NVIDIA's shady trick to boost the GeForce 9600GT

[wittekakker]

FIY, if you had your PCI-Express frequency defined as 110mhz and you wanted to run GPU at 720mhz, it should have been running at 720x1,10= 792mhz not 780... What programs did you use for readings?

Man I'm so excited with this that I really want a 9600GT NOW!

Do you realize with extreme cooling/voltage we could easily achieve past 1GHZ frequency?

THIS IS A MAJOR TWEAK!

Tweak? This is a overclock, nothing more, nothing less. 1Ghz is less great when the GPU seems to be less efficient then expected. K, you got the 1GHz screenshot, but performance wise it doesn't change a thing.
I used Rivatuner 2.06 and GPU-Z 0.1.5.

I've not heard of a card that uses a 25MHz multiplier anyway. They all use 27/54MHz multiples.

Crystal /= multiplier I've seen 29MHz too, but like only once until now. Custum boards are also affected.

[STEvil]

When looking at the PCB, the only crystal we could find is a 27 MHz one, so where is the magical 25 MHz coming from ?

As many of you know, the PCI-Express bus clock frequency which connects the card to the rest of the system is running at 100 MHz by default. What NVIDIA did is to take this frequency and divide it by four to get their reference frequency for the core clock.

On systems where the PCI-E bus runs at 100 MHz, this results in the 25 MHz used for the default clock indeed. But once the PCI-E clock is increased beyond that, the GPU will magically run at higher clock speeds, without any changes to the graphics configuration, BIOS, driver or any software.

On "normal" VGA cards, when you increase the PCI-Express bus frequency you increase the theoretical bandwidth available between card and the rest of the system, but do not affect the speed the card is running at. On the GeForce 9600 GT, a 10% increase in PCI-Express frequency will make the card's core clock run 10% faster!

I guess it's all true guys. I have been running 9600GT's clocked at 720MHz out of the box, but the card never remained stable in my tests. I got in another sample, same happened. Now, due to rushing things, I have left my sources wondering what the problem could be, but now it all seems to come down to the simple fact where my PCIe bus speed could have been the problem all the time. I have it overclocked to 110MHz all the time which makes the GPU run at 780MHz instead of 720MHz, downclocking the GPU solved it so I guess PCIe speed did affect card clocks. Makes sense, I really don't know why else my three samples couldn't clock further then 710MHz with 110MHz PCIe speed.


I have an Intel X38 chipset. Those wondering if it will work on any chipset, please spend some time reading the article. It counts only 4 pages and you will not have to spend time posting 10 times here on XS in order to find out. It is based on PCIe speed, nothing to due with Linkboost instead that Linkboost will now overclock your VGA GPU too further increasing the total system performance.

Quoted for those who miss it.

[grimREEFER]

who does this really effect? ppl who want to overclock their pci express bus and not worry about changing their graphics cards frequency's?
i dont understand the con's

[Luka_Aveiro]

If you still need extreme cooling/voltage to achieve 1Ghz+ frequency, why is pci-e based overclocking a "MAJOR TWEAK"?
On some mobos, increasing the pci-e frequency even causes the hard drives to malfunction.

Vai spammar para outro lado pah, alguém te perguntou as horas?

It's a major tweak because it increases the clock generator frequency, similar to FSB incrasing when overclocking CPUs.

On nVidia chispets, afaik, PCI-Express frequency does not affect in any way hard-drives malfunction, as NB<->SB is made by HT Link.

[Eastcoasthandle]

who does this really effect? ppl who want to overclock their pci express bus and not worry about changing their graphics cards frequency's?
i dont understand the con's

The execution of this from NVIDIA's side is less than poor in my opinion. They did not communicate this new feature to reviewers at all, nor invented a marketing name for it and branded it as a feature that their competitors do not have.
Even when asked directly we got a bogus reply: "the crystal frequency is...". No, there is no 25 MHz crystal and its frequency is not fixed either. I'm not accusing the sender of the E-Mail of course, I just believe he didn't know, maybe this fact wasn't communicated to the marketing team at all. However, if you would get such an inquiry wouldn't you look into this further if it was your job to properly promote a product?

source

[Luka_Aveiro]

Guys, the fact is rivatunner is buggy, and it does not read your final clock by increased PCI-Express frequency, appears it always multiplies the GPU internal multiplier by 27mhz, as it might refer to the physical existing clock generator (memory ones) according to techpowerup.

