Intel Core i7 3930K 4.8GHz w/FSX Quick Review

[SpiritFlyer 987]

 

I wonder what benchmarks would look like comparing other sandy bridge processors with relative speeds to the 3930K

 

We will find out shortly without having to guess as soon as we have a few SB-E data sets with the FSXMark11 Benchmark tests. It is amazing how well it has taken speculative subjectivity out of the comparitive performance equation. The results so far presents a compelling case for Sandy Bridge, whatever the model. It will be interesting to see what tests on several of the new extreme SB samples will show. Many people think that the next big jump for FSX will be Ivy Bridge that is just mere months away.

 

Kind regards,

[worldclassleader 0]

"Here you go: "ProcSpeed=9267"I think that speed is from before you overclocked, could you delete just that line from the config and let it load again, it should be much higher.It won't make the whole config rebuild, it'll just redo that line.

Here you go: "ProcSpeed=10425" If ithelps to know, I have Intel SpeedStep enabled as well, so perhaps that's why the number is low, as the processor runs at about 1600MHz when idle.

[Kosta 122]

Here you go: "ProcSpeed=10425" If ithelps to know, I have Intel SpeedStep enabled as well, so perhaps that's why the number is low, as the processor runs at about 1600MHz when idle.

Weird, mine is ProcSpeed=19707 with OC to 4.8. SpeedStep could be the reason though - isn't ProcSpeed taken into consideration when doing some calculations in FSX? And I have C1E enabled, but not SS.And it would be completely awesome if you would run FSXMark11 on this CPU with the intended settings! That way we could definitely say how much this one is better in terms of CPU speed... additional cores bring texture loaders, the better question is though, and this is what you claim, is more or same FPS with higher settings.

[worldclassleader 0]

Hello Jacob,I am happy to see that the new CPU seems to be performing as much as you expect it to.Would you be so kind as to take the time to run the standard FSXMark11 Benchmark tests that have accumulated significant objective comparative data on CPU/GPU packages, as it directly relates to FSX? If so, please insure that you follow the detailed setup instructions exactly as outlined, so that any results will be consistent with the established objective testing parameters. This will help insure that we can add and track the new CPU's exact FSX performance as other SB-E examples are tested, as they relate to each other and other SB versions, as well as different and previously tested setups. Please go to the AVSIM hardware section for more information at http://forum.avsim.n...other-hardware/Looking forward to reviewing your results as they compare with other similiar units that will be tested in the coming days and weeks. Thanks.Kind regards,

Hi Stephen,Thanks for your feedback and sharing the FSXMark11 benchmark software. Well, I've got some interesting results. First, here are my results using default settings with no add-ons: Min Max Avg Test 1: 23 60 37.78Test 2: 22 62 47.323Test 3: 30 60 46.833Test 4: 29 62 46.613What's really intersting is that my average was about 5FPS higher using REX 2.0 and Ground Environment Extreme Enhanced (North America). Also, I experimented with various Affinity Mask's, such as 62, 63, 3549, 4095, and 2730, but 1344 gave me the best results as noted in the benchmarks above.Lastly, I've used many processors to test FSX, including most recently, Intel's i7 920/930, 950, 970, 990X; and from Sandy Bridge (SB) the 2500K, 2600K, and 2700K. Compared to SB, the 3930K "feels" more fluid at the highest [density] settings and eye-candy (e.g., flare, shadows, etc) at 1920x1080. Therefore, I don't believe the FSXMark2011 benchmark is valid because the 3930K is superior to the 2600K at 5Ghz in demanding settings; at least in my experience. Perhaps there are tweaks I'm not aware of (?) that facilitate higher FPS, as some of the 2600K benchmarks on this website exceed my results; most notably due to higher Ram speeds and tighter timings.Here are the modifications I've made to my config file:[bufferPools]UsePools=1RejectThreshold=98304PoolSize=8388608HIGHMEMFIX=1AffinityMask=1344I hope this helps..and please feel free to contact me regarding any tips you'd like to share or if you need me to explore different test scenerios.Best,Jacob Edited November 28, 2011 by worldclassleader

[worldclassleader 0]

Weird, mine is ProcSpeed=19707 with OC to 4.8. SpeedStep could be the reason though - isn't ProcSpeed taken into consideration when doing some calculations in FSX? And I have C1E enabled, but not SS.And it would be completely awesome if you would run FSXMark11 on this CPU with the intended settings! That way we could definitely say how much this one is better in terms of CPU speed... additional cores bring texture loaders, the better question is though, and this is what you claim, is more or same FPS with higher settings.

