CPU GHz question

[nemokin]

May not be appropriate to ask in this forum,
but how significant are the changes to CPS speeds for FSX?

I have an original 3.7 GHz machine and had been running it
with overclock of 4.5.
My FSX under DX10 is running quite well at that setting (thanks to Paul and others' advice).

Knowing that CPU speed plays a significant role in FSX performance,
last night I changed my 4.5 overclock to 4.6.
I have yet tried FSX with my new speed so can't compare
but I was wondering, in general, how significant this 0.1GHz
increase is.

My reason for asking is, if the difference is important
then I am willing to see if I can push the speed more up to 4.7 (feels uncomfortable
pushing this beyond that). But, if not so big of a difference, then I'd rather
not spend the time/effort to pursue.

(I kind'a can imagine the response to be "give it the max speed I can,
no sense holding back regardless of impact.")

[mirage3]

Hello, i have original 4 GHz and the possibility to 4.8. the 0.1, i thing its not significant why your CPU is also very speedy! but it can help increase to low the crashes and other things. hope it help:confused:

[Paul J]

There's an old saying in the UK, Nemokin - probably all languages have it:

"Take care of the pennies, and the pounds will take care of themselves". In the USof A I guess it would be "Take care of the cents, and the dollars will take care of themselves."

I spend months preaching the value of every processor cycle: the DX10 How-To guide, and the other files and folders here (http://www.sim-outhouse.com/sohforums/showthread.php?86548-How-To-s-Files-and-Pics), echo this same principle. You have a 2700K - one of the best ever "overclockable" procs made. This one will run over 5.0 gig. First line, second item, under "Miscellaneous" here (http://www.sim-outhouse.com/sohforums/showthread.php?86548-How-To-s-Files-and-Pics).

Given: FSX is processor-bound. If you want to experience the best performance possible from your rig - become a fanatic: you will do whatever is necessary - cost permitting - to make that proc as fast as it will go - and that's only limited by stability issues, or how good your cooling system is. I cannot move mine to 5.1 GHz and keep it reliable. It stays cool, but it will BSoD at any time - not just with FSX, and I've tried every way I can think of to get it stable: I reckon 5.0 is as good as it gets for me - and so I compromise.

Our future FSX isn't going to go anywhere past where we are now, either, as I don't think there's going to be a further increase in proc speed (any time soon), as the current architectures are slowly moving to the tightly-coupled server-style multiprocessing technology to gain performance, with new apps following that route. FSX is still essentially a single-proc mode application with SP2 moving a few threads to other cores, so our DX10 is probably the last generation of FSX, being eventually replaced by P3D.

All the best,

pj

[nemokin]

Thank you Paul, I will try and see how my boosted 4.6 will perform and gradually move up to 4.7 then 4.8 without changing my vcore setting to see how stable it can be. 5.0 GHz would be a dream setting I suppose.

[Paul J]

Thank you Paul, I will try and see how my boosted 4.6 will perform and gradually move up to 4.7 then 4.8 without changing my vcore setting to see how stable it can be. 5.0 GHz would be a dream setting I suppose.

Almost not possible, Nemokin. It's not only vCore that has to be altered, but you will certainly risk BSoD's if you don't change the vCore - READ the manufacturer's mainboard overclocking manual (I've included the ASUS manual in my pinned post) and when you're happy that it's stable and the temps are stable then you can reduce the vCore.



5.0 GHz would be a dream setting I suppose. No - it's very good, but everything needs to be setup with the understanding that the cfg contains many parameters that will still bring on stuttering and low frames very easily. It's all about expectations and compromise, so that proc speed is key in what "experience" FSX will bring to you.

Watch the temps, be careful, one step at a time! It's an interesting side of FSX, isn't it! :encouragement:

All the Best,

pj

[TheFamilyMan]

I'm with PaulJ: push your clock as high as possible such that you can maintain 100% stability while staying within reasonable voltage/thermal limits.

Something extra to consider:
Clock scales cpu speed, at best, in a linear manner. So a .1 bump in a 4.5Ghz OC yields a 2.2% speed increase over running at 4.5Ghz. Now for a leap: say you average 40 FPS with your FSX setup; this .1 bump may increase your average, at best, to 40.9 fps (40 + 40*.022). As for how meaningful this really is in FSX, all I can say is "beauty is in the eye of the beholder". If that .1Ghz is an easy reach, definitely take it. Realize that every .1 step yields another 2.2% increase over 4.5, thus a .3 bump gives you a 6.6% increase: 40 -> 42.6. I'd gladly push my 4770k to 4.8 but my thermals can't handle it. Though I've reached 4.8 at 1.31v w/o HT, I'm too "chicken" to delid it.

[RockStarofRust]

=(I kind'a can imagine the response to be "give it the max speed I can,
no sense holding back regardless of impact.")

