The Value of Overclocking an Integrated System

[Spirit Flyer_old]

The Value of Over-clocking

Recently while helping a friend OC his i975, an online multiplayer tried to tell me and the others not to fiddle, over-clock or tweak their computers as this does nothing to really enhance performance and is dangerous to the equipment at best. He was somewhat rude, adamant and got quite worked up about it. All we could do was to be polite, not argue back, and wait for him to stop sputtering and go away so we could get on with it.

You have probably heard and read statements here, and elsewhere that have expressed similar comments whether it comes to adjusting the bios for the motherboard, over-clocking the CPU, GPU, RAM or anything else. The same goes for using prudent FSX adjustments. Everyone is equally entitled to have, and to express their opinions, but obviously some opinions work better than others. Objective benchmarks are the only way to tell the real difference between them.

PassMark Software (http://www.cpubenchmark.net) claims to have the results of over 200,000 CPU and GPU benchmark tests that are updated on a daily basis. They rate not only one type of CPU and GPU against other types, but allows an individual tester to measure the performance of his machine against others that are of the same make and model. That enables us to see just how our own computer stacks up to others.

No testing and measuring system is infallible, as expressed in their warning, but because of the huge numbers of tests run, they may be the most accurate: "Rarely is a graph completely accurate in what it is representing. There are many factors that can skew the results and make a graph misleading. As such it is necessary to have some background understanding of the data being presented."

Each test is done, whether it be on the CPU or the GPU as they are actually clocked when the test is run, so the average includes the results of both default and over-clocked systems, blending them into one. Since most PassMark adherents tend to be somewhat performance orientated, the average benchmark is probably a fair bit higher than what just stock clock settings would be on their own.

In accordance with these standard operating procedures, I tested my machine against the others to see whether or not over-clocking my i7 930 CPU to 4.3ghz (from 2.8ghz) and 480 GTX GPU to 850mhz (from 700mhz) would make a significant performance difference when measured against the same equipment (with average clocks) in other machines. Since no two machines are exactly alike, the average results of a large number of machines is the most important comparison point.

CPU BENCHMARKS

The average results for high end CPUs are as demonstrated below as of August 30, 2010, the date of the test. As you can see the machines with i7 980X as the CPU tested the highest at an average of 10,336 points. These include both stock and over-clocked chips. The cost of the i7 980X (at that date) was $999.99. My machine's i7 930 was rated way down to number 25, with a performance rating of nearly half of the leader at 5,824 points at a cost of $289.99.

When over-clocked from 2.8 to 4.3 the i7 930 takes off like a rocket ship performing at 9,017.9 points, placing it partway between the Xeon W3680 (at $1,059.99) and the Xeon X5680 (at $1,708.49), moving it from 15th to 5th in performance, which is just 1318.1 points back from being #1. All this for just $289.99 and a few minutes of self education and bios changes (plus cooling solutions).

GPU BENCHMARKS

Average results for high end GPU 3D Graphics tests are graphed below, as of the date of this test. Clearly the Nvidia GeForce 400 series cards have become performance stars, taking a sizable lead over their AMD ATI competitors. The GeForce 480 GTX leads the pack with an average G3D Mark of 3,496 points, some 749.42 points and 28.6% over the Radeon 5870 - a huge performance advantage. The rest of the graphic cards quickly fall away from there. Even the once dominant GeForce 285 GTX scores a mere 2,031 points, some 1465 points and 41.9% below it's successor.

When over clocked from 700mhz to 850mhz core clock, and 1950mhz memory, the 480 GTX jumps ahead of the average 480 GTX tested, to a remarkable 4,621.3 points. This represents an astonishing 1,125.3 points, or 32.19% improvement. Obviously this 480 GTX is getting shoved from behind by the super fast ram, drive and CPU. This places the over-clocked 480 GTX 2,751.3 points, or nearly 2.5 times better than my previous graphics card, a 275 GTX - a massive jump!

HARD DRIVE BENCHMARKS

Only recently have I bought and installed a solid state hard drive. I had no idea what a huge difference it makes in smoothness, speed and the general satisfaction it can bring. Truly, it is like having a new computer. Everything is BAM, SLAM, SCRAM - right now! Even though I do not have the very fastest drive available, my 120GB G.Skill Pheonix PRO has completed the traffic circle on my machine, removing any roadblocks that limit performance. It is an essential part of the performance team that helps make the high numbers of the other components possible. The solid state hard drive world is changing quickly, in both performance and cost. Here is a snapshot of the best of the best as tested:

As you can see, the Crucial C300 leads the way with 2,039 points, which is a healthy lead over it's nearest competitor at 1,654 points, a 23.27% advantage. The sampled top 28 SSD's all perform over 1,000 points, providing lightning data feed to the rest of the systems. There is no turning back at this point as SSDs are taking over the marketplace. Each month brings better drives at lower prices. At test date my new SSD was too new to have been widely benchmarked as a distinct catagory, but the 120GB G.Skill Pheonix Pro installed on my machine performed at a respectable 1141.7 points, placing it between the Patriot Torqx 128GB SSD and the Corsair CSSD-V128GB2.

