They may be right with more games in the future to make good use of more cores. but at the moment: Nope.

So why would I buy a CPU TODAY which may (or may not) be good in the future? :-)

 

I look at gaming benchmarks, see that the Intel 4 cores are the best at the moment - so I buy it.

6 or more cores may be good for other stuff - but gaming: 4 Cores with high clock speed.

 

But if someone is on a budget and wants the latest thing: Go with Ryzen, sure not the worst idea.

 

I really don't get this argument here- It seems to me like some people always think that their setup is the ULITMATE setup for EVERYTHING. This will never be the case regarding PCs as the usecases are so different and software is optimized in different ways.

 

And regarding the title "CPU choice for P3D now and in the future" - then the answer is clearly 7700K.

2 hours ago, swiesma said:

And regarding the title "CPU choice for P3D now and in the future" - then the answer is clearly 7700K.

And that is exactly what I like to hear 

4 hours ago, swiesma said:

And regarding the title "CPU choice for P3D now and in the future" - then the answer is clearly 7700K.

Let's assume you can foretell the next 5 years in P3D's development, which you can't of course beyond speculation.   And let's say you are comparing 7700K against SkylakeX-6.   You already know how 7700K will accommodate overclocking, so in this hypothetical we will assume SkylakeX-6 can overclock to 4.4Ghz and 7700K 4.9Ghz.  For single-core performance P3D, how much better will 7700K perform in this hypothetical set of overclocks?

I've maintained since it really matters most in the most demanding parts of the sim, when frame rates start to drop below 30 for example, a 10% increase in CPU performance will get you somewhere around 10% greater CPU-related performance.  This means if you're hammered at a super complex portion of a given flight plan down to a frame rate say of 22 w/ marginal smoothness, you're going to get that up to around 24 with slightly less marginal performance, emphasis on slightly.  This becomes hardly significant, and especially so when you start to factor in terrain loading performance which is coming out of extra cores, as well as concomittantly running other applications which can be assigned off of any cores P3D is playing in.   So at some point you're going to get to a point where more cores indeed outperforms less cores in the scenario described above.  Not in raw performance of the main threads for P3D, but in the total experience.

This being said if P3D goes 64 and is able to use much more VAS to pre-load textures before realtime commences during flying more cores may matter less.  I'll wait and see what Skylake X brings if anything in terms of overclockability and side-by-side single threaded compared to 7700K.  From what I gather Skylake X may be the only practical Intel 6+ core coming along for the next few years, which is why I us it in the above discussion.

11 hours ago, JamesHongKong said:

By the way, this is what I have settled on for my build:

• Intel Core i7-7700K 4.2GHz Quad-Core Processor (5.1Ghz, delidded from Silicon Lottery)

James, when you get it up and running post your analysis eh?  Unfortunately it will be nigh to impossible to assess how much total improvement came from the massive increase in GPU performance versus the relatively small performance change of your 7700K over your 3930K.  Overall, you should experience a very big total improvement.  Looking at single-core PassMark your SB-E over your new KL-K you could see close to a 20% improvement, and that starts getting meaningful for sure.  Add the 1080Ti and you will be enjoying quite an upgrade.  I don't see this level of upside in the 6700K extrapolated out to Skylake-X and that really has to do w/ how well SB-E overclocked compared to anything more recent (base clock to highest overclock, i.e.), so indeed you may have the best possible option for the next several years.  Let's hope P3D can exploit the massive bump in VAS.

So you still have to de-lid these chips still?  Remarkable.   I was pleased w/ SB-E to lose the need to dedicate transistors to GPU function as well as avoiding the need to de-lid.  Pretty amazing 5y later to be looking at only a 20% bump, however as I say it's meaningful when you can have decent performance w/o the bump. 

11 hours ago, Noel said:

James, when you get it up and running post your analysis eh?  Unfortunately it will be nigh to impossible to assess how much total improvement came from the massive increase in GPU performance versus the relatively small performance change of your 7700K over your 3930K.  Overall, you should experience a very big total improvement.  Looking at single-core PassMark your SB-E over your new KL-K you could see close to a 20% improvement, and that starts getting meaningful for sure.  Add the 1080Ti and you will be enjoying quite an upgrade.  I don't see this level of upside in the 6700K extrapolated out to Skylake-X and that really has to do w/ how well SB-E overclocked compared to anything more recent (base clock to highest overclock, i.e.), so indeed you may have the best possible option for the next several years.  Let's hope P3D can exploit the massive bump in VAS.

