Yet another sneaky Nvidia trick

[rgallant]

Originally posted by: taltamir

Originally posted by: SniperDaws

Originally posted by: jaredpace
nah, he went into his bios with the card clocked at stock speeds. He went to PCI-E frequency and changed it from 100 to 110. booted windows and ran 3dmark06.

Then he went back to bios and set the PCI-E back to 100mhz (which is default) ran 3dmark again, and got a lower score.

Correct


but i do have an auto setting for my PCIe would this make any diffrence.

i tried 110Hz with my card at 740/1850/975 and it locked up halfway through return to proxycon or whatever its called.

Going back to topic. This IS manually overclocked. He manually overclocked it on a P35 and it was unstable, and then he said the new drivers auto overclock it on a P35 mobo. It's not automatic if he goes into the bios, changes settings, and experiences crashes.

So it seems that it does not happen with a P35 motherboard. Only with an nvidia motherboard.

So an automatic overclock only when combining an nvidia mobo + video card that does not happen with intel mobo, and in fact, cannot be archived (Stably) with an intel mobo. And, which nvidia adventised as "link boost technology". [I conclude that this isn't a shady trick, but rather, a case where a feature was not taken into account by some reviewers.]

-What would have happen to Intel, if they added some auto overclocking to their chips ,say back in the P-4 days, so when benched ,up to the AMD's 64, they would looked better ? and those overclock's did not show during benchtesting .
-if you have the PCI-E locked at 100 it should stay that way. my two cents.
-It's more than a trick , if you use it to sell cards.
- when they come out with the overclocked cards will they turn that feature off, giving more head room to sell the cards at a higher price?

 

[apoppin]

Originally posted by: rgallant
-What would have happen to Intel, if they added some auto overclocking to their chips ,say back in the P-4 days, so when benched ,up to the AMD's 64, they would looked better ? and those overclock's did not show during benchtesting .
-if you have the PCI-E locked at 100 it should stay that way. my two cents.
-It's more than a trick , if you use it to sell cards.
- when they come out with the overclocked cards will they turn that feature off, giving more head room to sell the cards at a higher price?

Probably 'nothing' IF intel's "P4 Boost" was listed right on their web site as nvidia's is today
--AMD would [will] probably copy it. AND intel p4 customers back then - like 9600GT owners today - are jumping for joy that their midrange card can beat AMD's 3870 with a Link-Boost O/C.

No one gives a crap HOW the results are achieved ... nvidia did not make a pact with the Devil for "magic" O/C'ing
:roll:


... no more than AMD did :evil:

 

[Janooo]

Originally posted by: apoppin
...
i *only* reported that Link Boost is not new and is clearly for O/C'ing the GFX card that resides on that PCIe bus. No rescue needed. And, although all the nvidia fans call me a "AMD fanboy", i got attacked by AMD fans. i find that a little fickle. i haven't changed. i call it as it see it and when i am wrong i have always been glad to admit it.
...

The Link Boost did not overclock GPU core. It's for PCIe only. You don't have any prove about GFX overclocking.
You say you'd be glad to admit when you are wrong. So, please, prove the 7900GTX overclock or admit that you are wrong.

 

[BFG10K]

Wrong

i have some interesting news for you ...

Link Boost has been around for awhile .. since '06

Yes, it has been around for a while but again it has nothing to do with changing GPU clock speeds, only PCIe.

oK , if it is not for Overclocking the GPU, what is it for?

It's for raising the PCIe clock speed as was explained earlier. As for the GPU requirement, those are probably the ones nVidia has tested with an overclocked PCIe bus but again that doesn?t mean GPU clocks are ever raised.

AFAIK the 9600 GT is the first card to do this and also the first card to report incorrect frequencies through the driver.

there is a lot that is not well-publicized; doesn't mean it isn't true because reviewers don't know - i.e. some nForce chipsets ran the AGP bus at 16x when a nvidia GF4 Ti was in the system

You're not answering the question. The question was can you, please, point me to an article that explains how 7900GTX overclocks itself when paired with LinkBoost-enabled mobo? All I can find is that PCIe is overclocked but not video card.

Your answer was that AGP was raised which again has nothing to do with raising GPU clocks.

Again overclocking PCIe and overclocking the GPU clocks are two different things and again AFAIK the 9600 GT is the first card that does both which is why it?s significant. In particular it's even more significant as it's yet another piece of the puzzle why it runs so well compared to the 8800 GT despite having much less SPs.