Quick Analysis

In the same way as no man is an island, no piece of computer equipment can function by itself. Computer performance solutions require equal partnership with the other corresponding components in order to function at maximum capacities. It is when a fast CPU works with fast RAM and a fast GPU with a fast hard drive on a well tuned motherboard, adequately powered and cooled when controlled by a well adjusted operating system that good things happen. Good things never happen by accident, but by design. There are no equipment shortcuts that work. Performance always drops to the lowest common denominator every time on every machine.

I hope this helps to encourage those who want to maximise their Flight Simulation experience without breaking the bank. Computers are simply tools that need sharpening and wise application. So this is not about comparing machines, how one is better than another, but how it is possible to iron out hardware incompatibilities so you can hit far above your weight. Mine is still a work in progress. I know that many here have much better machines than mine, and some a bit less, but in all cases I trust that you will be able to benefit somehow from this post towards reaching that end.

Kind regards,

[Wolter van der Spoel]

Top Notch Stephen !

I made it a sticky

[Spirit Flyer_old]

Thank you Wolter. That is very kind of you. 

Kind regards,

[Paul Baumanis]

Superb.

Very informative, clear and concise.

Well done mate, and thanks.

[Argharg]

Can SSD drive be OC? I dont see why not. Etien

[Spirit Flyer_old]

Can SSD drive be OC? I dont see why not. Etien

Hi Etienne,

I tried googling "how to overclock an SSD" and went looking in the usual hiding places for those who would know if and how it is done, without finding a single word being written about it. Interesting question though!

Kind regards,

[TenBlade]

Wouldn´t surprize me if there were water cooled SSD´s soon

Bjorn

[Iain Emms]

Another really great and very imformatted post my Friend, Very good reading indeed and Thank you for posting these findings.

cheers

Iain

[davidv]

Very handy! Thanks!

[gheeD]

Great post Stephen and definitely deserves to be sticky!

I have personal benchmark results of my i7 930 at default clocks vs 4.2GHz, and the overclock was 10 fps better in FSX. I didn't test on other games, but i believe the raise was more than 10 fps in every game. (i was playing BC2 (bad company 2) and testing one issue, so i defaulted all clocks and the game was not fluent at all, like it was with OC with same settings in BC2) So overclocking is definitely worth it!

to overclock, there has to be something to overclock (as it firstly was really overclocking the clock speed of CPU = raising MHz) but nowadays it also consists everything that has (clock speed) MHz that can be tuned up from the factory default.

[jjjallen]

Stephen excellent writeup.  I agree that OCing gives added value to your hardware expenditures.  The problem may be that most users only dabble without the right tools or information.

My question is how do you find the sweet spot where all of the planets are aligned and your CPU / GPU are working in harmony?  I understand that the sweet spot will move depending on FSX workload and that the BP tweaks may be the best way to keep graphics requests flowing smoothly.

Finally, I can get my system to run with a much higher OC using a stress tool like Kombuster but have to downclock quite a bit to get stable in FSX.  To be honest, I am more concerned about good IQ and smoothness than a high FPS counter.

TIA for any tips.

jja

[alainneedle1]

What Stephen explained here is sooooo true...

jja...let me try to answer your question about getting the sweet spot.....

Let me give you an example......about 25 years ago I was drag racing (1/4 mile) in Canada and to me building a new PC is a lot like building a dragster ....so let assume your CPU (930) is your car motor.... your rams (1600MHz) is the transmission..... the mobo(EVGA E758-A1 3-Way SLI) is the body of the car..... the GPU (GTX 470) is the axel and the HDD is the fuel.....ok so you have a decent set up and you'r benchmarking some good numbers, let say 3.8GHz (1/4 mile in 11sec).....

Now here is the kicker...you'r asking yourself "can I shave 1/4 of a sec on the quater mile with what I have"....maybe.... so what do you have to do....like a race car you have double check everything... test test and re-test....is the timing on the motor tight enough.... meaning is the voltage well balanced on the CPU....can I raise the MHz on the memory +++ and the only way to know is by testing ...OCCT, Prime 95, (making a run 1/4 mile)....write down the number adjust something else and try again....better number after a run...GOOD...worst number....go back to what you had before and try something else...