Compared to the 2600K at 5GHz, the 3930K "feels" much more fluid to me; there's no question. However, I'm using DDR3 1600MHz ram and based on some of the other 2600K FSXMark2011 results, it appears that faster ram with tighter timings has a significant [positive] effect on FPS. Therefore, I think it's possible to net even more performance out of this CPU.Also, I can't explain why the FSX config file reflects a different "ProcSpeed" number. I'm going to try disabling SpeedStep tomorrow and see what that brings.I just posted my FSXMark2011 results in this post as well.

[Kosta 122]

I just posted my FSXMark2011 results in this post as well.

In which post?Be sure to post here: http://forum.avsim.n...9116-fsxmark11/Btw. what Affinitymask are you running now, and is HT enabled or disabled?

[worldclassleader 0]

In which post?Be sure to post here: http://forum.avsim.n...9116-fsxmark11/Btw. what Affinitymask are you running now, and is HT enabled or disabled?

Thanks...posted the results in that link. Checkout the benchmark results; I ran several different 'affinity mask" tests, but a decimal value of 1344 was best, and HT was enabled.

[dazz 69]

I'm sorry, but it doesn't make sense to me. With a 1344 affinity mask you are using 3 cores and core 0 for fibers, exactly how AF=14 works in quad cores. So you are not even using all 6 cores in FSX, if that gives you the best results (and it should - thanks Bojote) only goes to show what we already knew. More cores won't make a difference in FSXThose FSMark11 results are self explanatory, there's no difference with SB, and "it feels more fluid" is something we've all heard before.

[Kosta 122]

I'm sorry, but it doesn't make sense to me. With a 1344 affinity mask you are using 3 cores and core 0 for fibers, exactly how AF=14 works in quad cores. So you are not even using all 6 cores in FSX, if that gives you the best results (and it should - thanks Bojote) only goes to show what we already knew. More cores won't make a difference in FSXThose FSMark11 results are self explanatory, there's no difference with SB, and "it feels more fluid" is something we've all heard before.

I must completely agree to this one. Using 1344 on six-core CPU will give you the *same* result as the full quad core usage on conventional SB, 2x00K.Beside that, your scores are below what is usual for 2500/2600K chips.The more fluid, I fully agree with Dan. That's something noone can measure and is situation dependant (and your human situation too, including reaction of the eyes and all that).The only thing better you might achieve with this CPU is faster texture loading if your storage is allowing for it. That is also including flight-internal operations like texture loading, though they don't need more than 2 cores, and you won't see any real difference here.You have a nice chip there, yes. Set affinitymask to 1364 (if HT enabled) and enjoy it!*if HT is disabled, AM 62, which I recommend even (HT is useless for FSX)

[simon747 185]

My 920 3.8ghz with triple channel ram was smoother in low teens than the 2600k 4.6 in low teens thats for sure!! That is the only time it is smoother than the SB system. The 2600K allows me to use the NGX, for example in an add-on airport, sliders maxed with heavy ai (no way could I do that in the 920) and runs great, but if fps dip into low teens it is not as smooth as my previous platform when fully loaded. Thats for sure for me. But I am HAPPY with the 2600K over all as I RARELY get low teens!!! I would assume that quad channel and 2 cores would help smoothness. But this is all subjective and hard to judge, but I would assume with a loaded sim scenario two cores loading texture files from an ssd and two more channels of RAM would help the perception of smoothness. Not many 6 core cpu's will do 4.8 ghz though, and damn its expensive!

