NVIDIA's shady trick to boost the GeForce 9600GT

[ghost101]

If marketed correctly, this could boost nvidia's motherboard sales. Since linkboost is a feature of the motherboard, you cannot call this a lock in and have the usual anti-trust allegations.

However, not telling reviewers was clearly a mistake.

Also, it isnt unfair in terms of comparing the 9600gt to ati cards. Effectively, the stock frequency is higher than what people originally thought. But people on intel/amd motherboards deserve to know this.

[Jakko]

Nvidia plays dirty again.

And no, this is not just an undocumented feature.
If ati made drivers that overclock cards but make it appear as if stock speeds are used it's foul play as well.
The question is not whether or not some gpu's can handle the overclock (obviously ati cards could handle it as well), neither is this about nvidia making it easier to overclock gpu's (it has always been easy).

This is about ing up reviews and benchmarks, just like nvidia has done before. (Remember the early Crisis drivers that set your Crisis detail settings to low automaticly?)

No nvidia for me.

[hollo]

And leads to people like me being up to midnight on a work night bashing their heads into the keyboard thinking they got horribly binned cards cause they didnt know this little hook.

haha i feel for you man

nvidia would have only had to attach a little note saying "PCI-e clock and GPU-Core clock are linked" to all the samples they sent to review sites to have stopped this from happening.

oh well
none of these companies are very good at supporting tweakers and overclockers, except maybe intel. AMD won't even release the AM2 pinout

[G80]

Gosh people, calm down.

[audiofreak]

So... does this mean that the base clock for those cards is PCI-Express bus frequency / 4?

Looks like just another way to save cents on BOM -- one crystal less on the board itself.

From that perspective, NVIDIA did the good thing because cards can be sold cheaper.

Unfortunately, the problem is they forgot to tell us so those people who overclocked their PCI-E bus will end up overclocking the GPU as well.

What I don't understand is this -- if reviewers tested 9600GT overclocking potential which method they have used to raise the GPU clock?

I mean if RivaTuner does that directly via PLL, then there is a chance they haven't actually managed to change the GPU clock at all. That would make all overclocking results invalid, right?

[op1e]

Cannot run stable at 110 pci-E even with stock clocks. 675 core 1050 vram and 105 pci-E gives best result for me.

[i found nemo]

And the hardware monitor in rivatuner, your readings are...?



This is the basis of TechPowerUP theory, it could just be a reading error, afaik.

....

[op1e]

The discrepancy in clocks is shown with or without increase in bus speed. What bothers me is the actual clocks are not shown, that I know of. 675 reads 734, I think, and is not stable with 110 set in bios. It is stable at 105, however.

[jimmyz]

I guess the part that bothers me is where will this stop. can amd now launch processors they claim run at 3ghz because they decided to change the math? there are organizations such as jedec that are supposed to govern these things.

[Luka_Aveiro]

....

Damn it man? have you read the whole thread or are you just nit-picking?

Here's your answer to that, it's a RIVATUNER BUG, READ IT.

Thank you.

I can't believe this is xtremesystems forums, all I see is noobs.

[v_rr]

Damn it man? have you read the whole thread or are you just nit-picking?

Here's your answer to that, it's a RIVATUNER BUG, READ IT.

Thank you.

I can't believe this is xtremesystems forums, all I see is noobs.

It´s not rivatunner bug.
Actualy is right the article in the first post.

[zakelwe]

I cannot see LINKBOOST in my Asus 780i BIOS nor is it in the evga 780i manual. Is this fully automatic then? If it used to be a BIOS option it is no longer there.

I would suggest that Linkboost set to disabled and PCI-e set to 100Mhz would be default BIOS options for most motherboard BIOS so I am not convinced how many reviews have been corrupted by this. It seems more like someone is excited to find this and other people are excited to beat nvidia with it

It should have been documented though as it can lead to instabilities if how the system works has changed.

Regards

Andy

[b1lk1]

I hardly see how this is nearly such a big deal that is being made of it. So what if it is clocked in any way that secretly makes it faster. I fail to understand the problem. After all, AMD's Ghz and Intel's GHz have always given different results.

I would be upset if they actually found a way to make 128 shaders appear as 64 shaders. That is what I have suspected about these cards. But so far, there has been plenty of rational explanations as to why it is so much faster with 1/2 the shaders. People need to be looking at this aspect, not a silly clock speed difference.

[ZoLKoRn]

On my default run ASUS 9600GT have this effect and this GPU clock show on everest from 120MHz pci-e by Striker II Formula

[Luka_Aveiro]

It´s not rivatunner bug.
Actualy is right the article in the first post.

OMG LOLOLOLOL

The article says rivatuners bugs because it reads the final clock by using the gpu multiplayer x 27mhz ALWAYS, as you only find 27mhz crystal clock on the PCB, BUT this is the memory clock generator.