Again if this is a known feature why isn?t nVidia telling us abut the GPU overclocking itself? And why is the driver masking the overclock by reporting incorrect clock speeds?

 

[apoppin]

Originally posted by: Janooo

Originally posted by: apoppin
...
i *only* reported that Link Boost is not new and is clearly for O/C'ing the GFX card that resides on that PCIe bus. No rescue needed. And, although all the nvidia fans call me a "AMD fanboy", i got attacked by AMD fans. i find that a little fickle. i haven't changed. i call it as it see it and when i am wrong i have always been glad to admit it.
...

The Link Boost did not overclock GPU core. It's for PCIe only. You don't have any prove about GFX overclocking.
You say you'd be glad to admit when you are wrong. So, please, prove the 7900GTX overclock or admit that you are wrong.

We are discussing the 9600GT OC - you seem to be hung up on the fact that Link Boost was mentioned as being for O/C'ing the PCIe bus the old 7900 resided on. IF you cannot make the connection that perhaps O/C'ing the PCIe would would ALSO o/c the GPU on that bus, you need to do a little research. it is very basic.

i do not have a 7900; i never had a link-boost enabled MB ... but i have reports there is nothing unusual about Link Boost increasing GFX card performance on nForce 590 and above MB for G71. *Evidently* it just wasn't initially supported on G80, until G94 and 9600GT arrived. i don't have anything in my current bookmarks.

And i will be very glad to admit i am wrong ... when i am wrong

===========

oK , if it is not for Overclocking the GPU, what is it for?

It's for raising the PCIe clock speed as was explained earlier. As for the GPU requirement, those are probably the ones nVidia has tested with an overclocked PCIe bus but again that doesn?t mean GPU clocks are ever raised.

AFAIK the 9600 GT is the first card to do this and also the first card to report incorrect frequencies through the driver.

OK .. i don't see a problem ... we KNOW the basics that O/C'in the PCIe bus also O/C's the GPU ... but not like Link Boost ,,, agreed! ,,, it is new to G94 that they apparently expanded on its capabilities .. the 'proof of the pudding is in the eating of it'

it is "inaccurate" because it is new ..
there will be tools to report it correctly even if RT doesn't

 

[taltamir]

you are aware of the fact that the only thing differentiating a 1000$ and a 200$ intel CPU is the fact that intel UNDERCLOCKS and LOCKS the cheaper CPU.
If intel gave you a "free overclock" then they are basically lowering their price across the board.

 

[apoppin]

Originally posted by: taltamir
you are aware of the fact that the only thing differentiating a 1000$ and a 200$ intel CPU is the fact that intel UNDERCLOCKS and LOCKS the cheaper CPU.
If intel gave you a "free overclock" then they are basically lowering their price across the board.

If i were like *some* of you guys, i'd be bitching and complaining that my stupid little $114 e4300 1.8Ghz CPU at 3.25GHz IS faster then AMD 's best, right now
:Q



or would i be complaining?

i am not complaining now for that IS the *reality*

You guys are missing the BASICS .. my low-end intel CPU IS completely UNlocked from intel as usual ... and it DOES keep up with any of intel's stock CPUs ... i can probably get 3.5Ghz, if i want to add voltage to the core

i'd love to get my hands on a Link Boost board for GX2

 

[BFG10K]

but not like Link Boost ,,, agreed!

Yes, that's the point. This isn't link-boost, this is something much more.

It?s undocumented GPU overclocking while the driver apparently tries to hide it by reporting incorrect clocks.

there will be tools to report it correctly even if RT doesn't

Huh? It's because of RivaTuner that we found this out as it can read the GPU clock directly. Anything reading the driver clock will have incorrect frequencies.

 

[bryanW1995]

luke, you are my son, come to the dark side...(cue cheesy music)

 

[apoppin]

Originally posted by: BFG10K

but not like Link Boost ,,, agreed!

Yes, that's the point. This isn't link-boost, this is something much more.