OK so after changing a little thing here and there in the bios your car is doing 10.75 sec on the 1/4 mile run....(4.00GHz) and everything is well tuned with all the testing and adjustment you did so you think you have a winner....BUT....you'r looking at the dude beside you and he's doing 9.50 sec....so you take a look at what he as under the hood of his car and you see a i7 980X @ 4.43GHz .....

Now you run to the nearest part supplier you buy and drop the 980x in your mobo and you'r like "HA! now i'll be able to do 9.5 sec on the quater mile" at 4.43GHz.....so you get on the track put the pedal to the metal and BANG.....BSOD .... you never passed 100mph you got oil on the track a bend frame and the transmission does not get in first gear anymore....OUPS...what happend.....well you forgot to take a look at the rest of his car (rig) thinking you were ok just buy changing the motor (CPU) but the other dude was running on 2000MHz Rams (bigger transmission) with a GTX 275 (axel) and a funny car frame (GIGABYTE GA-X58A-UD7) with an SSD (nitro).

So changing only one piece of hardware on your rig may improve your number ONLY if the rest of your hardware can keep up with the new one....look at my sig. this Q9650 is pushed to the max including the rams at 1200MHz timing at 5-4-4-10 so there is nothing else I can do with it ...let's say I change the Q9650 for a QX9650....I'm still screwed cause I'am running at 1200MHz on the rams and that's the max this mobo can take.

I put the new GTX 480 AMP! on it just to see what more I can get out of the rig....guess what.....nothing....with the 480 my whole rig became the bottle neck cause their is nothing else I can improve on it to keep up with the 480.

So jja to answer your question (finally LOL) the only way to find the sweet spot on your rig is by testing, testing and testing again until there is no more option because you have exausted all possibility and you have reach the best you can do with what you have.

[jjjallen]

What Stephen explained here is sooooo true...

jja...let me try to answer your question about getting the sweet spot.....

Let me give you an example......about 25 years ago I was drag racing (1/4 mile) in Canada and to me building a new PC is a lot like building a dragster ....so let assume your CPU (930) is your car motor.... your rams (1600MHz) is the transmission..... the mobo(EVGA E758-A1 3-Way SLI) is the body of the car..... the GPU (GTX 470) is the axel and the HDD is the fuel.....ok so you have a decent set up and you'r benchmarking some good numbers, let say 3.8GHz (1/4 mile in 11sec).....

Now here is the kicker...you'r asking yourself "can I shave 1/4 of a sec on the quater mile with what I have"....maybe.... so what do you have to do....like a race car you have double check everything... test test and re-test....is the timing on the motor tight enough.... meaning is the voltage well balanced on the CPU....can I raise the MHz on the memory +++ and the only way to know is by testing ...OCCT, Prime 95, (making a run 1/4 mile)....write down the number adjust something else and try again....better number after a run...GOOD...worst number....go back to what you had before and try something else...

OK so after changing a little thing here and there in the bios your car is doing 10.75 sec on the 1/4 mile run....(4.00GHz) and everything is well tuned with all the testing and adjustment you did so you think you have a winner....BUT....you'r looking at the dude beside you and he's doing 9.50 sec....so you take a look at what he as under the hood of his car and you see a i7 980X @ 4.43GHz .....

Now you run to the nearest part supplier you buy and drop the 980x in your mobo and you'r like "HA! now i'll be able to do 9.5 sec on the quater mile" at 4.43GHz.....so you get on the track put the pedal to the metal and BANG.....BSOD .... you never passed 100mph you got oil on the track a bend frame and the transmission does not get in first gear anymore....OUPS...what happend.....well you forgot to take a look at the rest of his car (rig) thinking you were ok just buy changing the motor (CPU) but the other dude was running on 2000MHz Rams (bigger transmission) with a GTX 275 (axel) and a funny car frame (GIGABYTE GA-X58A-UD7) with an SSD (nitro).

So changing only one piece of hardware on your rig may improve your number ONLY if the rest of your hardware can keep up with the new one....look at my sig. this Q9650 is pushed to the max including the rams at 1200MHz timing at 5-4-4-10 so there is nothing else I can do with it ...let's say I change the Q9650 for a QX9650....I'm still screwed cause I'am running at 1200MHz on the rams and that's the max this mobo can take.

I put the new GTX 480 AMP! on it just to see what more I can get out of the rig....guess what.....nothing....with the 480 my whole rig became the bottle neck cause their is nothing else I can improve on it to keep up with the 480.