So you still have to de-lid these chips still?  Remarkable.   I was pleased w/ SB-E to lose the need to dedicate transistors to GPU function as well as avoiding the need to de-lid.  Pretty amazing 5y later to be looking at only a 20% bump, however as I say it's meaningful when you can have decent performance w/o the bump. 

I will certainly run some tests when I have it up and running. Actually I could test out the improvements in P3D v3 as I may not have the 1080Ti for a little while yet and may resort to my 780Ti. That said I wasn't planning on putting v3 in my system and rather wait for v4 to come out.

I agree that there hasn't been much improvement since SB-E which has been a good thing for us. The incremental CPU developments have meant that our need to upgrade has been less. That goes hand in hand with optimisation on P3D rather than a revolutionary design change. However, if it does go 64 bit, makes even better use of GPU's (and maybe cores) then there should be a significant improvement in performance if we consider complex add on scenery and aircraft in locations now which are VAS restricted.

My plan is that this upgrade sees me through a couple of years as the CPU battle plays out and P3D optimisations are made more clear. The next upgrade then can be made in a more informed way - although I think the choice I have made is a very good one and I won't be sitting here disappointed. De-lidding on Kaby Lake is definitely worth it and I have had the company do it for me - they are also sending me a 5.1GHz binned CPU that they've tested. I'm not sure I'll run it at that speed 24/7 unless I can get decent voltages. Will be fun to test anyway.

I don't know how long you have been at it but back in the day of the very first Intel X486 processors sported clock speed in the double-digit mHz.  So for years you could expect a doubling of performance w/ almost every iteration.  Now it's 25% after 4-5 years.  It was really an expensive time for hobby computing, flight simming, and so forth, because of the rapid pace of improvement.  I remember my first box was pretty high end, was a custom built Windows 3.1 486DX 33mHz, ATI card of some type, a 15" top of the line Sony color CRT, 8 megabytes of ram, a 20 megabyte HDD, a 4x CDrom drive, a Sound Blaster, and the two floppy drives.  This box was around $2400 believe it or not, and those were uninflated dollars.. 

Nowadays we can hope for more multi-threading and code optimizations, and relatively meager improvements in single-core performance.  But the upside is you can get some service life out of whatever you purchase as long as you're as future-proofed as possible.   I'm going to hold pat unless the parts die and at that point I would have to see compelling proof that the best showing 4-core du jour is significantly better than 4-8 core options when it's time before opting out of more cores.  I still see some future-proofing potential in this approach, and if the cost for that hedge against a brighter multicore future is less than a 10% hit in single-core performance i'd likely once again go that route again because of the long service life I expect to get out of my next build, barring an unforeseen quantum jump in computing power, then all bets are off!

 

On 3/4/2017 at 7:59 PM, joemiller said:

 Forgive if you haven't consider this but......

..... The benchmarks all have one-thing-in-common... these new AMD CPUs are better than the Intel's on most areas and Intel in others. Upon overclocking one of these AMD  CPU's, the difference we will see (remember we all have different set-ups and play at different P3D and FSX settings) when compared to Intel's 7700K, will be around 1-2 fps. I do not see significant difference of 10, 12, 15, or even 20 fps between both Intel and AMD's. Even today with a high-end Intel processor going from 4.6Ghz to 5.0Ghz shows very little to no difference in performance increase. (I have built and OC  4 computers in the past 10 years).  Most will try to OC to the highest number possible only to feel better that their CPU got all the way to 5.0. ..But, ask them if they see any gains?  Nope- No gains!  In fact, they will go through all the trouble of delidding their CPU to get to 5.0HGhz only to find out , there is  almost no gains in fps between 4.5 to 5.0Ghz. 

 

On 3/4/2017 at 7:56 PM, joemiller said:

 

I don't intend to spoil your joy, but...

