Overclocking the GT 555M card.:-)

Well, it's not mentioned in the specs and no one here has managed to fry his 555M yet, so no one can really tell Earlier in this thread someone advised against permanent OCing, because by doing that he had fried his 8600M (or similar). ...But these were a badly manufactured GPUs and they were prone to fail even on stock clocks.

 

I seem to be experiencing this as well.

Anyone have a solution?

 

There are other threads related to fan behavior both on L702X and L502X, which has the same issue.

 

Ok, I got it. It's like a lottery and you never know the result

 

If you have the 144 chip version remember there are the same chip running at higher clocks and voltages being sold, so really all you need to worry about are temps.

 

Ok, so '144 chip' means 144 unified shaders, right? Then I got lucky Thank you for this info.

 

Yes, the same chip is used in the 460M which runs at 675Mhz at a higher voltage (0.95v). My chip which is just a small tweak runs at 1v and 775Mhz.

 

ES=Engineering Sample. Basically early Intel test chips that made it to market. He is referring to an engineering sample version of the i7-2720QM.

I always felt that as long as temps were under control, overclock to your heart's content as long as you aren't going crazy with the voltage. Since the 555m in the Dells are voltage-locked, little issue there. The temp I'd be more concerned about is the memory temp versus the core. Your core temp is more than fine, the question is what is the memory doing? I've yet to find a way to read the memory temp in my system though...

Chances are you are more than fine however. I forget what the stock memory speed is on this card (was it 900?), but if you aren't going far over that, good to go is my bet. Overclocking GPU memory in general has proven to give minimal at best gains most of the time.

 

But the 460 has 192 cuda cores so the 555 can never be as quick even with overclocking due to less physical cores/shaders. But it does overclock near its level.

 

Thank you for such detailed information, it really helps. I will try to measure how many extra frames I got with oc'ed memory clock. If there will be no noticeable gain I suppose it's better to keep it at default 900 MHz.

 

Is 900 indeed the stock? Good memory on my part then (ooo, pardon that pun, SO SORRY!).

I've seen plenty of benchmarks before indicating minimal fps gain with memory overclock. If it is 1-2, I wouldn't bother with the OC.

 

Has anybody exerience with modding BIOS here?
Some guys with a M14x, which has exactly the same GT555M, have managed to mod their BIOS for a voltage increase, resulting in clockspeeds of 850+MHz, and 3DM11 scores of P1900.
[Mod] Voltage increase Nvidia GT 555M - Page 7
Maybe we could do the same for the XPS 17? :wink: :wink:
BTW p1900 score in 3DM11 surpasses the GTX560M with 100 points...

 

All I was saying is don't be worried you are somehow damaging the chip by ocing it, so long as the thermals are fine.

 

I don't have 3DMark11, but here is a link to the 3DMark06 results.

Intel HD Graphics 3000 Mobile video card benchmark result - Intel Core i7-2630QM Processor,Dell Inc. 0XN71K score: 14359 3DMarks


I DO NOT have the 3D system, it's optimus, and it's 1600x900, not FHD.

EDIT: Nevermind...I could have SWORN I had the 2720QM, it's the 2630, so that's the difference.

 

Looks like you're doing OK, I see you got it above 14K which was one of your original concerns. From the looks of your results with the 2630 you're not to far off of the 2720 levels.

Good to see you got it sorted out.

 

Yea any of the quads should be similar. I was thinking you might have a dual, which would possibly have that impact.

 

Today i OC my gt555 to 800 mhz with nvidiainspector and the program downcloced to 797,i think that is the maximum without bios mod,wich is good,and oc the memory to 960mhz and score 1824 on 3d mark 11.For me thats enough.I will be more interested of bios mod that unlock the memory speed,because kingston hyperx pnp 2133 comming soon,and that will be nice boost.

 

While temperature is definitely the main factor in OC, it's not the only one.

There is a hidden danger to overclocking that's much more difficult (practically impossible) to monitor and that's current.

As a person with an electronics background, I can't count all the times I've had to closely study data sheets for individual components (capacitors, resistors, etc.) when building my own simple circuit boards and what the current or voltage tolerance was at each gate. Increasing the current by just 5% on any gate could be reaching or going past a component's tolerance and greatly reducing its life (maybe even frying it).

Imagine how many transistors are in a GPU and the current and voltage increases at each gate when you increase clock speeds (we're talking micro amp possibly even smaller increases). The change at any one gate could greatly affect the life of a single component and if that component fails, the entire device fails.

OC'ing is definitely 'do at your own risk' but the folks on this thread seem to have fared well since at least March. Maybe when you get to 12 months with no issues, I'll OC mine hehe

 

Thanks for the insight, LeeTone.

This leads to another question (to all): in case one fries his card, will Dell be able to tell it was being OC'ed and therefore refuse to replace it? (I realize I asked about this a while ago, but apparently no one had any thoughts nor experience to share.)

 

I doubt the dell techs are even intelligent enough to know what overclocking is tbh, from my previous experiences with them. But even if they did i doubt they would be able to tell its been overclocked, as long as you delete the software.

 

It's 797mhz because the clock generator does not do 1mhz jumps. There is no point to have that fine control. Try raising it further and see it jump up again.

 

To give you an idea, the maximum current on my CPU is listed as 115A, yes it will affect the life but the likely hood that one of your other components will fail first is higher.

Virtually the same chip is running at 1v and 775mhz at stock in my machine, so your chip is undervolted and underclocked to begin with.

 

I hope that is a typo...115A cannot be right. That's more current than a typical house power feed from the road (100A is typical)...an electric oven only pulls 50A max..., yes I realize that is at a higher voltage but still.

 

I know it's crazy eh? But that's not a typo.

 

Looks like I've gotten about as close as I could come...did alot of system optimization, driver updates etc...so I was able to get this OC'd at 770/1000/1540

NVIDIA GeForce GT 555M video card benchmark result - Intel Core i7-2630QM Processor,Dell Inc. 0XN71K score: 14804 3DMarks

 

It's not the total load voltage and current you have to worry about though. You can monitor that like any other electrical device.

A GPU/CPU isn't one component, it has billions of components inside it. Even if 99% of the transitors inside a chip are within tollerence, that 1% failing will cause that entire device to fail. The problem is that there's no way of monitoring each transitor.

The fact that a device comes underclocked to begin with, may suggest that all the components inside it are well within their tollerance levels but you can't know for sure without looking at the datasheet.

 

Voltage is the main killer, life expectancy will mostly depend on it really if we are going to put our practical engineering hats on.

Second to that is temperature, beyond that it matters little.

 

Both voltage and current can kill a device.

P=IV

 

REMOVED COMMENTS

EDIT: After thinking about it for a few minutes I don't want my comments to LeeTone to cause this thread to turn sideways so I removed them, this thread is about our OC'ing experiences and Increased Performance Levels. Not sure why LeeTone a guy with 4 posts is trying to steer this thread in a different direction, maybe he should start a new thread about the dangers of overclocking instead.

It's my card and I'm gonna run it the way I want too and if I damage it doing so then that's my problem.

 

That's a pretty good performance increase from what you originally had, now you'll just have to keep an eye on temps and determine if it's stable at those settings.

 

True current spikes would kill it, but they are far more sensitive to voltage.

It's not the power dissapated that kills a chip, it's the switching process breaking down the structure and that is sped up by temperature and voltage.

True a higher current increases the power and temperature but thats why you watch temperatures, if you keep them down current has little effect.

 

@ Hatchett (a guy with rockin' 80 posts... )

Why are you getting all jumpy? No one tells you not to OC your card...

LeeTone is not the first to make comments about the risks of OCing in this thread. In fact, he joined an ongoing discussion about safe temperatures and has made some interesting points - helpful for those L702X/GT555M owners, who are still hesitating. Separate thread would certainly make more users benefit from this discussion, but I'm sure there are quite a few such threads around NBR forums and if this discussion wasn't taking place in this thread, I, for one, wouldn't have come across it.

 

Not jumpy at all & Points Taken.

 

So far so good. I've not hit over 72C without a Chill Pad. I've found that turning V-Sync off in my games helps TREMENDOUSLY. I thought it had something to do with lip flap LOL, apparently not.

My Street Fighter IV Benchmark hit like 160FPS with it off...talk about mexican jumping beans! LOL.

 

agreed - I play all my games with v-sync off, mainly because any frame rate drops are far more pronounced with v-sync enabled... there 's slight mouse lag and your max fps will only ever be the refresh rate of the monitor.

would rather deal with slight screen-tear

and yeah - would listen to anyone who raises concerns/questions regarding overclocking... you can only really go on your own experiences and whether or not you're willing to deal with the potential consequences. I've had an old radeon card in a desktop die after 2 years - (possibly prematurely) due to overclocking but then again I've been overclocking the 8600m GT (a card notorious for overheating/death) for nearly 4 years now without issue.

in the end it boils down to "most people will be fine" if they overclock a card and monitor temps etc... but there'll always be that minority who will be unlucky :-/

 

I've noticed similar temps unless I am really pushing it hard like with Crysis 2.

My idle temps are lower than stock also, are you guys seeing the same?

 

Yay for single digit post count!

Apologies if my comments seemed like they were re-directing the thread, that was not my intention. I'm definitely not telling you not to overclock your card either (excuse the double negative).

As I was reading through the thread, I felt that there was a great amount of re-assurance that if a user would carefully monitor their temps, their card would be fine. I thought this was possibly misleading as a user might just as likely end up frying their card because of the reasons I've already mentioned despite observing relatively low temperatures.

It's probably important to note (and I've seen this disscussed in this thread as well) that manufacturing defects and other factors such as environment can also affect the life of a card possibly giving a stock card (under such conditions) an even shorter life than an OC'd card.

My basic conclusion is that OC'ing can never be done without increasing risks. However, careful monitoring and mitigation can counter this, potentially to the point (if you went far enough) where there the card is at less risk while OC'd than a stock card that's not treated with such care.

 

Do i dreaming something or with ssd games are sooo much more fluid ???

 

I think that would depend on how much the game tries to read/write to your drive in the middle of gameplay.

 

That is definitly something I have noticed in the past, I mainly play CODs but yeah everything just runs much cleaner and smoother - also I always used to get LOOOONG loading times between matches, so long that the round had already started by the time the screen had loaded up - all gone when on the SSD usually first there now.

 

Look, what you are doing is taking a small factor, and blowing it out of proportion when comparing it to the large ones.

You can take a chip, put a larger cooler on it and overclock it 33%, if the cooler drops load temperatures 10-15C it will actually last longer than the stock chip under the stock cooler.

Plus can I remind you again this exact silicon is running in other laptops at higher voltages and clocks, so your point does not even apply to this chip at all.

 

Again, just because you're running cooler, doesn't guarantee your chip will last longer especially when OC'ing. Temperature is not the only thing that can cause a chip to fail. To tell someone that sufficient cooling will completely negate the effects of increased electrical load on the chip is irresponsible.

The same silicon? Yes. The same circuit design? Definitely not.

 

Since you brought it up, irresponsible is telling people that they will burn up their cards using very general examples of basic electronics principles instead of facts.

Sorry but using Ohm's Law to support your position on this subject just isn't getting it done, it's garbage. I'm not saying Ohm's Law is garbage, i'm saying how your applying it to what we're doing is.

The above statement is very true, but how have we increased current or voltage by changing clock speeds?

You have talked about increased voltage and current multiple times while, where is this increase of voltage current coming from? We cannot change the voltage supplied to the chip so it's not coming from there. Don't confuse Overclocking with Overvolting.

We are changing Clock Speeds (Frequency) (Cycles per second). Now you cannot say how changing a frequency affects a circuit using Ohms Law, the truth is you have to know the EXACT circuit to determine how an increase or decrease in Frequency affects it, in some circuits increasing Frequency decreases current flow and when we decrease current flow we generate heat.

We are not changing the amplitude of the clock pulse so this increase in voltage or current isn't coming from there either. So once again unless you know the exact circuit you saying we've increased voltage or current is garbage. My guess is that you don't know what the circuits were discussing consist of, just like I don't. None of us can make a claim that we are or aren't affecting voltage or current unless we know the circuit/circuits in question.

What we are doing is asking the GPU and Memory to run faster than the normally do by adjusting the clock. The extra speed results in more work being done by the GPU and Memory in a given time period, increasing the overall performance.

What we can state as fact is that we are generating more heat. We also know that heat can damage electronics. Heat is one thing we can monitor and control which is why you hear it being discussed here.

When you start to exceed the CPU's, GPU's or Memory clock limits from a increased Frequency standpoint (Overclocking) they either crash or give undesirable results like artifacts long before there's actual hardware damage. Thermal damage is more likely to damage our GPU's which is why you hear about temps being discussed.

We monitor temps because we are trying not to exceed the Thermal Design limits of the GPU, we can measure and monitor it and control it. One other point on heat is that modern chips have temp sensors and can now monitor their temps and slow or shut themselves off to protect themselves.