[worldclassleader 0]

I'm sorry, but it doesn't make sense to me. With a 1344 affinity mask you are using 3 cores and core 0 for fibers, exactly how AF=14 works in quad cores. So you are not even using all 6 cores in FSX, if that gives you the best results (and it should - thanks Bojote) only goes to show what we already knew. More cores won't make a difference in FSXThose FSMark11 results are self explanatory, there's no difference with SB, and "it feels more fluid" is something we've all heard before.

Same here...I can't explain it either, but having used [several] 2600K's at 5GHz and 980x/990x with the exact same [density] settings on many builds for customers in the Chicagoland area this past year, the 3930K is definitely my choice for FSX today. I don't discount the validity in your statement either Dazz; I just cant explain the performance increase...I just can't. Plus, I don't know how FSX works "under the hood," other than setting the Affinity Mask.=)

[dazz 69]

Same here...I can't explain it either, but having used [several] 2600K's at 5GHz and 980x/990x with the exact same [density] settings on many builds for customers in the Chicagoland area this past year, the 3930K is definitely my choice for FSX today. I don't discount the validity in your statement either Dazz; I just cant explain the performance increase...I just can't. Plus, I don't know how FSX works "under the hood," other than setting the Affinity Mask.=)

well, who knows. I may be missing somethingDo you happen to own the PMDG744? I have another benchmark with it we could try if you want

[worldclassleader 0]

I must completely agree to this one. Using 1344 on six-core CPU will give you the *same* result as the full quad core usage on conventional SB, 2x00K.Beside that, your scores are below what is usual for 2500/2600K chips.The more fluid, I fully agree with Dan. That's something noone can measure and is situation dependant (and your human situation too, including reaction of the eyes and all that).The only thing better you might achieve with this CPU is faster texture loading if your storage is allowing for it. That is also including flight-internal operations like texture loading, though they don't need more than 2 cores, and you won't see any real difference here.You have a nice chip there, yes. Set affinitymask to 1364 (if HT enabled) and enjoy it!*if HT is disabled, AM 62, which I recommend even (HT is useless for FSX)

Thanks for the tips Word Not Allowed! I just ran the FSXMark2011 benchmark having applied the 1364 Affinity Mask you suggested and the test results for all four tests were identical to my initial tests using an affinity mask of 1344. So...go figure....lol...=)

well, who knows. I may be missing somethingDo you happen to own the PMDG744? I have another benchmark with it we could try if you want

I'd be glad to give it a shot! I own the following FSX PMDG add-ons:MD-11NGX-737 800/900747x and the 8i expansion model

My 920 3.8ghz with triple channel ram was smoother in low teens than the 2600k 4.6 in low teens thats for sure!! That is the only time it is smoother than the SB system. The 2600K allows me to use the NGX, for example in an add-on airport, sliders maxed with heavy ai (no way could I do that in the 920) and runs great, but if fps dip into low teens it is not as smooth as my previous platform when fully loaded. Thats for sure for me. But I am HAPPY with the 2600K over all as I RARELY get low teens!!! I would assume that quad channel and 2 cores would help smoothness. But this is all subjective and hard to judge, but I would assume with a loaded sim scenario two cores loading texture files from an ssd and two more channels of RAM would help the perception of smoothness. Not many 6 core cpu's will do 4.8 ghz though, and damn its expensive!

That's really interesting; I wonder if the Triple-Channel memory on the X58 platform has a positive effect on FSX, compared to SBs dual channel? Ofcourse, then we got SB-E with Quad-Memory. I'm assuming this has been addressed some where on AVSIM as well..

[dazz 69]

Ok, thanks Jacob.The test is in Heathrow, with heavy traffic. Do you have any EGLL scenery, GEX/UTX Europe or Traffic addon like UT2 or MyTrafficX?

[worldclassleader 0]

Ok, thanks Jacob.The test is in Heathrow, with heavy traffic. Do you have any EGLL scenery, GEX/UTX Europe or Traffic addon like UT2 or MyTrafficX?

Sounds good Dazz....nope, I don't own any of the addon's you mentioned above.