..... The benchmarks all have one-thing-in-common... that these AMD CPUs are better than the Intel's on most areas and Intel in others. I can comfortably say that upon overclocking one of these AMD  CPU's, the difference we will see (remember we all have different set-ups and play at different P3D and FSX settings) when compared to Intel's 7700K, will be around 1-2 fps. I do not see significant difference of 10, 12, 15, or even 20 fps between both Intel and AMD's. Even today with a high-end Intel processor going from 4.6Ghz to 5.0Ghz shows very little to no difference in performance increase. (I have built and OC  4 computers in the past 10 years).  Most will try to OC to the highest number possible only to feel better that their CPU got all the way to 5.0. ..But, ask them if they see any gains?  Nope- No gains!  In fact, they will go through all the trouble of delidding their CPU to get to 5.0HGhz only to find out , there is  almost no gains in fps between 4.5 to 5.0Ghz. 

 

I see you making assertions but not citing any data.  

The irrefutable data that proves overclocking makes a difference in frame rate IN FLIGHT SIM (which is what we are all here to discuss) is here: 

It's not up for debate.  Hundreds of data points collected over the last 5+ years all say the same thing: higher clockspeed = higher FPS in ESP-based flight sims (FSX & P3D).  

On 3/6/2017 at 11:30 PM, JamesHongKong said:

Hey Noel, I only just saw your old question on voltages - I think I was running at 1.35v at max but I dialled it back and right now I've turned off the overclock as I'm not currently using the sim - saving it for my new build.

I got ok performance with my 3930k, there's no doubt. But my use case is perhaps different - I have an older 780Ti with it and performance has been fine although I do have to be conservative with the sliders. I do use X Plane sometimes and my system struggles with that for sure. I happen to be moving countries in the next couple of months and I know it'll be prohibitively expensive once I move to do any upgrade hence the urgency - in an ideal world I may have waited for the next Intel or next Ryzen or optimisations to Ryzen but I had to make a choice based on last week and the best set up for last week. As things stand, that is a 7700k overclocked to 5.1GHz coupled with a 1080Ti (when it's released). 

Ideally I would have waited for v4 which may be just around the corner but it is still going to be based on the ESP engine. Therefore any improvements with core management and usage are not going to be massive. I think a really well over clocked 4 core 7700k will serve me well even if they can make use of more cores. Or certainly until my next upgrade say in 3 years. Being able to hopefully not worry about VAS, dial up sliders and having some good add on scenery/aircraft that all perform smoothly is what I'm looking to achieve. 

I should also point out that my plans have changed somewhat on advice from Rob - I'm going to offload a number of the add-ons onto a spare networked computer I have hopefully allowing the 7700k to focus purely on running the sim. Therefore it is the right choice based on my urgent circumstances and today's best equipment for flight simming. 

 

Max, there was a question I was going to perhaps ask on other forums that you may be able to answer. I plan to overclock to 5-5.1GHz - I will have an Asus Maximus Code motherboard and understand there is a very simple 1 button overclock to get it straight to 5GHz. On my current system my clock speed would vary. So when I had it at 4.2GHz overclock it would ramp up and down to that. I'm not sure if that was the correct set up as I've been watching a few videos where people use it overclocked 24/7. Would it be ok to leave it at 5GHz without a variable clock speed? I assume if there is little load on the CPU it won't matter? Just something I want to check.

My CPU is on it's way and I already have my case and PSU. If anyone can recommend a way for me to be notified of the EVGA GTX 1080 Ti FTW release, let me know - that is the GPU I plan to go for. Had some great success with EVGA cards. I cannot find a way and want to be able to jump straight on to buy it when it's released.

 

I dropped off the forums for a few days there, sorry I missed your question.  

I choose to disable power saving features on my CPUs and run them at full speed all the time because those power saving features can actually contribute to micro-stuttering when they don't work correctly.  That being said, I change out hardware often.  Very often.  I don't keep anything around long enough to kill it so if you intend to keep your system for 5 years or so you should probably leave the power saving features enabled.  Also keep in mind I delid and water cool so my temps are significantly lower than what most people see with their off-the-shelf cooling setups and non-delidded chips.