So jja to answer your question (finally LOL) the only way to find the sweet spot on your rig is by testing, testing and testing again until there is no more option because you have exausted all possibility and you have reach the best you can do with what you have.

Alan, nice rant.  Check out this link that SpiritFlyer turned me on to. 

http://www.techreaction.net/2010/09/07/3-step-overclocking-guide-lynnfield/

When you see the chip layout you think about overclocking differently.  Like the inside of your race car motor

jja

[alainneedle1]

That's a good one for Lynnfield Bloomfield  CPU....as you know each CPU is different...take the 980x... it's a Gulftown and it's a little different....that's the one I'll receive this week ...here is a good article on how to overclock them...

http://i4memory.com/f55/intel-core-i7-980x-guide-optimal-cpu-vtt-memory-voltage-voltages-uncore-memory-frequency-ratios-23430/

No matter what you me or other will use a good overclock (meaning getting the most out of the CPU) is a lot of reading about what others were able to do with their CPU and a lot of testing....I should be able to start getting my new rig together in 2 weeks...I still have to buy a SSD for the OS and other small stuff...I expect to spend a full week on testing if everything goes according to plan.....

[jjjallen]

That's a good one for Lynnfield Bloomfield  CPU....as you know each CPU is different...take the 980x... it's a Gulftown and it's a little different....that's the one I'll receive this week ...here is a good article on how to overclock them...

http://i4memory.com/f55/intel-core-i7-980x-guide-optimal-cpu-vtt-memory-voltage-voltages-uncore-memory-frequency-ratios-23430/

No matter what you me or other will use a good overclock (meaning getting the most out of the CPU) is a lot of reading about what others were able to do with their CPU and a lot of testing....I should be able to start getting my new rig together in 2 weeks...I still have to buy a SSD for the OS and other small stuff...I expect to spend a full week on testing if everything goes according to plan.....

Kid in the candy store.  Have fun!

jja

[Spirit Flyer_old]

That's a good one for Lynnfield Bloomfield  CPU....as you know each CPU is different...take the 980x... it's a Gulftown and it's a little different....that's the one I'll receive this week ...here is a good article on how to overclock them...

No matter what you me or other will use a good overclock (meaning getting the most out of the CPU) is a lot of reading about what others were able to do with their CPU and a lot of testing....I should be able to start getting my new rig together in 2 weeks...I still have to buy a SSD for the OS and other small stuff...I expect to spend a full week on testing if everything goes according to plan.....

We all have been watching and waiting for this exciting new build to happen! I can't wait to see the results! I hope it's a smoking screaming screeching blur of thundering muscle car from the traps to the flag, setting all new FSX records!

Kind regards,

[alainneedle1]

That's a good one for Lynnfield Bloomfield  CPU....as you know each CPU is different...take the 980x... it's a Gulftown and it's a little different....that's the one I'll receive this week ...here is a good article on how to overclock them...

No matter what you me or other will use a good overclock (meaning getting the most out of the CPU) is a lot of reading about what others were able to do with their CPU and a lot of testing....I should be able to start getting my new rig together in 2 weeks...I still have to buy a SSD for the OS and other small stuff...I expect to spend a full week on testing if everything goes according to plan.....

We all have been watching and waiting for this exciting new build to happen! I can't wait to see the results! I hope it's a smoking screaming screeching blur of thundering muscle car from the traps to the flag, setting all new FSX records!

Kind regards,

HAha I can't wait either Stephen...I don't know about FSX record breaking but trust me ....there is nothing like the smell of burning tires......

This guy broke the sound barrier on land.....craaaaazy stuff

Thrust SSC - Still The Fastest Land Vehicle

[Spirit Flyer_old]

Update November 19, 2010

It has been some time since I began the work in this thread to try and place a relative value on the gained efficiencies of an integrated well-matched overclocked computer system, particularly as it relates to its use as a platform for advanced versions of FSX.

Since I ran the original tests I have swapped out my primary 640gb WD 64 cache drive and a single G.Skill FM-25S2S-120GBP2 solid state drive for FSX to two of these SSDs configured in Raid 0 containing both the OS and FSX. It seems therefore advisable that new objective tests be conducted to ascertain what benefits could be derived from such an upgrade.

