You de-lidding detractors seem to forget the minimum FPS factor, i.e. the worst case scenario. When your FPS get down to 10 at KJFK at night in the NGX and mine are at 13, which of us do you think is going to have the better simming experience? That's tangible, not subjective.

 

No, that's not how it works. At low frame rates, you still get the same FPS boost. so from 4.5 to 5GHz you still get about a 10% perf boost, which translates into a meager extra 1 FPS at 10 FPS. How is that going to make that much of a difference?

You guys can keep trying to overstate overclocking, IPC and memory performance in general and IB in particular. That's not going to change the fact that things are how they are

 

Yes, that is a good point, 10 frames per second in Techguys "KJFK at night" scenario, to 11 frames per second is pretty useless in terms of performance boost.

 

For Techguy, at 10 FPS to achieve 3 extra fames, he would require a massive 30% increase in frame rate. In other words... 5.85 Ghz. :biggrin:

No, that's not how it works.

 

Ok Dazz... I don't know what I'm talking about... I've never experienced the low end and how a few fps can make a difference in that range.

 

I am not talking at all about OC per se. So I am overstating nothing... just describing what I have experienced. And I am not alone on this... many here have observed this phenom with FSX... for one the Techguy just said the same same.

 

And the plain and simple fact... a proper delid will dramatically drop temps and allow a higher oc... what performance gains specifically I cannot address as I really don't care to do a lot of testing ("I'd rather be flying"). However, it is undisputed that higher freqs = better FSX performance. So I sign-on (in this case) to the Pirates of the Caribbean mantra of "take what you can... give nothing back."

 

In addition, I have also stated several times now, in two threads

 

Why you guys duping that btw? your conversations? One not enough?

 

Also btw... can't see I am in any "boat"... I just wanted to try the de-lid once (and lapping for that matter)... the one I am using now is "pristine" (no delid or lap).

But I don't see how a 11% clock boost can yield a 30% more minimum FPS. Sounds to me like the magical mystery boost or something.

 

Thank you!

 

For Techguy, at 10 FPS to achieve 3 extra fames, he would require a massive 30% increase in frame rate. In other words... 5.85 Ghz.

 

I believe it's worse than that, but my opinion is more based on observation of my own experience w/o rigid controlled testing, AND only testing Core 2 Quad processors in my testbed which employs decent ram and timings for its day, and a GTX280, also a killer card for its day. 30% increase in clock speed I maintain will not yield a 30% increase in frame rate. My sense is for every 1% increase in clock speed you may see .7% inc in frame rate. This of course depends on the other bottlenecks, and indeed may come from other bottlenecks. I think I can unequivocally say that you will not see greater than 1:1 scaling.

I thought about delidding my new i7 3770k which is arriving tomorrow with all the other ordered parts to complete a dedicated FS only machine. My i5 2550k rig is going under my 40inch TV in computer room as gaming/media centre.

 

I may have a go at the delidding process in 6 months or so. I'll make sure the chip works ok first and tempt fate when the next gen of i7's arrives this year. Hopefully the next gen might be worth upgrading to and if so and I make a botch job of delidding at least I can buy the new CPU to make up for it.

 

So this is the final word on wild overclocking and de-lidding, as far as I'm concerned:

 

Only do it when you need a good excuse to spend money on a new CPU. If you succeed? Hey! You've saved hundreds of dollars!

 

<br />For Techguy, at 10 FPS to achieve 3 extra fames, he would require a massive 30% increase in frame rate. In other words... 5.85 Ghz. :biggrin: <br />

 

Hi

worst case senario, that you have in FSMark11 "min fps" or iam wrong.

min max avg

2700K @4.6 mem 1600mhz 26.3 61.7 44.4

2700k @5.2 mem 2133mhz 33.3 69.0 51.1

27% 11% 15%

 

Overclock include mems approx 17%

 

its a 27% increase on worst case on a 17% OC

 

Hasse

Hi

worst case senario, that you have in FSMark11 "min fps" or iam wrong.

min max avg

2700K @4.6 mem 1600mhz 26.3 61.7 44.4

2700k @5.2 mem 2133mhz 33.3 69.0 51.1

27% 11% 15%

 

Overclock include mems approx 17%

 

its a 27% increase on worst case on a 17% OC

 

Hasse

 

Hasse. Minimums and maximums are instant readings in fraps and obviously minimums are more subject to statistic error since those are lower values. Are those results consistent at the same memory speed?

