?
Solved

Overclocking Graphics Cards

Posted on 2005-03-23
4
Medium Priority
?
1,252 Views
Last Modified: 2013-11-08
Are there some current guides on overclocking current graphics cards?  I see some stuff on graphics cards but it's old and I'm not sure if applies to current models.  If you have experience doing this, please explain the overall in's & out's.

I suppose overclocking would be defined as decreasing the safety margin between the actual clock speed and the point where instability begins to occur (cranking up clock speed).  To allow for production variations, this is probably set to a wider margin in mass-produced stuff than what could be done on an individually tuned case-by-case basis.  But anyway the amount of possible increase would seem modest?  I wonder if it's even worth the trouble.

I wonder why nvidia 6800 ultra clock speed is so much slower than ATI X850 (540mhz).  You see a lot of nvidia cards pre-OC'd from the supplier.
0
Comment
Question by:mark876543
[X]
Welcome to Experts Exchange

Add your voice to the tech community where 5M+ people just like you are talking about what matters.

  • Help others & share knowledge
  • Earn cash & points
  • Learn & ask questions
  • 2
4 Comments
 
LVL 69

Accepted Solution

by:
Callandor earned 1200 total points
ID: 13615443
0
 
LVL 12

Assisted Solution

by:GinEric
GinEric earned 800 total points
ID: 13617519
FYI

When Intel, AMD, National Semiconductor, GTE, RCA, and all the other chip manufacturers make a "batch" of salamis [silicon wafer sources that resemble a salami or bologna before they are spliced], they assign it a a batch number.

All chips manufactured from this salami have that run/lot number on the final product, the chip, be it a simple gate or a 64-bit microprocessor.

At the point that the first few chips are completed, they begin testing them.  One of the test is the speed/clock/load power stress test.

What basically is established is how much heat is generated for this run/lot and how it affects failure in the form of the chips "burning up."  That means any substrate circuit failing due to opening or short to another substrate circuit.

Most chips within a run/lot will fail at nearly the same temperature and load consistently.

This determines the "clock rate" of that particular run/lot of chips.

Over the speed, most chips will have a very short lifetime,caused by the excessive heat which cannot be dissipated by any amount of fans.  It is internal to the chip and it is determined by many variables.  But the run/lot stays pretty much within the limits set by the testing.

The actual clock speed at which a microprocessor executes bares little relationship to the pseudo-scientific advertised rating.  For example, no chip executes at 1 gigaHertz, that is just the frequency of the base crystal, it is not even the frequency of the clock.  The clock is stepped down by a series of excitation oscillators and gates packaged as one unit on the motherboard.

Considering that it takes usually more than a few nanoseconds for any electrical signal to pass through even a single gate means that it is not currently possible to achieve such execution rates.

The overclock has the same effect on all printed circuits on the motherboard, it tends to overheat them, shortening the life of the motherboard, sometimes drastically.

Overclocking a video can be worse, it can damage the CRT.  There have been instances of overclocking the video which have resulted in the "implosion" of the monitor, as this is one of the results of feeding frequencies to the high voltage circuits post Vertical amplifier and post horizontal amplifier causing the high voltage multipliers to resonate at an asymptotic frequency resulting in implosion.

The gain is minimal, if it exists at all, because the actual clock delivered to chip devices is some fraction of the crystal frequency, depending on what the phase characteristics of all components will allow.  Usually, it is a binary fraction, that is, 1/32, 1/64, 1/128, etc., because current crystal are around 1 gigaHertz, but current technology is between 1 mHz and 8 mHz onboard the chip.  There are different internal clocks for different functions.  For example, the memory transfer bus from cache may be able to operate at 166 mHz, but this is because of the speed of the one-gate proximity RAM onboard.  The CPU internally, on the other hand, may require longer write pulses, and thus be limited to one quarter of the RAM, or around 8 mHz.  The whole thing boils down to junction time barriers and gating propogation delays, combined with the number of clocks, internally, that it takes the microprocessor to execute one full instruction.  This is hardly ever anything more than about 8 mHz.

In addition, many microprocessors are self-timing, not allowing execution above certain predefined delay limits.  Many internal clock are generated based on single shots, a physically limiting time mechanism.

What this means:

Neither nvidia nor ati runs at anywhere near 540mHz.

The difference in numbers is an imaginary rating that refers only to the crystal, not the speed of video processing.   Such numbers are used by marketing and advertising to suggest that their boards are faster.  They are deceptive.  Those are just the resonant frequencies of the substrate crystal used to supply the eventual clock and sync for the monitors.