It?s undocumented GPU overclocking while the driver apparently tries to hide it by reporting incorrect clocks.

there will be tools to report it correctly even if RT doesn't

Huh? It's because of RivaTuner that we found this out as it can read the GPU clock directly. Anything reading the driver clock will have incorrect frequencies.

so, BFG, do you really think that the performance increase that we see by O/C'ing the PCIe bus is NOT related to Link Boost?
... read the description of Link Boost again

i am quite certain nvidia is aware that RT would read the GPY core speeds directly .. i am not privy to the "mind of nvidia" {except by psychic link of course } .. so they need to explain their "reasoning" as to why they did not tell the reviewers yet.

imo, this is a Tempest in a Teapot ... if AMD did it the nvidia fans would cry foul ... a 'lose-lose' situation for their PR .. they need to explain

Here's my prediction:

In a week or so we will see a Press Release from NVIDIA extolling the praises of LinkBoost and more papers and releases explaining the benefits thoroughly ... and a week later AMD will announne it's version
*stink boost* ... or whatever
:roll:

Originally posted by: bryanW1995
luke, you are my son, come to the dark side...(cue cheesy music)

sorry to disappoint you .. the Emp already rejected me as Beyond Chaotic Evil ... too unrepently evil and especially too unpredictably evil .. and there is a -balance+
-you ought to meet my twin = B-O-R-I-N-G

 

[bryanW1995]

ouch! I didn't know that they had anything worse than Chaotic Evil...is that like, um, I dunno, an EVIL sea bass? or at least an ill-tempered one?



edit: Apoppin, here's the real problem with "link boost" or whatever they choose to call this gpu overclocking tool: even if the DO have a press release in a week or two about it, all of the ORIGINAL reviews are already done. MANY people have made a decision about 9600gt based on 1. a "sneaky" or at least "unknown to anybody outside of nvidia for weeks after release" overclock, and 2. newer drivers that aren't publicly available yet giving the card a decided advantage vs 8800gt (and the AMD cards since 8.3 should be out at the same time as 174..).

9600gt is a good card, but I'm VERY glad that I wasn't in the market for a card when it was released since I might have fallen for the hype.

 

[apoppin]

Originally posted by: bryanW1995
ouch! I didn't know that they had anything worse than Chaotic Evil...is that like, um, I dunno, an EVIL sea bass? or at least an ill-tempered one?

with a laser attachment, of course

[g-d L/S is broke again]

 

[lopri]

i *only* reported that Link Boost is not new and is clearly for O/C'ing the GFX card that resides on that PCIe bus. No rescue needed. And, although all the nvidia fans call me a "AMD fanboy", i got attacked by AMD fans. i find that a little fickle. i haven't changed. i call it as it see it and when i am wrong i have always been glad to admit it.

Originally posted by: apoppin

it seems to me the only DIRTY TACTICs are the PR assholes that passed this info along "as shady" to a very GULLIBLE TechPowerUp.com journalist.

"foul language", get real - is it only foul if i fucking say it? - or do i not have the privileges the other members do?

-[no f'ing way .. wtf -where are the NV PR guys? ... they know this crap better than i do (i guess)!

.. and i think AMD's viral PR first pointed this "shady" shit out ... not TechPowerUp ... they are the unwitting 'tools' to get this twisted 'message' out

you are MISinforming gullible "people"

edit .. since the balance of this forum has apparently shifted ... toward stupidity ... i guess i NEED to get a GX2 to counteract it ...

Frankly it is becoming embarrassing for me to be associated in ANY way with ATi fanboys ... it is becoming ridiculous .. i feel really stupid having to defend nvidia from retarded false charges

ANSWER the FACTS i posted FIRST before you do the DISTRACTION DANCE

 

[ArchAngel777]

The fact is, nVidia's own driver does not report the proper frequency of the 9600GT. It is the only card thus far <that we know of> that has responded this way. So, you basically have two explanations.

1) nVidia did this on purpose.
2) nVidia didn't know about the driver reporting incorrectly.

Now, we have to go with the most probable of those two and I think it is clear that #1 is the most probable since none of ther other cards accidently reported the incorrect clock speeds.

I think the issue that you and BFG are debating over isn't the same. BFG is attacking this from the viewpoint that it is wrong to hide the true GPU clock speed. Which is 'deceptive' to say the least. You, Appopin, are trying to say that nVidia has been open about the PCI-E linkboost technology, which is also true.

But, the issue here is the deception involved. Two those so far are:

1) Why did they feel the need to hide the true clock speed of the GPU?
2) Why is this only on 'one' card thus far and not publicised to reviewers? Other cards have been able to support this from the 7900GTX (Link) yet none of them had it. The first to get it was the 9600GT.

Now, even more further deception is the compression technology that the driver was given for the 9600GT. It is clear that the driver supported any of the G80+ cards, but they decided to write an INF for ONLY the 9600GT.

So we have some major cheating going on with the card to give it an unfair advantage.

