I'm starting to develop a Hatred for Nvidia GPUs

Do not feel awkward about it, NVIDIA makes it extremely easily to dislike (or even hate) them.

Here's a funny blog post of one of the Valve graphics developers (he worked at Portal 2 and Dota 2) who calls NVIDIA greedy fucking bastards (of course not literally, but in corporate speak, the idea is the same):

http://richg42.blogspot.ro/2014/05/the-truth-on-opengl-driver-quality.html

A is NVIDIA, B is Intel, C is AMD.

EDIT: I might be wrong, as vmN has kindly pointed out. B might be AMD and C Intel.

I think we've all largely agreed on that. As Morgoth said, we can only criticise the relative price/performance.

I'm pretty sure vendor C is Intel.

Really? What makes you say that? Because I honestly thought it's AMD because they have the two different drivers (the open source one and Catalyst) and because of this: "Anyhow, driver #1 is unfortunately pretty far behind on the GL standard, but maybe in 1-2 years they'll catch up and implement the spec as of last year". I know that the open source driver is at OpenGL 3 and it's working its way towards OpenGL 4 by implementing a feature here and there, people are estimating it will get there in a year or two.

And this: "They are the leaders in the open source graphics driver space, and their hardware specs are almost completely public". Hm... this might apply to Intel too, I know they published lots and lots of specs several months ago.

Although this :"These folks actually have so much money [..]" does make me think twice.

EDIT: am curious, why do you say it's Intel?

"They don't really want to do graphics"

Intel really don't have the great interest in graphics, however they are "forced" into getting better integrated graphics performance.

 

it's really just a distraction from their historically core business, but the trend is to integrate everything onto one die and they have plenty of die space to spare

AMD really doesn't have that much diespace to play around with. That is the major reason why AMDs APU doesn't feature L3 cache.

 

Vendor B sound much more like AMD than the other too.

Ok, I see your point.

Seeing as I have done a bit of overclocking with nvidia cards ill put my 2 cents in.

Now the GTX 770 as said is an overclocked 680. Its already runing beyond its optimal performance and overclocking it is hard. A lot of voltage is needed to push 770's and you can only get that with a modified bios and of cause that lowers its lifetime.

The thermal limit, Basicly it is in most cases pointless unless your doing extreme overclocking or your an idiot and took off the cooler.Its a safety pricortion and if your running your card stock, you are unlikely to hit the limit.

Power limit is another semi useless feature as even with max stock voltage and pushing the card as hard as you can, you are unlikely to see the limit. I cant even get 100% with stock bios, the 115% is useless at stock. Just another safeguard if something goes wrong.

GPU boost 2.0 also has a limit. pre defined for safe overclocking of the card but in reality is designed for longevity of the card so to underclock when the card is not needed.

All those limits can be bypassed with voltage unlockers and bios flashing. Once those limits are gone you get new limits

VRM throtling. This is where you are physical drawing 100% of the ampage that the VRM can provide, You can sort that with higher voltage which brings down the ampage but the total power consumption stays the same. When you hit the VRM limit, Thats when you know you are drawing max power for that voltage that the safe guards hold, Now increasing the voltage actually relieves the stress on the VRM as converting 12v to 1.1v is easier than down to 1v. It dosent reduce heat that much however so you still have to be careful. 

Because the GPU core needs so much wattage to function at different clocks, When it is not receiving enough wattage, the core cant process the data correctly and you gets artifacts.

Now there is another physical limit and thats the core. Silicon is a semiconductor, Basicly in pure state, Much like water. Very little electricity can flow through the silicon. When you have impure silicon, it can conduct electricity a lot easier. thats how a silicon chip works, Traces of impure silicon in a silicon wafer. However the physical limitation is like many things, If you heat it up enough, it will conduct electricity. And silicon starts conducting electricity at between 110 and 130 degrees if I recall correctly. So when that happens the whole chip shorts and the physical chip is no longer as it was. That is why heat is so important but remember the temperature readouts you get are not the same as the temperatures inside the chip and so your chip might be running at 95 on the surface but it could be hitting 100 inside or even more if the chip is poor.

Now heat is caused by the resistance in the chip, without resistance, No work would be done so its a fital part of a chip. This is why not all chips are the same. Resistance effects the overclocking potential of a chip, more resistance leads to higher required voltage and in turn higher temps. and just like pure silicon, the impure silicon has reduced resistance at higher temps. Thats where a balance of temperature and voltage comes in. Too little voltage and the circuits can not be complete, Too much voltage and the circuits will short and dead chip. So thats why extreme overclocking is such an art as CPU's and GPU's are overclocked so much at factory that they put these limitations in as its so easy to get the balance wrong.

Now going back to the 770, It still has potential but to get where they are now, They are binned extensively. The classy is binned even higher and so the fact that people can still overclock many of them is great but dont expect too much from those chips without better the air cooling and once you go there, it starts to become harder the balance resistance with temperature and voltage. With the newer GTX 780's. The core in them is more efficient when it comes to resistance and so they can push those cards harder. They are also relatively new compared to the 770's and so have a lot more headroom. Thats why the 770's dont overclock as well as 780's and people tend to get disappointed. 

Now GPU boost goes by past dater of benchmarks before tests so honestly not worth bothering with.

The thermal and wattage limits you do need to look out for under extreme overclocking. I have hit 140% power consumption and 99/100 degrees on the core. But that was running with 100% physical limit of power draw from the PCIE Bus and PCIE power connectors. 

Anyway I would say if you are looking for more performance from your 770. Take a look at this forum

http://kingpincooling.com/forum/forumdisplay.php?f=33

The guy K|ngp|n has the EVGA GTX 780 TI K|NGP|N dedicated to him asnd not only do they do basic overclocking on air, They do extreme overclocking. They have managed 2100 on a GTX 680 Classified if you feel up to it take a look. There is a lot of reading up to do and they do a lot of soldering to get higher voltages than the VRM allows but their good guys and they will help you out.

Ill be honest. The overclocking community on this forum is nothing. I dont even think Logan would touch overvolting to 1.35 even though its perfectly safe with most cards if you play it smart but then. Dont expect to hold those clocks without risking lonmg term damage to your card.

You can disable the boost function on nvidia GPU's using a bios mod.

The voltage limits can also be bypassed for the cards with a decent voltage controller, and for the ones that are resistor controlled, you have to look up the datasheet on the voltage controller, then find the resistance needed to boost the voltage.