26 minutes ago, TechguyMaxC said:

I dropped off the forums for a few days there, sorry I missed your question.  

I choose to disable power saving features on my CPUs and run them at full speed all the time because those power saving features can actually contribute to micro-stuttering when they don't work correctly.  That being said, I change out hardware often.  Very often.  I don't keep anything around long enough to kill it so if you intend to keep your system for 5 years or so you should probably leave the power saving features enabled.  Also keep in mind I delid and water cool so my temps are significantly lower than what most people see with their off-the-shelf cooling setups and non-delidded chips.

Thanks Max, no worries on the delay - always appreciate the input.

My CPU has arrived today - I got the guys I bought it off to delid it for me and they sent me a 5.1 GHz tested chip which I can play with - hoping for some nice results. Going to need to read up a bit on over clocking again - ideally want to get as low a voltage as possible on it but can't remember how to do the incremental testing that I did for my 3930k! I did run the 3930k at variable clocks but maxed at 4.3GHz for a long time and after 5 years I feel it's struggling. I think I will use profiles in the bios for when I'm flying to max it for that usage.

I thought about custom water loops but I have opted for the Corsair H115i in the end purely as this will be my first build in a long time - I thought custom water-looping may be more better to try out next time once I know I can get the basics working.

Saw your other post regarding RAM so glad I went with the speed I did (3200Mhz) - I did plump for 32GB but that's because I got a good deal. I don't mind a bit of overkill ;-)

Just waiting on my other components to arrive before getting things set up. I may load up a vanilla P3D v3 to test with but I'm holding out for a full install with v4 to be honest. 

On 3/7/2017 at 10:43 AM, Noel said:

James, when you get it up and running post your analysis eh?  Unfortunately it will be nigh to impossible to assess how much total improvement came from the massive increase in GPU performance versus the relatively small performance change of your 7700K over your 3930K.  Overall, you should experience a very big total improvement.  Looking at single-core PassMark your SB-E over your new KL-K you could see close to a 20% improvement, and that starts getting meaningful for sure.  Add the 1080Ti and you will be enjoying quite an upgrade.  I don't see this level of upside in the 6700K extrapolated out to Skylake-X and that really has to do w/ how well SB-E overclocked compared to anything more recent (base clock to highest overclock, i.e.), so indeed you may have the best possible option for the next several years.  Let's hope P3D can exploit the massive bump in VAS.

So you still have to de-lid these chips still?  Remarkable.   I was pleased w/ SB-E to lose the need to dedicate transistors to GPU function as well as avoiding the need to de-lid.  Pretty amazing 5y later to be looking at only a 20% bump, however as I say it's meaningful when you can have decent performance w/o the bump. 

 

Yes you got to delid to reach 51.Ghz  Does 5.1 makes a difference in P3D ? No! Not one bit. P3D needs a lot more than that to work well. However, our beloved milker (Intel), has only made a small improvement in the past 8 years to make us believe that it's the greatest thing, charging a bunch of money for a small "improvement." And this is what really irritates me... this practice of fooling people in order to get more money with the least benefit. 

 

1 hour ago, TechguyMaxC said:

 

I see you making assertions but not citing any data.  

The irrefutable data that proves overclocking makes a difference in frame rate IN FLIGHT SIM (which is what we are all here to discuss) is here: 

It's not up for debate.  Hundreds of data points collected over the last 5+ years all say the same thing: higher clockspeed = higher FPS in ESP-based flight sims (FSX & P3D).  

Both FSX and P3D would work best at 4.6 when compared to 2.0Ghz. However, tell them you just spent almost $3,000 with the greatest and the latest to reach 5.0Ghz, and they will laugh 24/7 at you.  To really make a difference beyond 4.6Ghz, you need to go up to at least  6.5Ghz and above; however, our beloved milker (Intel) is more interested in making a lot  of money very quickly rather providing us with a significant chip that will truly make a difference.  