1) 9600GT given special drivers that will work on all G9X video cards, but nVidia decided not to include the other G9X cards.
2) 9600GT getting an overclock that is not reported in the drivers. No other card gets this overclock, but it is quite possible for nVidia to do this with G8X/9X

But people have said 'why complain about a free overclock?' Well, to be honest, it isn't the free overclock that I am complaining about. It is the deception that goes with it. if nVidia sold me a card that said "650 - 725Mhz <variable>" on the box you would have absolutely no complaints from me. Furthermore the fact that the driver doesn't report the true clock speed of the card is enough to convict nVidia in ANY jury that they have commited a crime here. That crime is deliberate deception to the public .

 

[apoppin]

Originally posted by: lopri

i *only* reported that Link Boost is not new and is clearly for O/C'ing the GFX card that resides on that PCIe bus. No rescue needed. And, although all the nvidia fans call me a "AMD fanboy", i got attacked by AMD fans. i find that a little fickle. i haven't changed. i call it as it see it and when i am wrong i have always been glad to admit it.

Originally posted by: apoppin

it seems to me the only DIRTY TACTICs are the PR assholes that passed this info along "as shady" to a very GULLIBLE TechPowerUp.com journalist.

"foul language", get real - is it only foul if i fucking say it? - or do i not have the privileges the other members do?

-[no f'ing way .. wtf -where are the NV PR guys? ... they know this crap better than i do (i guess)!

.. and i think AMD's viral PR first pointed this "shady" shit out ... not TechPowerUp ... they are the unwitting 'tools' to get this twisted 'message' out

you are MISinforming gullible "people"

edit .. since the balance of this forum has apparently shifted ... toward stupidity ... i guess i NEED to get a GX2 to counteract it ...

Frankly it is becoming embarrassing for me to be associated in ANY way with ATi fanboys ... it is becoming ridiculous .. i feel really stupid having to defend nvidia from retarded false charges

ANSWER the FACTS i posted FIRST before you do the DISTRACTION DANCE

Originally posted by: lopri
I am leaving this thread and will leave it to fellow forum members.

Why don't you keep your word?

i next expect to find you snipping my single words out of context to piece something together since you have no ability to debate the issue .. it is a distraction dance
:roll:

What you're doing, is not debating the issue, but chaotically assaulting it. Take it down a few notches. Calm down.

Anandtech Moderator - Keysplayr2003

 

[lopri]

Oh the mod informed me that things were cleared. But still, I didn't need a word of my own to make a point.

 

[apoppin]

Originally posted by: lopri
Oh the mod informed me that things were cleared. But still, I didn't need a word of my own to make a point.

i missed your "point"

try again ..

 

[bryanW1995]

Ok, maybe I'm just stupid, but I just spent 15 minutes trying to snipp apoppin's posts into something that could convict him. Or at least put him on trial. Or get him fined. Or something. Here's what I've come up with so far: "point" your single piece at my laser attachment

do I win the prize???

Seriously, though, apoppin, why are you so vehement in your defense of nvidia in this regard. I don't consider myself a "fanboy" of anybody, be it intel/nvidia/amd/samsung/or darth maul, but archangel has made a pretty good summation of the facts. Let's just admit that nvidia pulled a slick marketing trick and be prepared for it next time. I'm sure that amd is kicking themselves and saying "why didn't WE think of that", this is just representative of how nvidia has outmanouvered all other video card companies for the past 10 years.

 

[Janooo]

Originally posted by: apoppin

Originally posted by: Janooo

Originally posted by: apoppin
...
i *only* reported that Link Boost is not new and is clearly for O/C'ing the GFX card that resides on that PCIe bus. No rescue needed. And, although all the nvidia fans call me a "AMD fanboy", i got attacked by AMD fans. i find that a little fickle. i haven't changed. i call it as it see it and when i am wrong i have always been glad to admit it.
...

The Link Boost did not overclock GPU core. It's for PCIe only. You don't have any prove about GFX overclocking.
You say you'd be glad to admit when you are wrong. So, please, prove the 7900GTX overclock or admit that you are wrong.

We are discussing the 9600GT OC - you seem to be hung up on the fact that Link Boost was mentioned as being for O/C'ing the PCIe bus the old 7900 resided on. IF you cannot make the connection that perhaps O/C'ing the PCIe would would ALSO o/c the GPU on that bus, you need to do a little research. it is very basic.

i do not have a 7900; i never had a link-boost enabled MB ... but i have reports there is nothing unusual about Link Boost increasing GFX card performance on nForce 590 and above MB for G71. *Evidently* it just wasn't initially supported on G80, until G94 and 9600GT arrived. i don't have anything in my current bookmarks.