I'm done with this conversation, I went out and educated myself about what overclocking was and the benefits and risks, I suggest you and others go do the same before making the decision to OC or not.

Like I said before this thread wasn't about encouraging or discouraging anyone from doing anything, it was about those of us that have already made that decision to OC our cards and our results.

Carry on.

 

As someone with an electronic engineering degree I can tell you what Hatchett said is true.

You can say the average current consumption will go up, but really electron migration is not likely to be an issue with these chips so much.

 

Are you saying that it's possible in make something move faster without suppling more energy to do so? What do you think is making your clock run faster? What do you think is generating the additional heat? Think back to physics 101. In almost all CMOS circuits, the current is generally directly proportional to the clock speed.

Pass average? Didn't you build any PCBs during your program? Given your supposed background, how could you agree that it's possible to increase the amount of work without providing more energy?

Agreed. Back on topic.

 

So how many percent are people overclocking this card compared to default settings?

 

Kroenen on page 1 has increased the core clock up to 760MHz which is about a 29% increase.

It seems like many others on this thread are using that same clock speed.

It might be good to increase your clock speed in staged increments and observe the performance/temperature at each stage.

 

Yes, loading times are improved with SSD, that is one of their benefits. FPS will not increase however, don't believe it if you hear it.

Back to overclocking. I'm thinking about bumping mine soonish. I just don't use this system to game unless I am not home, I have a 560TI in my desktop for that

 

OK now you've pissed me off with this Gem, I was gonna let this die until I saw this and by the time this conversation is over your should offer Meaker a public apology. You'll probably owe me one too but don't bother since it won't mean jack to me coming from you.

Yes I'm saying that you can move something faster without applying more energy to it.

If I want to do more work then yes you need to increase the amount of energy, in our case we're not increasing the amount of work we are doing we're just getting our work done faster.

It's time (pulse per second aka frequency) that we're changing, not the total amount of energy or the amount of work to be done.

EXAMPLE:
We both push a 10 lb ball up a 20 degree slope, I take a break every 30 seconds because I'm an old fart and you don't cause you don't have to cause you're young. It takes the same amount of energy to get the ball up the slope regardless of how fast it happens.

Time, pulses per second, minutes, hours. aka Frequency Go back and take a look at pulse widths and their definition.

EXAMPLE:
The Job = work = filling up a 5 gallon bucket 1 quart at a time.
I put 2 quarts in the bucket every minute until it's full. You on the other hand put 4 quarts in the bucket every minute until it's full. The frequency that we put the quarts in the bucket was different but the total amount of energy required to put 20 quartz in a bucket was the same regardless of the speed at which the work was done.

Transistors dissipate heat when you turn them on and off, why because we're impeding current flow, when we impead current flow we generate heat (Think Resistors). If the transistor is on then current flows freely, as we shut it off we start to impead the current flow until we stop it (Well almost) since there's leakage current. When we turn it on and off more times per second then we generate more heat.

OK Lets.

Power = Work/Time

Work has nothing to do with the amount of time that this force acts to cause the displacement. Sometimes, the work is done very quickly and other times the work is done rather slowly.

EXAMPLE:
With our GPU's the task is to display 100 frames, we can display those frames at 30 frames a second or 60 frames a second, One will finish faster, either way the job (Total amount of work) to display 100 frames didn't change, neither did the amount of energy it took, only the rate at which the work was performed changed.


The statement below is where you're confused

Here's where the problem lies, we didn't increase the amount of work, we just changed the rate at which we do work.


EXAMPLE:

A game has 20 million frames to display (That's the work that needs to be done).

Now I display that 20 million frames at a frequency of 30 frames a second.

Then I increase the frequency (The number of frames per second) display it at 60 frames a second.

The total amount of work didn't change it was still to display 20 million frames and the amount of energy required to do said work didn't change either.

We just got our work done faster by turning transistors on/off faster (See examples above) and in the case of displaying graphics smoother and less jumpy.


@LeeTone before you go insulting someone publicly you may want to check yourself first, maybe you should do a little brush up on electronics and physics before feeding your ego.


My apologies to the readers of this thread for responding to LeeTone, I just couldn't let this slide when he turned it personal by attacking and insulting Meaker and he's just plain wrong to boot.


/Rant

 

Hatchett, im no expert but what you say does make sense compared to what leetone is saying. And i think people need to know the pros and cons of overclocking so these discussions are good, but not when it starts getting personal.