FSXMark11

[SAAB340 28]

   Hello everyone ! I just came  back to FSX after a long time and  the obviously conclusion is  I need to upgrade. Right now I have a xeon E5450 running at 3.6 Ghz. Average score in fsxmark is 21.5. Not a good playing experience with addons.

 

   First, considering money is problem, will be i5 4690k enough, or I should make the effort for i7 4790K ? I am truing to figure, on same system, what is the gaining in FPS if using i7.

   Memory should be anyway at 2400 speed. I have a GTX 650, planning to upgrade for GTX 960.

   Any advise is welcome !

Welcome back, and welcome as a poster finally. I know its been 7 weeks since your post so you might already have built a new system. There is little/no difference in FPS between an i5 and an i7 at the same clock. Hyperthreading on the i7 can be used for loading in photo scenery slightly quicker in FSX, so i5 is the way to go on a budget.

 

Skylake has now been launched and I'd go for the i5 6600K with 3000mhz DDR4 RAM and a Z170 motherboard unless you are on an extremely tight budget. DescendDescend has now shown that there is a small improvement going for Skylake over Haswell. As you'll have to upgrade all of CPU, MOBO and RAM you might as well go for the latest platform as well.

 

As far as the GPU goes, I'd do the CPU, MOBO and RAM upgrade first if you're really tight on money. The GTX 960 is probably a great match for FSX though, and most other games as well if you are on a 1080p 60fps display.

[WebMaximus 372]

I'm considering upgrading to Skylake too and reading the suggestion in the post above to go with 3000MHz RAM is there a special reason for that?

 

Reason I ask is I read everywhere that it's highly unlikely you'll see any difference going from let's say 2400->3000MHz under normal usage. If you on the other hand are looking for setting records in memory speed benchmarks you would probably want the highest speed possible which I think for the time being is somewhere around 3800 MHz.

[WebMaximus 372]

Here's an example where you can see what difference it will make going from 2133-3200MHz.

 

http://www.guru3d.com/articles_pages/msi_z170a_gaming_m7_review,16.html

[mpknemo 0]

I'm considering upgrading to Skylake too and reading the suggestion in the post above to go with 3000MHz RAM is there a special reason for that?

 

Reason I ask is I read everywhere that it's highly unlikely you'll see any difference going from let's say 2400->3000MHz under normal usage. If you on the other hand are looking for setting records in memory speed benchmarks you would probably want the highest speed possible which I think for the time being is somewhere around 3800 MHz.

 

It's true, but FSX is not normal usage :wink:

 

Most benchmarks show that FSX benefits from faster memory speeds, and it translates into better fps. Check out post #666(!) for example.

 

Markus

[WebMaximus 372]

It's true, but FSX is not normal usage :wink:

 

Most benchmarks show that FSX benefits from faster memory speeds, and it translates into better fps. Check out post #666(!) for example.

 

Markus

 

Well, that FSX (and maybe also P3Dv2?) shouldn't be considered as normal usage is true :wink:

 

However looking at the post you referred to the memory speeds that were compared were much lower where the top speed was 2400MHz. That you will see a difference between let's say 1600->2400MHz is something most will agree with. However when you start going above 2400MHz I don't think you will see that much of a difference anymore.

 

Then it's not all about speed but also timings. For instance it would be really interesting to compare 3000+ MHz CL16+ RAM with for instance 2400MHz CL14.

[TechguyMaxC 516]

Here's an example where you can see what difference it will make going from 2133-3200MHz.

 

http://www.guru3d.com/articles_pages/msi_z170a_gaming_m7_review,16.html

 

Memory performance is only relevant when memory-bound, and none of those tests are memory-bound.  If they tested FSX/P3D, they would definitely see a difference attributable solely to memory speed/latency.  FSXMark has shown us that, time and time again.  

[TechguyMaxC 516]

Has anyone run the test lately with a 355 version driver?  I'm getting results less than my original Ivy Bridge build right now with a faster CPU, faster RAM, and a faster GPU.  I can only attribute that to some kind of driver problem...

[DescendDescend 25]

If your original results are the top one on the excel sheet then maybe you didnt set wideaspect=true when doing them as they seem slightly inconsistent with the machines beneath yours. Maybe your current test more accurately reflect the true performance. Just speculating.

[TechguyMaxC 516]

If your original results are the top one on the excel sheet then maybe you didnt set wideaspect=true when doing them as they seem slightly inconsistent with the machines beneath yours. Maybe your current test more accurately reflect the true performance. Just speculating.

I went back and re-tested multiple times after checking the config file.  Even with a 3 FPS difference due to wideviewaspect being misconfigured my current setup should be a good 20% faster than the old one with a newer CPU (Haswell vs. Ivy Bridge), 300MHz higher CPU speed (5GHz vs. 4.7GHz), and faster RAM (2400 vs. 2133).

 

Also, I'm not just comparing to my own old results, I've compared to other people's as well.  There's a 4670k system @ 4.6GHz on page 45 that is producing almost exactly the same results as I'm seeing now, and I've got an extra 400MHz on the CPU and 800MHz on the RAM.  No way is that accurate.  Some of Hasse's tests that got FPS in the 50's are more like what I would expect to see.  I've already confirmed the CPU speed and even disabled Speedstep to ensure Turbo was working at the full 5GHz.  

