +10.


Everyone knows that intel reigns supreme in IPC(not so much at the moment)/Clocks as such it is expected to have faster single core performance, but who knows if faf actually doesn't somehow benefit from ryzen Uarch.

So what were the actual counterparts...?

Also... you do know the bench number can be manually replaced? And; if you OC, run the bench, shut down, remove OC... it still shows the low bench score. That's part of the problem.

[SpoCk0nd0pe]

In every other single threaded application Ryzen consistently loses vs the Intel's of the same price range, so unless proven otherwise, I'll take that the same holds true for Ryzen's performance in FAF.

Quote from Uveso from the Github link you provided on page 2:

Well, i am saying this in all benchmark discussions.
CPU power is no longer our main problem. (Since we have fast CPU's like Intel I7 etc)
As already stated; If the game goes under +0 simscore then all CPU's are still under 25% load. (Intel I7 at least)
This means, we don't have a problem with calculation speed.

I did some Hardwaretest on my PC's that convinced me, that memory (RAM) speed is way more important in these days.

So i suggest to add memory test's like tabledeepcopy etc. to get a propper PC rating.

It may be as I suspected: Sim speed does not seem to be limited by single thread performance, it seems to be limited by memory hierarchy performance. Since the 2600x comes with 16 mb of L3 cache, being faster then a 8mb cache I7 is quite realistic. Single core benches like pi fast mean squat in this context.

Erandii, maybe you could provide some data on overclocked ram speeds. Since this reduces ccx and memory latency, this could make quite the difference.

Maybe the 5775C is the best CPU for SupCom.

[E8400-CV]

In every other single threaded application Ryzen consistently loses vs the Intel's of the same price range, so unless proven otherwise, I'll take that the same holds true for Ryzen's performance in FAF.

Quote from Uveso from the Github link you provided on page 2:

Well, i am saying this in all benchmark discussions.
CPU power is no longer our main problem. (Since we have fast CPU's like Intel I7 etc)
As already stated; If the game goes under +0 simscore then all CPU's are still under 25% load. (Intel I7 at least)
This means, we don't have a problem with calculation speed.

I did some Hardwaretest on my PC's that convinced me, that memory (RAM) speed is way more important in these days.

So i suggest to add memory test's like tabledeepcopy etc. to get a propper PC rating.

It may be as I suspected: Sim speed does not seem to be limited by single thread performance, it seems to be limited by memory hierarchy performance. Since the 2600x comes with 16 mb of L3 cache, being faster then a 8mb cache I7 is quite realistic. Single core benches like pi fast mean squat in this context.

Erandii, maybe you could provide some data on overclocked ram speeds. Since this reduces ccx and memory latency, this could make quite the difference.

Maybe the 5775C is the best CPU for SupCom.

Sounds interesting, but if so; at what point does it become a memory limit? Because the -U CPU's don't cut RAM frequency when they decrease core clocks. At least, not that I know.

[IceDreamer]

It will depend on load type. I've been doing some performance testing again the past three days and I'm beginning to come to the same conclusion as others: The simspeed is dipping into negative territory despite none of my CPU cores being over 40%, and this is on a laptop (i7 4720HQ @ 3.6). The only way I can spike the CPU load to 100% is to select a few hundred units with range rings turned on, which I know to need a huge amount of math calculations. The rest of the time though, it's slowing down despite very little core load. At the same time, RAM and cache use are very high indeed.

Now I'd need some expert input, but IF it is true that metatable lookups and big table copy or move operations use memory and cache a lot, this would be a big reason for the observed low core use, because LUA lookups are really, really slow, and yet the FA LUA is doing a massive, massive number of LUA lookups. I shall know more in due course.

Also, I fully plan to purchase a 2700X, and I will at that time also have access to an 8700K. I'll be able to publish comprehensive relative speeds between those two, and this 4720HQ.