[Zorlac]

I hope the GPU-Z author is willing to update the utility to show driver reported clocks and actual clocks.

[largon]

That always happends, and is because clock speeds arent linearly selectable, they have jumps, if this is all the basis of TPU review means they havent messed with cards in a long time.

The fillrate measurements W1zzard made are conclusive:
PCIe link OC-&#37; = G94 GPU OC-%

[Luka_Aveiro]

FYI, a friend of mine has a 9600GT and I am asking him right now to do some tests regarding the confirmation of this theory, please wait

[Cybercat]

who does this really effect? ppl who want to overclock their pci express bus and not worry about changing their graphics cards frequency's?
i dont understand the con's

I don't understand either. Apparently NVIDIA's the big bad wolf for implementing this.

People should be glad it can be done on chipsets other than NVIDIA's. If NVIDIA blocked it off to only 590i and 680i chipsets, then you can say it's a little shady.

[Tuvok-LuR-]

I wish I had a 9600GT in my hands to confirm this, it just seems too good to be true

I missed the good part

[WeStSiDePLaYa]

I think something is wrong with TPU's bench system.

They are the only people who have this "problem".

Also, linkboost support is now gone since 590sli.

Also, since when did linkboost increase the card clocks?

Linkboost never increased the card clocks, only the pci-e and chipset buses.

Also, since when is the core clock dependent on pci-e bus?

I don't think TPU knows much,. and they are just spreading misinformation.

[destr0yer]

Hello

I have a e-VGA 9600 GT and this "feature" confirms!

1&#186;, rivatuner report:


this clocks is equal for PCI-E @ 100 mhz or 110 mhz, but the 3dmark 2006 fill rate multi texturing is not equal!

PCI-Express=100mhz
1-Results at 675mhz GPU: 16349
2-Results at 743mhz GPU: 18284

PCI-Express=110mhz
3-Results at 675mhz GPU: 17948

at 743 mhz, rivatuner report 800 mhz core. Its REAL clock or bug?

if true, its amazing tweak at 125 mhz the core clock is pushed to 844 mhz! Interesting, Graphics OC "driveless"

but, higher PCI-E frequencies is dangerous to the hard disks?

[WeStSiDePLaYa]

Also, does this "conspiracy theory" even make sense?

Why would nvidia want to show lower clocks? It does not make sense. There is no reason for it.

And raising the pci-e frequency has given me a performance boost on ALL my cards from the 7900GS up. But it does not change the core clock.

[mascaras]

And raising the pci-e frequency has given me a performance boost on ALL my cards from the 7900GS up. But it does not change the core clock.

yes , but only performance in Max OC of Card , not in bench scores or games with "default" card clocks .. (like 9600GT )

regards

[Eastcoasthandle]

Also, does this "conspiracy theory" even make sense?

Why would nvidia want to show lower clocks? It does not make sense. There is no reason for it.

And raising the pci-e frequency has given me a performance boost on ALL my cards from the 7900GS up. But it does not change the core clock.

It makes perfect sense. You should never rely on your PCIe frequency as a de facto to OC your GPU as going to high (on some MBs) can effect stability (in particular your HDD). The PCIe frequency was never designed to OC the GPU.

[Frank M]

I don't understand either. Apparently NVIDIA's the big bad wolf for implementing this.

Think about reviews then.
Without the reviewers and the readers knowing it, they are comparing an
overclocked 9600GT with the other cards, which are still at stock speeds.
It's easy to perform well that way. In the sports world, this would be called
doping.

[Luka_Aveiro]

Also, does this "conspiracy theory" even make sense?

Why would nvidia want to show lower clocks? It does not make sense. There is no reason for it.