 

Hasse. Minimums and maximums are instant readings in fraps and obviously minimums are more subject to statistic error since those are lower values. Are those results consistent at the same memory speed?

 

Actually, it depends on the measurement chosen.

Actually, it depends on the measurement chosen.

 

What do you mean? minimum readings can easily be skewed by a single sort of imperceptible stutter in fraps. If we're talking about FSMark11, then the measurement criteria must by FRAP's right?

I'd like to install all the components on the mobo outside the case and run the pc for test purposes before any modding and delidding. But how to launch it without Power button (that is on the case)?

ASUS Maximus V Extreme here.

 

Thanks,

Dirk.

 

PS: This one is easy - short the pins with a screwdriver.

Hi

worst case senario, that you have in FSMark11 "min fps" or iam wrong.

min max avg

2700K @4.6 mem 1600mhz 26.3 61.7 44.4

2700k @5.2 mem 2133mhz 33.3 69.0 51.1

27% 11% 15%

 

Overclock include mems approx 17%

 

its a 27% increase on worst case on a 17% OC

 

Hasse

 

 

 

It's essential I would say, to get to the bottom of exactly the sort of performance gains we can achieve by iincreasing our overclocks from 4.5 to 5 Ghz. For Ivy Bridge of course.

 

It's no good guessing. We need a reasonable number of individuals, not just Hasse, who's systems are capable of reaching 5Ghz, to do the tests for us and settle this once and for all.

 

And not just FSMark 11, we need this confirmed or refuted using several testing methods.

We need a reasonable number of individuals, not just Hasse, who's systems are capable of reaching 5Ghz

 

Sorry too scared. :Nerd:

 

I'll take Hasse's word for it... or the Bimmer's :im Not Worthy:

 

LOL or I'll do what you do Martin... and say I am happy right where I'm at. :LMAO:

 

 

PS: This one is easy - short the pins with a screwdriver.

 

Should have been even easier Dirk... (and maybe a tad safer)... does it not have both power and reset buttons on the board?

Well... I've just done a quick test.

 

From 4Ghz to 4.5Ghz

 

Scenario was 737 NGX, at EGLL 27L. Weather was flightsims Gray and Rainy.

 

The results.

 

4GHz: VC = 23.5 Outside = 30

 

4.5 Ghz: VC = 26.5 Outside = 33.5

 

4Ghz to 4.5Ghz = 12% increase in clock speed. 13% increase in frame rate.

 

Allowing for errors, clock speed increase correlates to frame rate increases.

 

Someone else will need to test from 4.5 to 5, as I can't reach those frequencies, but it looks definitive to me. I highly doubt, that 4.5 to 5 would magically jump to a 27% increase in performance.

 

Looks like if you increase clock speed by a given percentage, you get the same percentage increase in frame rate. Just as I had assumed.

 

If this is correct from 4.5 to 5, then I see no reason to reconsider my choice not to delid.

4Ghz to 4.5Ghz = 12% increase in clock speed. 13% increase in frame rate.

 

Then the bottleneck in your machine is the CPU--congrats, that makes it simpler. The scenarios are identical, i.e., it's the same 'loaded' scenario on a fresh reboot?

Scenario was 737 NGX, at EGLL 27L. Weather was flightsims Gray and Rainy.

 

Martin, help me set up a test that will help me appreciate what I can expect if I do a Haswell upgrade, which I am seriously thinking of not doing as I can't tell if this really will be worth the time, cost, headache, etc.

 

 

Here are my critical configuration settings:

▪ All sliders hard right except the following

▪ Road traffic at 8%

▪ Air Traffic and Airport Ground Traffic at 30%

▪ Autogen at DENSE

▪ Cloud distance at 100M

▪ Water at one notch left of hard right

▪ GPU settings for my 1920x1200 display

▪ AF 16x

▪ AA 8xS, 8x Supersampling

I keep FSX at UNLIMITED, and use nV Inspector's frame lock at 30, and use fiberframe manually set a .33 (this may not be necessary but I use it in case UNLIMITED overrides it. I have best overall perf at UNLIMITED and don't suffer from blurries unless I'm in over my head.

 

Can you set your machine up like this, leave everything the same, including UNLIMITED so I can compare apples to apples?

 

Next, are you using ORBX FTX England? If so, I have it too and could put the NGX on 27 at EGLL. Change weather to Clear, so we can truly standardize the comparison. Tell me what you see thru the VC. Leave your machine at 4.5Ghz if that what it can do.

I'll have a look tomorrow for you. Or could be the day after. If I don't make progress with my novel, I'll be very depressed. :smile: