caught blindsided

[underclocker1]

Ok, here is the scoop. Took one of the gt9800x512 cards out of my FX machine. I don't play many games, so, not needing two sli'ed cards together for internet surfing, I put the card into my daughter' would be FX, well it is a quad core anyways. A week ago, I bought her a new game. She set the graphics to max out. It ran ok, in fact really good, but then the game asked her to upgrade the nvidia driver, by clicking on a upgrade button, where ever that was, maybe in the game? The upgrade was successful. The game went in to near tilt speed, but within about 5 minutes, the system blue screened. I discovered the gt9800 hot enough to boil water on, & the bench marks in vista went through the roof for the video. I put this huge 4 pipe cpu cooler on the main heat sink with grease to cool it off. Well that works for about 15 minutes now before it blue screens. I've concluded that somehow the gt9800 was overclocked by whatever means available, somehow by the game she was playing. She found on the web that this is a common problem, when the video driver is upgraded with the overclock ports enabled by the upgrade driver. I can't make no sense to it other than the card is running full tilt speed, & running at least 200 + degrees before it blue screens. Help! How do you revert back to the original clock profile? That is to say, how do you down clock the card back to a more normal frying eggs speed? This is no joke. I am sure the card is maxed out overclocked at this point. It is in a state of thermal run away, it is clocked so high. I have about 15 to 18 minutes from a cold start to do what I have to do, before it blue screens. I have a 25 cfm blower I am going to mount on the cabinet to blast the cpu sink, to try to buy more time before the blue screen. Help! Help! I sure hope someone out there can tell me what to do, & that is, I hope not, drop dead..... Lay it on me, point for point. I follow good directions very well. Thanks, & happy gaming......

[Waco]

How hot is it actually getting? Use something like CPU-Z or Hardware Monitor to check temperatures. Does the fan spin up properly when it gets hotter under load?

 

The driver isn't overclocking it...but it is possible the fan isn't ramping up properly.

[bishop245]

What game did this to the system. Go get MSI afterburner or EVGA precision and see if you can lock the clock speeds that way.

[underclocker1]

What game did this to the system. Go get MSI afterburner or EVGA precision and see if you can lock the clock speeds that way.

 

Ok.. Hum? I will try replying to both of you at once. The original & the added on 4 pipe cpu sink fan pull a wheel stand when you first power up the system, but I would say the on board video card fan doesn't break any speed records once it boots to the desktop, down the track to the blue screen, it stays the same speed, not real fast at all. Now the game, well she has a whole closet full of them, but she thinks the game 'Assassin's Creed II was the recent purchased game, that she was playing when all of this happened, yesterday. By the way, she has a Samsung flat panel 2233RZ, & it was previously set to 60 HZ refresh, but now I find it running 1680x1050, at 120 refresh. I tried to switch the the refresh back to 60 hz, but it keeps re-setting back to 120 refresh again, all by itself. ? As far as I know, the 120 refresh is for 3D games with the Nvidia glasses, but I am not sure of that. We do have the glasses & the infared sync pyramid, but the game she was playing is not a 3D as far as I can tell, such as Avatar 3D. Does that tell you anything?

[underclocker1]

Ok, we just found MSI Kombuster & installed it, but now I have to pull the standby g-force card & re-install the gt9800 to have the kombuster take some measurements. This will take time to do, with all of the rebooting, et cetra. Stand by while I do all of that.

 

Ok. Looking at this MSI Kombuster prog. app. She said this app was listed as the afterburner app you suggested. I still have a cut rate video in the system, & likely it might not be showing any clock settings for this card. Since I don't want to kill the gt9800 from over heat, I am treading lightly on the boot up time. So, is there specifically a place I will find a clock lock down command in the afterburner app, & does the kombuster app sound like the same app? I, just was updated, we have the Evga precision installed & ready to go. Now you say to try to lock down the clock frequency? I hope my memory of last night is correct. I believe I read on this website about a gt9800 version manufactured by Gigabyte, that was marketed with software that could access the PLL of the video card, to set the clock multiplier, memory, & bus speeds directly with this software. Does anyone know the name of that app, & most especially, where I can download this prog. from? The name of the app would be a great help. I might track it down from there. It always helps to know what needle you are looking for in this web cloud.

I think for the moment, I am going to hold off putting the card back into the system, until I find out more about this. One other thing I forgot to mention earlier. After a cool off period, when I boot the system, after a blue screen, & I repeat, that happens when the area of the processor gets so hot, including the add on cpu sink, that I can't touch it, but for a moment without getting burned. After reboot, the windows message says the blue screen was caused by 'the video driver stopped responding'. It goes on about being corrupt. So I did a clean install of the driver from Nvidia website, after un-installing the original driver in system properties. No change, what so ever! The video bench mark in vista is 7.2, when originally it was a 5.6 or a 5.9? Help! That card cost me a lot of money. I am at least going to try to recover it. I'll be here looking for suggestions for a spell. Fire away.