You have a much better processor than I do! Paul j has a much better system than I. Because of my budget restraints (WIFE :redfire: due to tripping over all the discarded monitors...) I compromise!

An Intell i7 990X is a first gen i7 @ 32nm. I am cooling on air, (Zalman CNPS 10X Extreme also old school), my board is very slow (ASUS P6T SE) yet with all my compromises 4.6ghz spring,

summer, fall, today 5/14/2014 (100'(F) 38'(C) and 4.7ghz winter, I think we have reached the pinnacle of FSX. FSX has never ran or looked better than NOW!

Nothing fancy and not a video guy, plus we have to compromise with limitations of Fraps or Shadow Play or your preferable capture software...https://www.youtube.com/watch?v=o9tw0psnwzM .

In Overclocking there is no such thing as...well, do not want to offend...best stability yes. Safeguards yes, (ie. high temperature shutoff via...well you know (google makes us all genius').

The ratio of overclock vs. the ratio of output for FSX depends on our ratio of wallet vs. common sense and is finite!!! (For the majority of us anyway...yeah I'm talkin' to you John T.)

Been flying digitally and physically since the late 70's, been building computers for a hobby since 1982. Although low posts here and quite a lurker, please! An American idiom, like in the movie"Donnie Brasco", "Forget about it"!

OR.

"I can't get no...satisfaction":banghead:

[Paul J]

:redfire: hahahaha! "Though I've reached 4.8 at 1.31v w/o HT, I'm too "chicken" to delid it."


Me too! No way, Jose'! :very_drunk:

I'm sticking with the 2600K until I find a cheap 2700K to replace it. Currently- 5.0 gig at 1.432 (2600K max heat with FSX -> 55C)

[noddy]

I'm also running a 2600K at 4.5Ghz and I wish mine would run at 55C, it runs to 75C with FSX.

My machine is a Dell Alienware R3 liquid cooled, 680GTX, 8GB of Ram.

[Paul J]

I'm also running a 2600K at 4.5Ghz and I wish mine would run at 55C, it runs to 75C with FSX.

My machine is a Dell Alienware R3 liquid cooled, 680GTX, 8GB of Ram.


Home-built, Noddy, using very good bits and pieces (https://www.dropbox.com/s/x6tx8cckeqeiyay/Cooling_system.jpg): I've been (using in a few other builds) and recommending the Corsair H100i on the forums, but, to be honest.... I don't think it's very capable. (sigh) Any of the latest procs will overpower it at 4.5 - 4.6 gig, making it pretty useless for getting some real performance out of FSX.

By the way - I reckon you're producing some v e r y nice pics..... :encouragement:

All the Best,

pj

[TheFamilyMan]

Calling the Corsair H100i "not very capable" is a bit unfair. It's Haswell that's the killer; no off-the-self cooler, air or closed loop water, can handle cooling it at OCs around or beyond 4.5Ghz, unless you got very lucky in the silicon lottery (which BTW I did) or do the TIM mod, i.e. delid. At least a Haswell running at 4.5Ghz is as IPC capable as SB at 5.0Ghz, plus it supports faster RAM, PCIe 3.0 and chipset native SATA III and USB 3.0.

[Paul J]

Calling the Corsair H100i "not very capable" is a bit unfair. It's Haswell that's the killer; no off-the-self cooler, air or closed loop water, can handle cooling it at OCs around or beyond 4.5Ghz, unless you got very lucky in the silicon lottery (which BTW I did) or do the TIM mod, i.e. delid. At least a Haswell running at 4.5Ghz is as IPC capable as SB at 5.0Ghz, plus it supports faster RAM, PCIe 3.0 and chipset native SATA III and USB 3.0.

Yeah - I guess so, Rod - I don't have a Haswell on real cooling, so it's hard to judge. It's a pity that Intel did this, as, on one hand they produce such a fine piece of high-tech hardware, then blow it's capabilities out of the window with the crazy-paste.

Is the Haswell as capable, at 4.5 gig - as the SB is at 5.0 gig - when using FSX? That would be my question, and I don't know - I've read posts where that's been written, but I've seen no empirical evidence to support any truth in the matter. Interesting, and maybe some rich fellow will do the comparison some day. If I could be confident of the outcome - I would de-lid one to try, but I can already hear my wonderful wife's words... :censored: - and that baleful glare... (which she inherited from her mother. :hatchet: ..) I think I would be in for a World of Pain, and that same chisel - hammer or OLFA-X-Acto knife might be used against me..

All the Best!

pj

[lownslo]

Yep, the Haswell (and before that, IB) TIM config is a joke. If I was thinking about upgrading now (and I ain't... my wife told me so! :banghead: ) I'd wait for the Devils Canyon i7. 5.0 GHz on air from some early testers who like the TIM config. Only Intel knows how many IB and Haswells were sold before they finally brought the consumer a proper TIM solution.

Greg

[Capt. Speirs]

:redfire: hahahaha! "Though I've reached 4.8 at 1.31v w/o HT, I'm too "chicken" to delid it."


Me too! No way, Jose'! :very_drunk:

I'm sticking with the 2600K until I find a cheap 2700K to replace it. Currently- 5.0 gig at 1.432 (2600K max heat with FSX -> 55C)


Isn't there a point when a 2700K at 5.0 GHz runs the same as say a 3770K at 4.5 GHz? Somewhere I read a comparison of overclocked CPUs and it was quite remarkable how well the 3770K works, even at a slower speed compared to older CPUs clocked faster. I mean, think about it, if cheaper CPUs work better why buy newer ones?

[Paul J]

Isn't there a point when a 2700K at 5.0 GHz runs the same as say a 3770K at 4.5 GHz? Somewhere I read a comparison of overclocked CPUs and it was quite remarkable how well the 3770K works, even at a slower speed compared to older CPUs clocked faster. I mean, think about it, if cheaper CPUs work better why buy newer ones?

True, Ric - but that's using apps that respond to SMP processors. A large database's requirements may require several multi-core processors producing astounding throughput: Another configuration may be as a web server required to handle several thousand consecutive connections: A spreadsheet app may need a single-core, and, as we know - FSX was initially programmed as a single-core app until SP2 was released. That would have been a huge undertaking, and similarly LM is rebuilding FSC/ESP to use DX11, with proper ground shadowing and cloud shadows - neither of which work in FSX.
We are talking about the relative performance in terms of which proc is best suited to run FSX, and when someone quotes performance figures which may well exceed those of the 2600/2700 K-series i7's, we need to see how those figures are arrived at.

Now - if we were to run FSX using the DC proc that Greg mentioned - at 5.0 gig - I'll bet $50 that it will blow the socks off my SB at the same speed, because it'll be built to handle a hundred other apps, videos and connectivity all running at the same time as FSX, while FSX will be locked solid at 30, over Manhootin, in the NGX, 4096, 8/8ths cloud, 500 miles LOD, and so on, with four monitors running Nvidia's G-Sync! Well - ok - maybe just speculatin' a bit... Things like bus and ram speeds, higher firmware versions, proc cache sizes, etc., all have an impact, and soon enough I/we will be looking to replace what we've got.

Great topic.

All the Best,

pj

[Lucas]

Hello friends,

after reading a few posts on other forums, I'm not sure now -> should I turn the EIST and C1E options OFF while I have OC i5 2500K to 4.3GHz and Auto Vcore? (I have FSX's performance in my mind of course).

I know that these options "saves" energy by decreasing CPU clock, but do they can cause stutters?

Lucas

[Paul J]

Hello friends,

should I turn the EIST and C1E options OFF while I have OC i5 2500K to 4.3GHz and Auto Vcore? (I have FSX's performance in my mind of course).

I know that these options "saves" energy by decreasing CPU clock, but do they can cause stutters?

Lucas

Hoi Lucas:
This quote from the ASUS Overclocking (https://www.dropbox.com/s/f6d1ftthunf2n1g/Official%20ASUS%20P8P67%20Series%20Overclocking%20 Guide%20and%20Information.pdf) manual states:-


"It is NOT advised to make adjustments to Cstates as this can considerably affect hard drive throughput performance ( especially SATA6G SSD or Sandforce 2 based SSD ). It is recommended that all CPU power configuration states be left on their default parameters. Overclocking tests have shown internally no increase in multiplier scaling when adjusting these values.
* under special cases with high multi capable CPUs and synthetic high load applications ( Linx, Prime, Occt ) it may required C states to be disabled. This has generally only been confirmed for some 51-54 multi capable CPU’s. K series overclocking benefits from adjustments to Digi+ VRM options.
Sooo.... EIST and the CPU, C1E, C3, C6, etc., (the "C" states) should always be ON with the K series procs. Ramping up and down doesn't happen fast enough to create stutters or micro-stutters. Small hesitations, maybe, but there are a thousand other things that may cause that, too - power-saving scheme (set this to "High Performance") being one of them.