Consequently here are the accurate results of a battery of tests using the widely respected PassMark software provided at the below link:

Download PassMark Performance Test 7 Software

http://www.passmark.com/products/pt.htm

This is the complete results of the tests on this computer attained on this date of November 19, 2010, as follows:

Chart 1

Overall PassMark System Rating as of  19/11/2010

Chart 2

CPU PassMark Rating as of 19/11/2010

Chart 3

2D Graphics Mark Rating as of 19/11/2010

Chart 4

3D Graphics Mark Rating as of  19/11/2010

Chart 5

Memory Mark Rating as of 19/11/2010

Chart 6

Disk Mark Rating as of 19/11/2010

Chart 7

CD Mark Rating as of 19/11/2010

Published PassMark Bench Test Results 19/11/2010

Chart 8

Average CPU Performance Results as of 19/11/2010

http://www.cpubenchmark.net

Comparison of Chart 8 with Chart 2

This computer's 4.32Ghz overclocked i7930 CPU ran the gauntlet at 9023.9 points, just under the Xeon X5670, up from the default value of 5,832, some 3,191 points at +54.7%

Chart 9

Average GPU Performance Results as of 19/11/2010

[img width=540 height=700]http://fsfiles.org/flightsimshots/images/708GPU_Ratings.png

http://www.videocardbenchmark.net

Comparison of Chart 9 with Chart 4

This computer's 480 GTX overclocked to 850mhz passed the test with a 3D Graphics mark of 4,600.9 points which places it nearly 30% and 25% respectively over the average tested 480 GTX and 580 GTX.

Chart 10

Average Hard Drive Performance Results as of 19/11/2010

[img width=521 height=700]http://fsfiles.org/flightsimshots/images/358Hard_Drive_Ratings.png

http://www.harddrivebenchmark.net

Comparison of Chart 10 with Chart 6

This computer's twin G.Skill FM-25S2S-120GBP2 solid state drives configured in Raid 0 produced a disk mark of an extraordinary 3,908.1 points, which is 117% faster than when tested by PassMark as a single drive against others in masse. The value of removing system bottlenecks is even more dramatically demonstrated when the original SSD with only FSX installed, working with the OS on a mechanical drive, only produced a modest 1141.7 points, nearly a quarter of what it is producing when configured in RAID 0 with an equal teammate (see original graph in first post). Obviously these second placed (Chart 10) solid states drives, as good as they seem to be, are even better in RAID 0 configuration, at least in my computer.

This also explains the performance of the CD in Chart 7. The ability of the CPU and the Disk and the CD to transfer data could be described for once as a genuine superhighway, to use an old cliche.

Message to my FSX friends:

Please do not evaluate or use this information to make equipment decisions whatsoever without carefully reading the analysis and conditional findings and explanations found in the original post of this thread. It is not just about having the latest and best piece of equipment, but having a well matched integrated system that works well with itself. Please read more about that here:

http://orbxsystems.com/forums/index.php?topic=26354.0

As always, this information is believed to be true and accurate as of this date, however it is not warranted to be so. Any use of this information to make changes in computer hardware, configuration, or software is entirely at your own risk. I hope this might help you in your quest for excellence in your FSX hobby.

Kind regards,

[jjjallen]

Thanks Stephen!  This is really useful info.

jja

[peterhayes]

Stephen

Great post that covers a lot of specifications that may help FSX to run well.

Does anyone have a measure of the data transfer rate accross the PCIE bus is that where the bottleneck is?

To me the one other measure(s) that we never get is how the FSX code (plus its interactions with 3rd party add-ons) relates to memory allocation and the virtual address space.  (Poor performance, stutters OOM errors even in a 64-bit OS)

To me that could show definitively the optimal settings to run FSX and which 3rd party addons speed up, slow down or overload the process.

IMHO this post from MS http://msdn.microsoft.com/en-us/library/aa366781(v=VS.85).aspx shows what we may need to get a grasp on to fully understand how FSX can perform and possibly what we can do to optimise what hardware we have.

I live in hopes (I should say HEAPS) that we can tame the beast.

Thanks again for the specs.

Regards

PeterH

[alainneedle1]

That's a good one for Lynnfield Bloomfield  CPU....as you know each CPU is different...take the 980x... it's a Gulftown and it's a little different....that's the one I'll receive this week ...here is a good article on how to overclock them...

No matter what you me or other will use a good overclock (meaning getting the most out of the CPU) is a lot of reading about what others were able to do with their CPU and a lot of testing....I should be able to start getting my new rig together in 2 weeks...I still have to buy a SSD for the OS and other small stuff...I expect to spend a full week on testing if everything goes according to plan.....

We all have been watching and waiting for this exciting new build to happen! I can't wait to see the results! I hope it's a smoking screaming screeching blur of thundering muscle car from the traps to the flag, setting all new FSX records!