So, yeah a clock higher than 2.0 or 1.1 is better. But, to lead folks to believe that @5.1 all your dreams will come true is like telling someone that eating 5.1 cotton candies  will get you full enough than eating 2.0 cotton candies.  (it's all fluff) 

2 hours ago, joemiller said:

Both FSX and P3D would work best at 4.6 when compared to 2.0Ghz. However, tell them you just spent almost $3,000 with the greatest and the latest to reach 5.0Ghz, and they will laugh 24/7 at you.  To really make a difference beyond 4.6Ghz, you need to go up to at least  6.5Ghz and above; however, our beloved milker (Intel) is more interested in making a lot  of money very quickly rather providing us with a significant chip that will truly make a difference.  

So, yeah a clock higher than 2.0 or 1.1 is better. But, to lead folks to believe that @5.1 all your dreams will come true is like telling someone that eating 5.1 cotton candies  will get you full enough than eating 2.0 cotton candies.  (it's all fluff) 

Yeah Joe, I'm well aware of the incremental upgrades to CPU technology since I last built a system. I'm not building one and expecting a huge improvement over what I have - I expect to see a noticeable improvement going from a 3930k specifically (which runs really hot at 4.3Ghz) and a 780Ti to a 7700k at 5.1 Ghz with a 1080Ti - regardless of the incremental updates there have been CPU design improvements (which will benefit me) and big GPU improvements. So at 5.1Ghz I will see improved performance in sim and lower temperatures. I'll have the current latest technology and yes Intel have got my money this time but that is because I don't see there being any real choice today. I would hazard a guess that a 7700k running at 4.3 Ghz will probably beat my 3930k at the same clock purely on improved internal design improvements in the last 5 years. It may not be much but it'll be there. Also, it's not about simply increasing clock numbers from 4.6 to 6.5 Ghz in my case (which is what this thread is about) - it's going from 4.3 Ghz on old architecture to 5.1 Ghz on new architecture - and a new GPU. I'm current 3 (or 4?) CPU generations behind and 3 GPU generations behind - a good case for an upgrade. In fact I'd go so far as to pat myself on the back for the system I currently have - I future proofed myself well!

The next leap forward may now be triggered but it'll be a while before that potentially plays out. So for today there's no point getting annoyed that Intel have been running an effective monopoly and indicating that there's going to be minimal improvement in my upgrade. I'm confident there is going to be a decent bump on what I currently get and as long as I'm not unhappy spending the money to do that, that's really all that counts. :-)

PS I don't believe anyone in this thread or others is saying that anyone's dreams will come true at 5.1Ghz - it just happens to be the best current CPU for simming and people are suggesting it will continue to be. That was the answer I need to the thread title "CPU Choices for P3D Now and in the Future"

@JamesHongKong You did exactly the right thing. A delided 7700k is exactly the right thing for P3D.

And well... I don't see P3D to make better use of more cores. They would have to change the whole engine for that. Won't happen.

And where exactly is Ryzen faster then "all other Intel CPUs"? All benchmarks I saw suggest otherwise, except for 2 or 3 games that make good use of more cores.

 

Oh yeah and I'm upgrading from 16GB DDR3 at 1600Mhz to 32GB DDR4 at 3200Mhz (albeit the 32GB is perhaps unnecessary) - that combined with everything else suggests my upgrade is a decent one! ;-)

8 hours ago, joemiller said:

Both FSX and P3D would work best at 4.6 when compared to 2.0Ghz. However, tell them you just spent almost $3,000 with the greatest and the latest to reach 5.0Ghz, and they will laugh 24/7 at you.  To really make a difference beyond 4.6Ghz, you need to go up to at least  6.5Ghz and above; however, our beloved milker (Intel) is more interested in making a lot  of money very quickly rather providing us with a significant chip that will truly make a difference.  

So, yeah a clock higher than 2.0 or 1.1 is better. But, to lead folks to believe that @5.1 all your dreams will come true is like telling someone that eating 5.1 cotton candies  will get you full enough than eating 2.0 cotton candies.  (it's all fluff) 

Who's they?  Who said anything about spending $3000 to get to 5.2GHz?  I spent precisely zero more than you or anyone else has on their 7700k as I already own literally everything I needed to perform the operation and to cool the CPU.  Talk about strawmen...