 

Oh! First. I want to thank both of you for your' help. While it is always possible the card has failed for other unknown reasons, we found the settings in the Evga precision app. to adjust the GPU clock, memory, & voltage. I think that is the prog. I need to use. I am on my way looking for the clock frequency, & voltage defaults. If that does not fix it, then I will square file the card. Either way, good or bad, I will post what the outcome is, in the next day or two. Many thanks for the help. "Every day, hot or cold, north or south, east or west, somebody somewhere runs into a snag with their toys. Hopefully the card did not fail, & the problem is related to basic clock programing. I will let you know in short time if the fix was in the bag, but for now... Many thanks again. I know a few, not many, gamers, & I will tell them about this site. Thank you, underclocker1

Edited October 24, 2012 by underclocker1

[Stonerboy779]

Punctuation, spaces, paragraphs... anything please?

 

Your chunks of text are way too large.

[Waco]

Did you ever check the fans?

[underclocker1]

Punctuation, spaces, paragraphs... anything please?

 

Your chunks of text are way too large.

 

Stonerboy,

Please forgive me for the "fat chunk" ascii binary file, often read as text in a blog. More often, there is a limit to how many characters can be input to a blog file, I think it is only 160 characters maximum on cell phone, & twitter messages.

 

I am a speed typist. I can easily keep up with a 150 word per minute dictation, while I can transcribe from hand written text, to keyboard copy, up to, & exceeding 200 wpm. However, I am a slow speaker! Why? Because I formulate my spoken words, & thoughts, not once, but more than once, before I speak my peace verbally. I do not have to pre-plan my words ahead of time, what I want to say, sort of like "not" memorizing a speech.

 

On the other hand, Stonerboy, I have all of these various limitations, such as the 160 character text limits, that cause me to run together many thoughts, to keep the idea in focus, with as much detail possible, in the shortest text space. The objective is to give the reader all available information to make a judgement on what I am asking, or telling.

 

Stonerboy, when I am dictating from my thoughts directly to the keyboard, most often my mind bypasses the second, or even third, formulation of thought. The input to the keyboard, then contain' fragmented thought, as well as, incorrect "Punctuation, spaces, paragraphs", for the sake of continuity of thought VS character space.

 

Since I actually think slower in mental text, than I physically type, there are times when I actually type the next thought from mentally composing it, before I complete the previous thought in typed text, purely because I have already typed the fore thought, but was already formulating the next thought, resulting in a cross type of text expression. This is a problem, to the untrained reader, but not that much of a problem to a subjective reader. However, there is still a problem with the assembly of the thought, even to the subjective reader.

 

Most outstandingly educated subjective readers, have no problem assembling the text expressions, into processed information that can be used to reach a plausible answer to a question, based on, fragmented information. That is, if the fragmented information contains enough fragmented information to come to an acceptable conclusion.

 

Stonerboy? Are you with me? Do you understand, that I did not intentionally overload your' thought processes with my previously posted text, yesterday?

 

If that be the case, I am not sorry, but I will in the future, consider your' request to place all text into a proper diction state. Not just for you, but for everyone reading my text. Ok? Is that a deal?

 

Stonerboy, I have decided tonight, not to reply to all of the other blog'.

 

While they all have merit, I can not answer every blog for one reason. I have already answered most of the questions asked last night, & as well, the answers were gifted to me by those blog' with out a doubt sent me on the path of the correct discovery. I am only answering your' blog tonight, with the answer to all of the questions you want answered. I found out just what I need to know, to come to a ratified conclusion, to my question' last night. I intend to keep the text answer in context so everyone reading this blog reaches the same understanding.

 

Last night I stated, that I took a #2 single wide gt9800 out of my race car FX machine, & installed it in my child' near FX machine, so she could play games faster in her PC. The gt9800 cards were factory installed in my FX, as SLI bridged 2x512mb. The FX came with settings in the Nvidia control pannel default allow the program (game) to control the video card. I use the FX only for surfing the web cloud, not for gaming! I removed the #2 card, & installed it into my child' PC, say one year ago, so she could play higher graphic level games on her PC. I clean all of the PC' in the house somewhat every 2 months for dust collecting in the heat transfer sinks.

 

Aug. 2012 is the last recent cleaning.

 

A few weeks ago, I gifted her with a high res game, maybe high enough res to cause an overheat of the GPU? Ok.

 

Last night, I requested information from anyone, to tell me the name of the software used to access the Nvidia GPU, thinking the card was overclocked somehow by the game my child was playing when all of this happened. Thank' to this website, someone gave me the software name Evga Precision, along with a bench marking app., ?Ko?, don't remember the name. Without those two app., I would not only have never figured it out what happened, but I could not have reached an answer what to do, so this failure never happens again.