Kind regards,

Hi! Stephen, that is a very good comparaison, very informative and very interresting.....

This is a good tool to see where can be your weakest hardware, I've never tried it before so I gave it a go...what do you think?

[Spirit Flyer_old]

That's a good one for Lynnfield Bloomfield  CPU....as you know each CPU is different...take the 980x... it's a Gulftown and it's a little different....that's the one I'll receive this week ...here is a good article on how to overclock them...

No matter what you me or other will use a good overclock (meaning getting the most out of the CPU) is a lot of reading about what others were able to do with their CPU and a lot of testing....I should be able to start getting my new rig together in 2 weeks...I still have to buy a SSD for the OS and other small stuff...I expect to spend a full week on testing if everything goes according to plan.....

We all have been watching and waiting for this exciting new build to happen! I can't wait to see the results! I hope it's a smoking screaming screeching blur of thundering muscle car from the traps to the flag, setting all new FSX records!

Kind regards,

Hi! Stephen, that is a very good comparaison, very informative and very interresting.....

This is a good tool to see where can be your weakest hardware, I've never tried it before so I gave it a go...what do you think?

Hi Alian,

The fruit of your quest for the best has yielded a termendous crop. What a spectacular performance the Z-Drive demonstrates for incredible speed! Once again, this is another and most impressive validation of the value of having an integrated system where components are matched and mated for compatible yield, where each part enhances the performance of the others. Here is proof positive that the performance value of the whole can be significantly greater than the sum of the performance value of it's components.

I love it! The faster the better.

Kind regards,

[alainneedle1]

Thank you Stephen, I have to say I'm pretty happy with this PC.

What I did before building this rig is to research all the best component I can put together (as of today) without creating a bottleneck somewhere and if one is created I tried to make it as small as possible.

I always work around the mobo first and fill it up with the best he can take, the one I was looking at (and had for three week) was the Asus P6X58D...why this one...because you can get the same result for FSX as the one I have or the Rampage III (under normal conditions)  but adding a PCI-e SSD card including my GTX 285 and the Zotac GTX 480 (covering three slots) I was running out of PCI-e slots......now I have to say I'm not convinced anymore about the stability of the Asus vs Rampage III and the Gigabyte UD-9  been the same at high speed (CPU and memory wise).....I think you really pay for what you get....I am able to run this poppy at 4.67GHz with the memory at 2000MHz with a timing of 7-7-7-21 1T on air....now that's telling about the mobo...

Does the UD-9 worth $300.00 more for a little more stability at high speed....hell no....but you know how it goes....you have to pay to play.

As I can seeso far my bottleneck is the GPU because this z-drive is fast as $h.. and the CPU + memory at that speed can take a lot of beating....and don't we forget...in two years my rig will be called an antique....

[Spirit Flyer_old]

Hello everyone,

I encourage you to run these tests and compare each part of your computer's performance against the other parts in the same system. That way you can usually pinpoint if and where you may have chokepoints. If you are not too shy we would like to see your results too!

Here is the address for the software that is free for a month's trial:

http://www.passmark.com/products/pt.htm

Kind regards,

[AVI8TOR]

hello all

I am new to this anyways i am currently wanting to overclock my cpu using the system i have below,now i was  thinking of getting the v8 coolermaster to do so.Or do you think another cooler may be more suitable for my system specs.Gigabyte GA-EX58-UD3R Mother Board

Intel CORE i7 920/2.66GHz/4.8GT QPI/8MB CACHE – with standard Intel heatsink and fan.

3 x Corsair 2Gb DDR3 1333Mhz

LG 22x DVD+/-RW

750Gb Seagate SATAII HDD with 32Mb Cache

1 x Gigabyte Radeon HD4870 1Gb Graphics Card (which I’m about to replace with the 2 x HD5670 1gb Cards)

Windows Vista Home Premium 64Bit

Antec 300 Series Case

Asus 750W PSU

your advise would be greatly appreciated..

[Spirit Flyer_old]

Updated Dec. 15, 2010 with 580 GTX

480 GTX  out, 580 GTX in

Unigine Heaven DX11 Benchmark

Download Unigine here:

http://downloads.guru3d.com/Unigine-Heaven-DirectX-11-benchmark-download-2414.html

PassMark Benchmark

PassMark 580 GTX Benchmark

Download PassMark here:

http://www.cpubenchmark.net/

Download 3DMark Vantage  here:

http://www.futuremark.com/benchmarks/3dmarkvantage/introduction

Download 3DMark11 here

http://www.3dmark.com/3dmark11

Kind regards,

[AlexMD11]

Hi all,

Hope all of you have had a nice Christmas! So I ran the PassMark Performance test and I got the following results:

So my overall score is 2172.8 and I have the feeling that is rather low compared to SpiritFlyer's. Our system specs do not seem to differ that much.. would it be my harddisk drive that is making such a big difference? Now that I know this result, how should I proceed to increase performance/find bottlenecks?