 

There are a lot of possibilities here, I'm running a newer card with a newer driver than just about anyone who's ever run this test.  For all we know, the latest Nvidia drivers messed something up.  One other thing I observed is that a few weeks ago I got X-Plane on sale through Steam and I immediately set about benchmarking it.  The first few runs were showing great performance and then, all of a sudden the performance dropped from say 260FPS to 190FPS and I could never get it back up again.  3dmark, Furmark, and CPU stress testing performance haven't dropped off though so it's not a system-wide issue, just sims.  

 

Other possibilities to look at:

I'm on Windows 10, having upgraded from 8->8.1->10 with no clean install in between (though realistically that shouldn't cause a problem, it's not like we're talking about XP or Vista here)

I run a 3d monitor, however I turned off 3d in the Nvidia control panel.  

I have HT on where just about everyone else had it off, including my own 3770k tests.  

I have a 2nd monitor and have disconnected it before running FSX but that could be an issue as well.  

 

I can test all these things myself, I am a technician so it's not that hard to go through the possibilities one by one and eliminate potential causes.  I'm looking for a specific answer to my question to see if other users have experienced a similar problem.  

[TechguyMaxC 516]

Have tested now with wideview off and on, definite increase with it off compared to what I was seeing when I made my first post, but still lower performance than it should be.  Unless of course literally EVERYONE that's tested recently has actually had wideview off...  But even then Idahosurge's 4.6GHz 4770k results on page 44 are nearly identical to what I see now except my maximum FPS is higher.  I've got 400MHz on him, there's no way my system isn't being held back by something.  

 

I re-tested with HT off and affinitymask @ 14 instead of 84, virtually identical performance so I'm crossing HT off the list.  Have deleted and re-created the config file and ran it through Venetubo's site again, no difference (except with wideview off).  

[dincusor 0]

  Hello SAAB340, thanks for replying. The new skylake line is the way to go, foe sure. I am waiting to see some results for the i5 6600k. I will have the money prepared for black Friday. And thank you all for sharing your experience!

[westman 257]

Results from Skylake 6700K at 4.9Ghz and 5.0ghz, watercooled customloop avg load temp 60C

 

Run fast mems 3733 mhz Cl 17-16 give approx 4fps compared to a 3000mhz kit

 

The skylake is a monster compared to Haswell even Haswell-E

 

Test a Asus 580 Matrix-P get very GPU bound , but it show the diff to the 2700k with same GPU the diff in min fps is approx 8fps.

 

The other tests a Special Asus 980 Strix ( engenering bios ) 1500mhz gpu /2000mhz mem.

 

Settings by the book and set wideaspect=true

 

 

Hardware Configuration:

 

CPU: Intel 6700K@4.9Ghz

 

RAM: 2x4GB DDR4 3733 CL17

 

GPU: GTX 580 Matrix P

 

HDD: Samsung SSD

 

Frames, Time (ms), Min, Max, Avg

 

16193, 300000, 39, 68, 53.977

 

16045, 300000, 39, 66, 53.483

 

15999, 300000, 39, 65, 53.330

                               AVG 53.6

 

 

Hardware Configuration:

 

CPU: Intel 6700K@4.9Ghz

 

RAM: 2x4GB DDR4 3733 CL17

 

GPU: GTX 980 Strix OC

 

HDD: Samsung SSD

 

Frames, Time (ms), Min, Max, Avg

 

18991, 300000, 39, 85, 63.303

 

18722, 300000, 39, 86, 62.407

 

18572, 300000, 38, 86, 61.907

                              AVG 62.54

 

 

Hardware Configuration:

 

CPU: Intel 6700K@5.0Ghz

 

RAM: 2x4GB DDR4 3733 CL16

 

GPU: GTX 980 Strix OC

 

HDD: Samsung SSD

 

Frames, Time (ms), Min, Max, Avg

 

2 19400, 300000, 39, 86, 64.667

 

3 19333, 300000, 38, 86, 64.443

 

4 19141, 300000, 39, 87, 63.803

                                      AVG 64.3

 

 

 

[TurboKen 221]

 , but it show the diff to the 2700k with same GPU the diff in min fps is approx 8fps.

 

 

 

Hi westman,

 

Is the Approx. 8 FPS from 2700k to 6700k at the same clock?

 

I need to run this again to see how mine compares, as I just upgraded from from a 2600k to a 4690k, definitely smoother.

[westman 257]

Hi westman,

 

Is the Approx. 8 FPS from 2700k to 6700k at the same clock?

 

I need to run this again to see how mine compares, as I just upgraded from from a 2600k to a 4690k, definitely smoother.

6700k @4.9 and 2700k @5.2 the difference at min fps is 8fps , Max fps is the same you bottleneck a Gtx 580 with 5ghz 2770k and approx 4.3-4.4 ghz 6700k.

 

You se the difference with 980 min fps is the same as with 580 but the Max is close to 20fps higher

[gboz 157]

What do you think of incorporating this into FSXMark11?

 

http://www.avsim.com/topic/476198-free-tool-to-find-your-fps-drops/

 

gb.