You can only shoot as many pixels as the manufacturer of the screen will allow, and you can only shoot them as many times per second as the laws of physics will allow.

In any pixel, there are actually three color dots.  A good magnifying glass on any trinitron product will show these three different colored dots, making up one pixel.

"MPEG-1 provides images of 240 lines x 360 pixels/line, digital transfer rates up to 1.5 Mbps, and compression ratios of about 100:1. MPEG-2 provides a higher quality picture - 720 horizontal lines x 480 vertical lines (pixels/line). The MPEG-2 standard is used for HDTV, and will be used for cable broadcasts. "

So you can only get a higher speed, if you think about it, with an HDTV monitor.

The frequency to the monitor is set at about 6 mHz and in that bandwidth it has to transfer all pixel changes.  Sure you can probably find one that is 12 mHz, but then, what good is it on a 1.5 mHz cable or DSL connection?  Even the 6 mHz one is stepped down first to accomodate the lower Internet bandwidth.  After being double side band suppressed, the result is sampled at about 1/4 the actual video transmission frequency of 3.3 mHz.  So you get at most about a 768k video frequency.  Which is why games look more like games and not like a movie on television which has a much more "real" look to it.

This, of course, is not going to speed up either your computer or your video card.

Because they are all limited by the physical properties of the monitor itself.

And this is just the tip of the iceberg in why it's basically a waste of time to overclock.

If your car engine can't go over 7,000 RPM before the engine breaks apart, then giving it more and more gas is not only a waste of time, it can result in the failure of the critical component, sometimes with disastrous and/or catastrophic results.

The whole area of overclocking is actually wishful thinking on the part of a lot of computer users, and some gamers, not unlike the special effects [all done with film, by the way], for all the movies about supercomputers, superhackers, the pretty graphics inside of a mundane silicon chip, virtual reality, etc..  The Virtual Reality is the film, not the computer graphics.

Film density, by the way, is in the trillions of pixels per inch.

Laws of Physics again, I'm afraid, as film records directly to the atoms and molecules, and not to a RAM, hard disk, or some other electronic device.

Wouldn't it simply be more fun to just play the game?


0
 
LVL 23

Expert Comment

by:sciwriter
ID: 13617629
It is easier to overclock CPUs than it is for graphics cards -- CPUs (AMD at least) have jumpers to short, graphics cards rarely do.  They will usually only work at ONE bus speed and that is it.  YOu can try setting the AGP bus to 8x (if the card is 4x) see if it can handle it -- but you will likely find it cannot.  So you have 2 choices --

1. buy a faster bus graphics card, only if the MB will take advantage of it, or
2.  Live with the bus speed you have, and plan your next system purchase now....
0
 
LVL 69

Expert Comment

by:Callandor
ID: 13618682
GinEric,

>Overclocking a video can be worse, it can damage the CRT.  There have been instances of overclocking the video which have resulted in the "implosion" of the monitor, as this is one of the results of feeding frequencies to the high voltage circuits post Vertical amplifier and post horizontal amplifier causing the high voltage multipliers to resonate at an asymptotic frequency resulting in implosion.

I think you have the two effects mixed up - overclocking is independent of what frequency you set you video refresh rate to.  I agree that you can damage a CRT with improper signals, but an overclocked video card can use the same refresh rate as an non-overclocked card, eliminating overclocking as a cause of damaging CRTs.
0

Featured Post

Connect further...control easier

With the ATEN CE624, you can now enjoy a high-quality visual experience powered by HDBaseT technology and the convenience of a single Cat6 cable to transmit uncompressed video with zero latency and multi-streaming for dual-view applications where remote access is required.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

Great sound, comfort and fit, excellent build quality, versatility, compatibility. These are just some of the many reasons for choosing a headset from Sennheiser.
A clone is a duplicate copy. Sheep have been cloned and maybe someday even people will be cloned, but disk cloning (performed by the hard drive cloning software) is a vital tool used to manage and protect data. Let’s look at what hard drive cloning …
If you’ve ever visited a web page and noticed a cool font that you really liked the look of, but couldn’t figure out which font it was so that you could use it for your own work, then this video is for you! In this Micro Tutorial, you'll learn yo…
In this video, Percona Solutions Engineer Barrett Chambers discusses some of the basic syntax differences between MySQL and MongoDB. To learn more check out our webinar on MongoDB administration for MySQL DBA: https://www.percona.com/resources/we…

765 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question