As I managed to test this bug with destr0yer, he's readings at stock clocks (675mhz) were 729mzh with rivatuner monitoring. Do you know why? Because:

675mhz:25mhz=27; multiplier27x27mhz= 729mhz

Then, I asked him to raise core clock 25mhz, so the multiplier would increase by 1. Like, 700mhz:25mhz=28
So, multiplier28x27mhz=756mhz, and he confirmed that happened, rivatuner was now reading 756mhz with hw monitoring.

Did you understand it or should I make a draw?

God damn it man, read the damn thread!

[highoctane]

It´s not rivatunner bug.
Actualy is right the article in the first post.

From the article:

"Please also note that RivaTuner's monitoring clock reading is wrong. It uses 27 MHz for its calculation which is incorrect. When the PCI-E bus is 100 MHz, the core clock is indeed 650 MHz on the reference design. A RivaTuner update is necessary to reflect GPU clock changes cause by PCI-E clock properly though."

[wittekakker]

However, not telling reviewers was clearly a mistake.

Board partners don't even seem to know this, or are they also playing the same game?

I can't believe this is xtremesystems forums, all I see is noobs.

Yeah, disappointing, even with all the proof in front of their eyes they keep on telling its otherwise. Like... GET A CARD AND SEE FOR YOURSELF!

[mascaras]

The discrepancy in clocks is shown with or without increase in bus speed. What bothers me is the actual clocks are not shown, that I know of. 675 reads 734, I think, and is not stable with 110 set in bios. It is stable at 105, however.

its stable with 105mhz because with 110mhz the clocks are higher



675Mhz with PCI-E 110mhz , real clock = 742,5Mhz

675Mhz with PCI-E 105Mhz , real clock = 708Mhz


regards

[STaRGaZeR]

I can't believe this is xtremesystems forums, all I see is noobs.

Because this is a thread for noobs and fanboys, you know. In one hand, NVIDIA fanboys playing their game. In the other hand... mixture

[v_rr]

OMG LOLOLOLOL

Did you understand it or should I make a draw?

God damn it man, read the damn thread!

Insulting other users like that you hardly pass or point of view and no one take you serious

And yes the article is right. If you chage the PCI-E frequencie it will overclock the card:

675Mhz with PCI-E 110mhz , real clock = 742,5Mhz

675Mhz with PCI-E 105Mhz , real clock = 708Mhz

This leads to various problems. Instability, not advertized by Nvidia this thing and mostly is starting to appear some problems with the cards in some users.

[metro.cl]

I cannot see LINKBOOST in my Asus 780i BIOS nor is it in the evga 780i manual. Is this fully automatic then? If it used to be a BIOS option it is no longer there.

I would suggest that Linkboost set to disabled and PCI-e set to 100Mhz would be default BIOS options for most motherboard BIOS so I am not convinced how many reviews have been corrupted by this. It seems more like someone is excited to find this and other people are excited to beat nvidia with it

It should have been documented though as it can lead to instabilities if how the system works has changed.

Regards

Andy

NF780i shouldnt have linkboost on, since it uses a bridge and the actual PCI-Express works a lot higher to give more bandwith.

On NF680i at least EVGA the Linkboost is disabled at default, this has been for a long time, since bios 17 or 18, now they are in bios 31+.

[WeStSiDePLaYa]

I cannot see LINKBOOST in my Asus 780i BIOS nor is it in the evga 780i manual. Is this fully automatic then? If it used to be a BIOS option it is no longer there.

I would suggest that Linkboost set to disabled and PCI-e set to 100Mhz would be default BIOS options for most motherboard BIOS so I am not convinced how many reviews have been corrupted by this. It seems more like someone is excited to find this and other people are excited to beat nvidia with it

It should have been documented though as it can lead to instabilities if how the system works has changed.

Regards

Andy

Nvidia pulled linkboost months ago.

There is no more linkboost.

[zakelwe]

Then if no linkboost and default PCI-e clocks are 100Mhz by default then why are a few people claiming nvidia are cheating in reviews? Did they tell reviews to put the PCI-e clock up?

It just seems to me that a few people want to nvidia because for some reason they do not like them, perhaps due to their actions in the past.

And Luka_Aveiro, do us all a favour and shut the feck up. All you've done on this thread is demeened other people because they don't seem to share the same viewpoint as yourself.

Regards

Andy

[GAR]

780i's default pci-e clock is 125mhz when using a pci-e 2.0 card....thats why in the bios you only have pci-e_3

[purecain]

the pci clock frequency becomes an issue when it is tweaked to one sides advantage. if this adnvantage can be reproduced for both sides, then a true like for like comparison can be made as to which is the superior product...

IN THIS CONTEXT ONLY , having gpu boosting bios settings for one card is unfair.

From a chipset performance standpoint this is good news, as it shows that all areas of the chipset that can be improved have been....

i've always kept my pci clock at 110-120 anyway as we all know it provides some performance....