Auto vCore? Never let the pc decide on the vCore - this is a sure way to burn the processor. Always set this voltage manually. Again, quoting that ASUS Overclocking Guide:-


Regarding Voltage Scaling – Internal binning of both D1 and D2 parts we discovered consistent voltage scaling patterns. 1. For K series parts, the stock voltage supplied will allow for consistent overclocking generally up to a multiplier of 43x. There is potential for the multi to be raised to 44x depending on the load induced. This default voltage range be approx 1.240 to 1.260 under load. 2. Increased range between 44 to 47x multipliers will generally require a voltage range between 1.30 to 1.375V with an LLC recommended setting of high to ultra high. 3. Increasing the range between 48 to 50x multiplier will generally require a voltage range between 1.40 to 1.500 with a LLC recommended setting of ultra high. 4. Increased range between 50 to 52 (52 generally considered peak max multiplier except for rare 54x parts) will generally require a CPU voltage range between 1.515 to 1.535V with LLC at Ultra High and potential fine adjustments to the CPU skew range. Overall a key item to note is the best voltage to oc scaling range potential for the turbo multiplier is 1.400 to 1.425 vcore. Using this voltage range with an LLC recommendation of ultra high will generally provide the best scaling potential with proper load temperatures*.

If you check the links in the pinned post at the top of the DX10 forum - there is an "Overclocking the i7-9xx and the i7-2xxx series procs" with three examples of settings which can apply to your proc.

All the Best,

pj

[Lucas]

Thank you Paul.


there is an "Overclocking the i7-9xx and the i7-2xxx series procs" with three examples of settings which can apply to your proc

Do you mean 4.6 (4.8; 4.9) settings.pdf documents?

Regards,
Lucas

[Paul J]

Thank you Paul.



Do you mean 4.6 (4.8; 4.9) settings.pdf documents?

Regards,
Lucas

Yes - these were made as I overclocked the 2600K, by following the ASUS guide, plus some other input from the overclocking forums.
They will serve as good info that is quite usable by most of the i7's.

[Lucas]

Ok, thank you, but do you think that they will be good for my i5 2500K?

[Paul J]

Ok, thank you, but do you think that they will be good for my i5 2500K?

Yes. There are some components missing from the i5, but the Guides are still usable. All you have to do is print them, and go into the BIOS - you will then see all of this. If anything looks strange, or you don't understand it - just Google for the overclocking forums, and either find the answer or ask someone.

pj

[Lucas]

Thaks again for help, Paul. My target is 4.6 - maybe I can do it ;)

[lownslo]

More info about the Devils Canyon CPUs. The i7 looks to be the gem of the lot... clocked at 4.1 GHz and hopefully a better TIM strategy than IB and Haswell. Hope it overclocks well.

http://www.techpowerup.com/201116/multiplier-unlocked-pentium-g3258-core-i7-4790k-and-core-i5-4690k-listed.html

Greg

[nemokin]

Okay folks.
Since I started this thread, I have tried to overclock my maching from orginal 3.5 to 4.7
I have been flying my FSX in DX10 at 4.5 for about a month now at a stable environment.

Although I have yet flown my FSX under 4.7 (since I just tried this boost this morning),
I have now achieved the attached results.

Intel Burn Test with stress level at high and running it 15 runs
with the success message appearing at the end.

I have set my vcore at 1.470V to achieve this result. When lowered to 1.460V,
I would get BSOD during my Intel Burn Test (high stress level at about run #6).

My worry is that my CPU temp goes up to about 95 degrees (although not during the whole test)
when using my Intel Burn Test.

Of note, my current CPU temp (just using the web browser to type this) is around 40 degrees
but my Core speed is spooled down to 1.6GHz.

Are these indications that my PC is stable now at 4.7GHz?

(I have liquid cooling by the way)
8626

[nemokin]

Improved results to my previous setup.
Instead of having my vcore "fixed", I've now switched to "offset" and now
my vcore temp is lower around 1.38V instead of 1.46V.
With Intel Burn test, CPU temp goes up to 91 degrees but never 96.

I've test flown my NGX for a 2 hour flight and no problems.
Monitoring CPU temp, it stays between mid 60's to mid 70's degrees
most of the time, with a max at 82 degrees.

Will now fly with my new 4.7GHz setting for a while.
Different weather at different time/day setting to compare pre and post
4.7, nicely smoother at higher clock speed at least today's flight.




Okay folks.
Since I started this thread, I have tried to overclock my maching from orginal 3.5 to 4.7
I have been flying my FSX in DX10 at 4.5 for about a month now at a stable environment.

Although I have yet flown my FSX under 4.7 (since I just tried this boost this morning),
I have now achieved the attached results.

Intel Burn Test with stress level at high and running it 15 runs
with the success message appearing at the end.

I have set my vcore at 1.470V to achieve this result. When lowered to 1.460V,
I would get BSOD during my Intel Burn Test (high stress level at about run #6).

My worry is that my CPU temp goes up to about 95 degrees (although not during the whole test)
when using my Intel Burn Test.

Of note, my current CPU temp (just using the web browser to type this) is around 40 degrees
but my Core speed is spooled down to 1.6GHz.

Are these indications that my PC is stable now at 4.7GHz?

(I have liquid cooling by the way)