 

The first fragmented answer was? Is the fan 'ramping up'? I noted throughout the history of cards installed in my FX, the fans 'never 'ramped up' so to speak, except when the system was powered up for boot. At this power up level, the fans in the system went to maximum speed for some short period of time. My conclusion was the system was testing the cooling fans by way of bios install hardware state at time of power up. Surely I am correct! My future text in this blog might explain exactly why!

 

The Evga precision told me first, right up front.... The gt9800 video card was set to "manual fan speed". This fan speed was set to 30% speed. Checking the Nvidia control panel, I found the default settings to be controlled by the game program. I went into my FX machine with the same downloaded Evga program, & found the video card #1 set to the same 30% manual fan speed. Both cards came from the factory installed in my FX, with this hidden manual fan setting. So the "ramp up" was not possible under the default factory settings......? Well maybe the 30% on the first card, & the 30% on the second card, might have been good for the games, or programs as a SLI bridged group state, but alone, the #2 GPU was overheated from the manual settings of 30%, & subjected to game overclock, without any control of the cooling fan!

 

That means, if someone used a game fast enough to overclock the GPU while playing the game, the fan would not "ramp up", causing an overheat of the GPU!

 

Any questions Stonedboy?

 

While I thought the card was overclocked at first post to this website, I found out that my child had already benchmarked the card in vista, resulting in a 7.2, while she was playing the game. She paused the game, then ran the vista benchmark, where I originally found the card to be 5.6 or 5.9, on previous vista test. She said she paused the game because it started to lock up, so to speak, how she described it. She knows what a lockup of the video is! She knows how to recognize an impending video failure, as she has 15 years of experience on a hardware computer using every OS. She was telling me it started refreshing, is the right word, or flashing to blank screen, then back to the game screen, over & over again. When it stopped flashing from blank screen to game desktop, the game minimized with a popup from the system tray, stating there was a Nvidia driver update. Click on the update to install. She said, 'she clicked on the install' then after a time spell, the system said the install was successful. Within 5 minutes or less after that point, the PC blue screened. She had not even started to play the game again. It was still minimized.

 

When I tested the card after downloading the Evga app. late last night, I found the card default 30% fan speed on manual, not automatic, or so to speak, ramp up. I then went to my race car, FX, downloaded the app. & tested the #1 video card. I found the first card also set to manual 30% fan speed.

 

No question in my mind, the prog. app. controlled the clock speed, ramping up the GPU & memory clocks, but by factory default of the fan control as in manual setting of 30%, was not able to 'ramp up' the cooling fan to match the GPU load. Am I using the correct terms, so everyone can understand my text'?

 

Stonerboy, I do not want to leave you out of this conversation text. So I want to address you directly with this question, if you would gift me an answer? Are you having any problem with my text, where I am not using the best of the best language, needed to understand my diction, here in this post? Don't be shy. If I am over your' clock speed, please tell me, along with everyone else reading this post.

 

I did not realize the implication in the name I chose to log on to this website, at the time I registered. Underclock is an understatement to behold by the reader.

 

With that said, everybody not in the know, should either go south for the winter, or read this text more than once. I will start with the clock, a divide by 'N' PLL. The divide by 'N', is the key to the whole ball of wax. The PLL is possibly the least understood digital circuit, & is the reason why video cards fail!

 

In the case of my video card failure, it was not necessarily as a result of the fan not being game controlled to 'ramp up', as most games are low profile graphics, but on the other hand, high res games that reset the GPU clock on the fly, called programed game over clocking, that is the second reason behind, failures of a video card of any manufacture.

 

To understand how the PLL divide by 'N' works, one must understand exactly what the relationship between the master clock & the multiplier really means. Master clock, multiplier x = operating clock of the processor. Example: master clock = 6mhz x 12 multiplier = 72mhz GPU. or CPU, a given state. How far can the PLL control this 72mhz state through the divide by 'N' ? 1%? 5%? What is 5% of 72mhz? That is a lot of control at that low frequency! Not impossible? In digital control, this would be just another day, or is it? There still needs to be a a relationship to the output, in this case, the video card, say 1680x1050. Not every clock times x, memory times x throughput bus ratio, will divide to the 1680x1050 with the highest graphic' speed possible. Not only that fact, but this GPU clock/memory frequency must be recorded in a type of flash memory on the video card circuit for it to function in a game controlled environment. The game needs to revert to default after the game is terminated. The Evga program needs to be set by clicking on save. That is if you click to save, then progress to the game for play, the video card is overclocked by the game over the preset of the Evga program, but when the game exits, the video card reverts back to the preset clock saved in Evga.