Earlier I pointed out in another topic how the GTX580 did not seem to be making any huge difference for me over my previous GTX285; I guess that the 3d mark score confirms this...

[Spirit Flyer_old]

Hi Alexander,

A couple of things that might help are to run the test with hyperdrive turned on and your 580 GTX clocked as high as it is stable.  The highest I ran mine in these tests was 900 Mhz, but be sure to test yours before putting it under this kind of heavy strain. Make sure you have a good data disk already in your CD before starting.

Also, if a Passmark test program pop-up informs you that your settings are not optimal for the test and asks if you want to change them, say yes. It will then automatically adjust something somewhere. If no pop-up, then your settings are already set right. Afterward I reload my 8xSQ profile before starting FSX again just, to make sure.

Except for perhaps your drive, your system looks like it matches or exceeds mine, so you should do a lot better than these numbers I should think. Have you tried the Unigene  Heaven or the Vantage benchmarks?

Try again!

Kind regards,

[AlexMD11]

Hi Stephen,

Thanks for your response. As you suggested, I overclocked my GTX580 to 920MHz which ran completely stable for 30 minutes in MSI Kombustor.

After that, I tried the Heaven benchmark and got the following result:

So as compared to you, my FPS are on average 10 lower. In addition, I felt that there were microstutters all over the place, but I am not sure whether that is normal for this benchmark or not. Not smoooth at all!

I ran the Passmark thingy once more, which resulted in to the following figures:

So with the overclocked GTX580 it is all a bit higher, but I am still not convinced that this would be the max for my rig.. What I don't understand is how my CPU scores around 7500 points whereas yours leads to a value of about 9000! What could explain the difference between my 950 @ 4.3 and your 930 @ 4.3?

Thanks again!

[Spirit Flyer_old]

Hi Alexander,

Before I forget, Let me advise you not to run Kombuster, Furmark or OCCT, or any other benchmark program that gains control over the voltage setting in your equipment.  According to both Nvidia and AMD these programs, as well as others that they refer to as "power viruses," will destroy your video card. As a matter of fact nearly all the benchmark testers have ceased using them altogether. In addition, the 580 GTX now has a built in alteration that detects these programs. and throttles back the power.

Here is a link that explains it in a bit more detail:

http://www.geeks3d.com/20101109/nvidia-geforce-gtx-580-the-anti-furmark-dx11-card/

So, Why the differences?

The explanation for the different results of our two computers which contain some similar components, comes right back to the reason this thread exists: the value of overclocking in a well balanced system. Your system is not balanced, which is amply evidenced by the poor performance of your hard drive.

Those numbers you posted in both tests indicate that the performance of your machine's individual components are not working to capacity because they are being held back by your hard drive, and possibly by other hardware factors.

Your CPU runs 20% lower with the same speed, which means there is either something wrong with it, or it is suffering from bottlenecks from either the motherboard, the hard drive, the ram, power supply, or other factor, or a combination thereof.

Although your 2D application runs 19.51% faster than mine, your 3D graphics card, despite being a 580 GTX, is 16.11% slower. It is being handicapped by a lower functioning CPU, feeding it less information, as well as it perhaps being pushed beyond it's own limitations. Because it could not run "Heaven" smoothly, I suspect it was overclocked well above it's stable setting.

Your ram performed 34.84% lower, either because it is faulty, or is restricted by the slower CPU, the sluggish hard drive, improper bios settings, insufficient power (probably not, or it would just blue screen), a motherboard problem or a combination thereof.

The Raid SSD setup in my computer is 655.85% faster than what you have installed. Your hard drive is starving everything else, and therefore must be the major, if not the only problem with all the other components under-performing. My guess is that it is not the only factor, but it is the most obvious, and the first thing to replace.

The final result is that despite having a similar CPU and GPU, the two computers have widely different performance capabilities, with the SSD supplied unit scoring 49.84% higher. More than that, your recounting of the roughness of the Heaven benchmark (mine was butter smooth) indicates that your video card may not be able to overclock to what you tested it at. Whether it can or cannot overclock so high is just the luck of the draw, or, some other factor may be at play, such as a corrupted or faulty driver. This can explain why your Heaven test scored both 26% lower in FPS and overall scores, while your minimum frame rate was less than 50%.