Man, you all read too fast

This is the deal:
Core clock is affected by PCI-Express frequency. Why?
Because there's no crystal clock generator chip in 9600GT's.
So how is the clock generated? By using a quarter of PCI Express frequency.

Do you got it 'till here? Good.

Now, what if I increase PCI-Express frequency??? What's the real life consequence?
You raise your core clock. If you increase PCI Express Frequency to 110mhz, you will bump your core clock by 10&#37;.

As my friend destroyer (ol&#225;, meu ) has shown, the results speak for themselves...

Apparently shader clock is not affected by this overclocking method, as 743mhz core achieved by PCI-Express frequency is slower than 743mhz manually overclocking.

Once AGAIN, linkboost has nothing to do with this, it olny has to do that it could have raisen core clocks a bit higher than a regular mobo.

[WeStSiDePLaYa]

It makes perfect sense. You should never rely on your PCIe frequency as a de facto to OC your GPU as going to high (on some MBs) can effect stability (in particular your HDD). The PCIe frequency was never designed to OC the GPU.

I don't think you understood what I was saying.

[Lightman]

I think something is wrong with TPU's bench system.

They are the only people who have this "problem".

Also, linkboost support is now gone since 590sli.

Also, since when did linkboost increase the card clocks?

Linkboost never increased the card clocks, only the pci-e and chipset buses.

Also, since when is the core clock dependent on pci-e bus?

I don't think TPU knows much,. and they are just spreading misinformation.

Did you know that author of that article is also a creator of GPU-Z, Atitool, SPDtool, SysTool, etc??
I think he knows quite a bit about GPUs and chipsets

Lets wait till we can confirm this by independent sources. Validation should be easy, just test FillRates at different PCIe freq...

[kryptobs2000]

If this were a 'trick' then I don't see how it really benefits nvidia? All the same, it seems like 90&#37; of the people posting here didn't actually read why it does this. I'm not going to bother retyping, or requoting, so many of you seem to just be ignoring the numerous other explanations so w/e.

My question is, now that we know why this happens, to those claiming this is a benefit, how so? It's not like you can't overclock the cards otherwise. You were never limited because of the crystals clock before. This makes no difference. If your card can reach 800mhz it doesn't matter how you achieve that clock, 800mhz is 800mhz (for videocards anyways). Am I wrong?

[hollo]

destr0yer's results add up the same:

25mhz x 27 = 675mhz in rivatuner overclocking (correct clock)
27mhz x 27 = 729mhz in rivatuner monitor (needs a program update)

and 3dmark fill-rate increases (nearly linearly) with PCI-e frequency.

PCI-e 100mhz, 3dmark fillrate = 16349
PCI-e 110mhz, 3dmark fillrate = 17948 vs 16349 x 1.1 = 17984

[grimREEFER]

Think about reviews then.
Without the reviewers and the readers knowing it, they are comparing an
overclocked 9600GT with the other cards, which are still at stock speeds.
It's easy to perform well that way. In the sports world, this would be called
doping.

here's the part i don't understand, dont reviewers leave the pci express frequency at stock?
this seems like it was just an innocent attemp by nvidia to get motherboards to do hardware overclocking of graphics cards without relying on the drivers, thus adding another tool in the already bloated overclocker's toolbox.

[mascaras]

Lets wait till we can confirm this by independent sources. Validation should be easy, just test FillRates at different PCIe freq...

confirmed !!


PCI-Express=100mhz
Results at 675mhz GPU: 16349


PCI-Express=110mhz
Results at 675mhz GPU: 17948




w1zzard from techpowerUp its correct

[destr0yer]

Man, you all read too fast

As my friend destroyer (olá, meu ) has shown, the results speak for themselves...

Apparently shader clock is not affected by this overclocking method, as 743mhz core achieved by PCI-Express frequency is slower than 743mhz manually overclocking.

Once AGAIN, linkboost has nothing to do with this, it olny has to do that it could have raisen core clocks a bit higher than a regular mobo.

(Olhó o luka, como estás )

Testing @ 743 mhz, with Shader Clock defaut (1675), unlinked mode, the results are equivalent at the test 100% defaut. The Shader clock is already raised with PCI-E @ 110 mhz