 

So to speak, if the clock/memory frequencies are not close enough to be locked onto by the PLL, the PLL will try to lock onto the next nearest combination, while running under it's current phase lock. So long as the system video driver does not totally stop, (crash), this close match in the clock/memory of the video configuration will remain somewhat, but not totally stable, under some, but not under all load conditions. This is what causes instability in the video card. Also,the extreme ratio's will fail the GPU from restarting the video driver, but not necessarily cause the GPU to stop functioning while it is running in service before restart. There are times when the video driver will recover from these conditions. The key is restarting. If the PLL is at the wrong phase lock frequency saved in Evga when the PC is shut down, or the system experiences a blue screen crash during a game overclock, the GPU may not start when you reboot the Pc. This likely results in a failed video card. General recovery may not be possible.

 

This is likely what happened to the #2 gt9800 card, as when it blue screened, the GPU was scaled to an overclock condition by the game, & never recovered after the blue screen.

 

 

Underclocking the GPU is more dangerous than overclocking, because the PLL can lock on a smaller margin of frequency, (spectrum bandwidth) as the frequency of clock is reduced. Therefore, underclocking is likely to kill the video card more often than overclocking, given the range of the PLL, if the PLL can not lock while the video card is booting during PC startup.

 

Are you with me Smokeboy?

 

Moral of the story.... If you overclock to play a game? Well that is ok. If you go too far and the card crashes, well that might not be so good. So long as the system does not crash during the overclock, well that is good. But, however, after playing the game under a saved clock frequency in Evga, you need to use the Evga program to reset the video card back to default, before you shut down the PC. This is really the rule to follow.

 

If you do not do that, you risk the possibility of the video card not restarting the next time you boot the system. Most especially if the card was down clocked, but not without the good probability of the card failing on a restart from overclock.

 

Trust me. I am not telling you the wrong information. Those of you that think I am wrong, maybe you think you know more? Do it to it. See for yourself if I am giving you the wrong information. Do a max overclock & save it with Evga, or for that matter, a minimum underclock of slightly better than 8% or more of the default GPU frequency, & see if your' card will reboot, after it has cooled off.

 

From this point on in all of this text, I am issuing a warning to all 'Overclockers', as well as, 'Underclockers'.

 

Do not shut down your' PC without reseting the GPU to default with the Evga control program first and for most. A game could be crashed upon exit, & leave the GPU overclocked without your' knowing it.

 

To do so, as in shutting down the PC with the video card in an overclock state, can risk permanent failure, of the video card, & this is more than just some prank possibility.

 

Any questions Stoned?Blood?

 

your' to prove wrong,

underclocker1

[Stonerboy779]

Not to be a dick or anything but do ease up and try to post in a more concise manner. You clearly have a decent intellect and great english skills, so lets use them

 

What I gather from your essay per say is:

 

1 - The fan speed seems to have defaulted to a manual 30% setting apposed to the general setting most GPUs have which is to ramp up with higher loads (temps)

2 - Your daughter was able to work out that the GPU was locking up (this is most likely related to heat issues stemming from inaadiquate cooling from a fan been locked at a low speed)

3 - You may have some driver instabilities too

 

Now then my first suggestion is to use EVGA Precision (which you already have installed) to set up a fan curve. You can manually set points where you would like the fan speed to be at certain temps. It something you need to really play with to find the best balance of noise and temperature.

 

Once you have this sorted run a few runs in a benchmark like 3D Mark 11 while monitoring the temps and the fan noise you should be able to tell if it now ramps up properly. If however it is still overheating etc, there are other courses of action that can be taken mainly a good dusting and a TIM replacement.

 

It would also be in your best interest to uninstall nvidia drivers and then run a driver sweeper before doing a clean install of the video drivers.

[bigball1]

Im just going to point out a couple points even thought your dealing with Stonerboy who knows his . just like Waco.

 

1. You monitors native res is 1680 x 1050 @120hz, so it will always revert back to this setting. Google EDID.

 

2. Its a 9800, this is 5 generations back now so i wouldn't be overly worried if this card is on its way out.

 

3. Setting your overclock setting back to normal before you shut down. "If you don't do this is will cause failure". Where on earth did you every hear of this??? Most of us on here including me have our systems overclocked. This is including the CPU and GPU. A system not rebooting with a overclock means its not stable, which could mean to high of a clock speed or not enough voltage, there are a few more then that but those are starting points.

[Hyper]

You said something about putting a different cooler on the card. I might have missed it but what cooler? does it cool all the card (gpu, memory, vrm's) or just the GPU?From what your describing it sounds like a heat problem, which could also be case air flow related.

[Waco]

You said something about putting a different cooler on the card. I might have missed it but what cooler? does it cool all the card (gpu, memory, vrm's) or just the GPU?From what your describing it sounds like a heat problem, which could also be case air flow related.

It sounds like he has a manual fan profile set somehow. All he needs to do is set the fan to ramp up with temperature as others have already described and he'll probably be fine.