Is it worth to upgrade CPU?

[spartansociety]

So you just repeated what I said, but veiled it as a disagreement?

Ivy crushes this game as much as a Skylake does. The bottleneck is literally the engine. A Skylake might see minor improvements but not substantial ones. You're much better off getting a Z170/Z270 chipset for the modern interface ports than for any hoped for improvements in late game FPS drop.

What im saying is IVY will not 'crush the game', it is likely the reason for the slowdown. Upgrading to a 7700 that does 5ghz will result in an improvement and might be worth it for someone with who is happy to put decent dollars into his PC. This is especially the case now that Intel has dropped their prices after the Ryzen release. I know I certainly saw an improvement in several paradox games going from a 2500K to a 6700k (partly a better clock, partly IPC, possibly also the reformat and faster ram).

I would agree that for most people the better idea it just to wait for a few more updates or play slightly smaller maps.

 

[Kat Tsun]

What im saying is IVY will not 'crush the game', it is likely the reason for the slowdown. Upgrading to a 7700 that does 5ghz will result in an improvement and might be worth it for someone with who is happy to put decent dollars into his PC. This is especially the case now that Intel has dropped their prices after the Ryzen release. I know I certainly saw an improvement in several paradox games going from a 2500K to a 6700k (partly a better clock, partly IPC, possibly also the reformat and faster ram).

I would agree that for most people the better idea it just to wait for a few more updates or play slightly smaller maps.

Without wealth of benchmarks (PDS games are not known for being benchmark staples) we can't be certain, honestly. I am simply pessimistic about the ability of a CPU upgrade to provide any real performance improvements.

If you have data that says otherwise, I'd be happy to change my opinion on the matter.

Other than that, I can offer 3d stats measurements between my AM2+ Thuban and AM4 Raisin, perhaps next weekend, if everything goes as planned. Given the trainwreck that is Bulldozer, the Thuban should be a reasonable substitute for it in performance.

 

[spartansociety]

1. That stuff has been optimised for multiple threads because they tend not to have dependencies.It really has nothing to do with the purpose. It is a tad easier to design a program that converts video using multiple threads than it is to designed a real-time game where many events rely on knowing what has already happened. Also, I was referring to Naples which is clearly a server chip.
http://www.digitaltrends.com/computing/amd-naples-server-cpu-radically-changing-datacenter/

2. It is not close to 40 percent behind. In a game where the GPU is not the most important thing, you will see a difference at all resolutions. You are basing your comments on forums and reviews which look at FPS games, and maybe a Total War. The advantage of low resolutions is that it removes the GPU bottleneck in games which are GPU bound, thus making it easier to see differences in performance. Gaming performance will also improve over the next few weeks on Ryzen after a windows patch etc (windows is alternating the use of cores in a way which hurts Ryzen atm).

3. I heard they were making them at a Samsung fab as well.

 

[Kat Tsun]

1) We're not talking about servers, even if AMD has been pretty desperate to dislodge Nvidia and Intel dominance in that market, though. We're talking about desktops. Keep the discussion about Summit Ridge, not hypothetical releases given AMD's stock has somehow managed to plateau even this early.

2) I said "20-40%", not "40%". In general it's 10-20% slower than Kaby Lake on 1080p, sometimes as much as 40%, which puts it somewhere near Ivy Bridge in game performance. Ditto for singlethreaded performance on average. This says more about the state of CPUs (or Intel's inability to achieve higher gains in performance) as a means of boosting FPS, than it does Ryzen's under performance. It says that Ivy Bridge is still competitive, which benchmarks tend to bear out.

I'm not sure where FX 8150 lands on that, the only benchmarks people use are FX 8370, but it's probably comparable to an AM2+ 1090T, which is still competitive today in most titles, including Stellaris.

3) It must have changed since they renamed it from Zen, then. Last I heard they'd wanted Samsung but couldn't get them because of some weird contractual thing with GF, but GF managed to licence the FinFET process from Samsung. Now it seems they have both Samsung and GF fabbing their chips apparently.

 

[spartansociety]

Without wealth of benchmarks (PDS games are not known for being benchmark staples) we can't be certain, honestly. I am simply pessimistic about the ability of a CPU upgrade to provide any real performance improvements.

If you have data that says otherwise, I'd be happy to change my opinion on the matter.

Other than that, I can offer 3d stats measurements between my AM2+ Thuban and AM4 Raisin, perhaps next weekend, if everything goes as planned. Given the trainwreck that is Bulldozer, the Thuban should be a reasonable substitute for it in performance.

I would be interested in doing some myself. Different core count, ram speed etc. The problem is paradox games tend to be so hard to benchmark. I don't really know how to create a reproducible benchmark, as the game will change every time. Maybe I could create a universe covered with my empire and time the amount of time it takes for a year to pass on the highest speed. If I get the chances I might see how accurate this is over the weekend.

It would be interesting seeing how my 6700k goes against your new CPU.

 

[The Founder]

Someone doesn't play on a high enough resolution, then.

CPUs rarely increase performance in games substantially. You will always be better off upgrading to a new GPU, unless your mainboard doesn't support it, in which case you'll need a new CPU, unless by happenstance your new mainboard supports your old socket I guess. They're basically tertiary upgrades after mainboard and GPU if you want to increase your frames per second.

GPUs are almost always the bottleneck in frames. Games simply aren't big enough number crunchers to tax CPUs very much, unless you have a strange case like you're crunching PhysX on your CPU instead of your GPU (for whatever silly reason). Stellaris isn't a very intense game anyway. An Ivy Bridge will crush it into submission as much as a Skylake in any realistic measurement, so a new CPU is not going to be a panacea for any sort of end game FPS drop. That's inherent to the version of Clausewitz engine that Stellaris runs on.

I only run the game on Maximum settings at 1366x768 Resolution and get stable 60 FPS.
Stellaris is not anywhere near GPU taxing enough for the GPU to be the bottleneck. Usually it is not, but in this case. The Graphics are kept simple and non-taxing by design.
If they drew more then 50 missiles and thier trails, there might be a case for the GPU being sometimes taxed. But they are smart enough to have avoided even that.

Btb, I actually made basic games in the XNA Framework. And I can tell you that if the Update Routine runs to long it can easily delay the Draw Routine from firing and thus drop Framewrate. Like you can see in the Loading screen if you have FPS counter enabeled when there is no way in hell the GPU could be taxed (just the usual delays slowdowns from waiting on a HDD providing data dropping it to 1 FPS).
In the end the Draw and Update Routine always have to share one thread. And sometimes you just have to make a blocking call.