It's all about the numbers

All these numbers point to one or more serious bottlenecks in your system, with the primary culprit being your hard drive. In any case, you should replace your primary drive with a solid state drive that can run the operating system and programs much faster. That should make everything hurry along a whole lot better. Re-running the same tests after a new drive is installed should reveal the rest of the story.

You have a great machine and are a long way down the road to having excellent performance. Match things up a bit better and you will be laughing in glee. I am not a guru or expert in any fashion, so I could be wrong in some or all of these conclusions, but that is what it looks like to me considering the data that you provided. I wish you well in your pursuit of improvement, and will help in any way that I can if asked.

Kind regards,

[AlexMD11]

Wow Stephen, thank you so much for this clear explanation! I had no idea that a harddrive could serve as a bottleneck to this extent so I never considered replacing it.

The thing its though that the affordable SSDs have rather limited capacity and especially in a RAID0 config I can't how my full FSX install with addons will fit. I also do not have a lot of knowledge on hard drives, so what kind of setup would you recommed or do you have any good resources I can consult?

As for the videocard, perhaps it is clocked to high, but I also experience the microstutters while not overclocked, so I guess it is still something with my setup that is not correct.

Thanks again for all your effort!!

[Spirit Flyer_old]

Wow Stephen, thank you so much for this clear explanation! I had no idea that a harddrive could serve as a bottleneck to this extent so I never considered replacing it.

The thing its though that the affordable SSDs have rather limited capacity and especially in a RAID0 config I can't how my full FSX install with addons will fit. I also do not have a lot of knowledge on hard drives, so what kind of setup would you recommed or do you have any good resources I can consult?

As for the videocard, perhaps it is clocked to high, but I also experience the microstutters while not overclocked, so I guess it is still something with my setup that is not correct.

Thanks again for all your effort!!

Alexander,

What hard drive do you have now and how big is your root FSX folder?

Kind regards,

[AlexMD11]

Hi stephen,

Currently I have 1tb WD green caviar with a separate partition (300gb) for fsx while the root folder is about 60gb. After some research I couldn't resist and ordered two 120gb SSDs which should be delivered tomorrow . I intend to use one for my OS and one for Fsx so that I have some room for expansion. Hope this will help hate this hardware addiction I can't resist . Only thing I need now is I guide how I can install all of this properly and I hope I will be fine .

[Spirit Flyer_old]

Hi stephen,

Currently I have 1tb WD green caviar with a separate partition (300gb) for fsx while the root folder is about 60gb. After some research I couldn't resist and ordered two 120gb SSDs which should be delivered tomorrow . I intend to use one for my OS and one for Fsx so that I have some room for expansion. Hope this will help hate this hardware addiction I can't resist . Only thing I need now is I guide how I can install all of this properly and I hope I will be fine .

Hi Alexander,

It just so happens that I have the same drive as you do as one of my primary storage drives. In order to determine how it might effect your performance, I set it up to be tested by the AS SSD benchmark program which accurately measures various read and write speeds. Here are the results for a WD 1T Green Caviar 7200 rpm Drive

Here are the retested results for my primary hard drive which consists of 2 x G.Skill 120g Pheonix Pro SSD in Raid 0:

As you can see the SSDs perform between 5 times and 332 times faster for the read function and from 2 times to 240 times faster for write. Altogether the mechanical drive only rates between 2.9% and 3.75% as high as the SSDs.

It is no wonder your system is crippled. You have some of the best equipment in the world smothered and rendered helpless by such hopelessly antiquated technology as a mechanical drive. It just won't work.

Even the best of the 7,200 rpm drives are scarcely better. Here is the result of the test on another one of my back-up drives, the WD 640g Black Caviar 64 cache 7200 rpm drive:

You are really going to like the difference in performance with your solid states. I advise you not to try to ghost or anything, but go for a fresh OS and FSX install.

Good luck and please let us know how it goes.

Kind regards,

[AlexMD11]

So I got my two SSDs in the early afternoon and did a fresh install of Win7 and FSX. The loading times are amazing and the sim is already so much smoother!!! Thanks for all the advice The Passmark score is now up to 3000, though my CPU and GPU are still not as productive as yours (they still have the same values of 7500 and 4700, respectively). Now if I could only get rid of the stutters induced by VSync, I would be the happiest man!

Thanks again!

[recores]

What I can say is clear even to a child. Thank you very much for the exhaustive material.
 

[W2DR]

Did anyone notice how old this post is? OC and SSD? Give me a break...................

[Nick Cooper]

Hello Doug,

almost everyone, I would think.

I'll put it back to sleep.