And i will be very glad to admit i am wrong ... when i am wrong

===========

oK , if it is not for Overclocking the GPU, what is it for?

It's for raising the PCIe clock speed as was explained earlier. As for the GPU requirement, those are probably the ones nVidia has tested with an overclocked PCIe bus but again that doesn?t mean GPU clocks are ever raised.

AFAIK the 9600 GT is the first card to do this and also the first card to report incorrect frequencies through the driver.

OK .. i don't see a problem ... we KNOW the basics that O/C'in the PCIe bus also O/C's the GPU ... but not like Link Boost ,,, agreed! ,,, it is new to G94 that they apparently expanded on its capabilities .. the 'proof of the pudding is in the eating of it'

it is "inaccurate" because it is new ..
there will be tools to report it correctly even if RT doesn't

You keep talking about the Link Boost. You say it's around since 2006 and you keep saying it's related to the overclocking of GPX's. That's false. It never overclocked a gpu core before.
There is no difference in overclocking with Link Boost and overclocking PCIe on any other MB. The frequency is calculated in the same way on both.

 

[lopri]

Originally posted by: ArchAngel777
The fact is, nVidia's own driver does not report the proper frequency of the 9600GT. It is the only card thus far <that we know of> that has responded this way. So, you basically have two explanations.

1) nVidia did this on purpose.
2) nVidia didn't know about the driver reporting incorrectly.

Now, we have to go with the most probable of those two and I think it is clear that #1 is the most probable since none of ther other cards accidently reported the incorrect clock speeds.

Originally posted by: ArchAngel777
But people have said 'why complain about a free overclock?' Well, to be honest, it isn't the free overclock that I am complaining about. It is the deception that goes with it. if nVidia sold me a card that said "650 - 725Mhz <variable>" on the box you would have absolutely no complaints from me. Furthermore the fact that the driver doesn't report the true clock speed of the card is enough to convict nVidia in ANY jury that they have commited a crime here. That crime is deliberate deception to the public .

A possible theory:

I wonder if this is NV's 'quiet' message towards Intel. They probably couldn't advertise "Our video cards work faster on our motherboards!" without worrying about Intel's wrath. But it's been known that NV has been seeking Intel's QuickPath license for some time with some frustration. NV wanted to put pressure on Intel and put forward on their own in the Platformization battle, and realized one weakest link on Intel chipsets - their PCIe lanes aren't able to handle high frequencies. This gives NV a perfect chance to advertise their platform without advertising, by way of 9600 GT (and possibly more to come), especially to non-overclocking general public. "NV cards just work better on NV boards"

NV chipsets, while not the most popular, are capable of high PCIe clocks thanks in part to their early adoption of HyperTransport (note that Intel's DMI is essentially a PCIe x4 lane, and it might share the clock generator with the PCIe lanes - though I am not sure about this), and this gave them the opportunity of LinkBoost ver. 1 in the past, and now the revised ver. 2, which they will argue their way through with, in case they face a criticism. If this theory stands, NV will prove itself once again to be smart in taking advantage of, well, whatever advantage they've got but at the same time they are walking in an extremely dangerous line. Because as easily as can this be seen as LinkBoost Ver. 2, this can also be seen as unfairly crippling competitor's platform. (which could be accompanied by a lawsuit)

 

[taltamir]

they could counter that by allowing anyone's boards to activate a linkboost state... provided that it is capable of sustaining the increased pcie speeds. Which they know their competitors can't.

 

[apoppin]

Originally posted by: bryanW1995
Ok, maybe I'm just stupid, but I just spent 15 minutes trying to snipp apoppin's posts into something that could convict him. Or at least put him on trial. Or get him fined. Or something. Here's what I've come up with so far: "point" your single piece at my laser attachment

do I win the prize???

Seriously, though, apoppin, why are you so vehement in your defense of nvidia in this regard. I don't consider myself a "fanboy" of anybody, be it intel/nvidia/amd/samsung/or darth maul, but archangel has made a pretty good summation of the facts. Let's just admit that nvidia pulled a slick marketing trick and be prepared for it next time. I'm sure that amd is kicking themselves and saying "why didn't WE think of that", this is just representative of how nvidia has outmanouvered all other video card companies for the past 10 years.

Who is defending nvidia? .. i am defending my own right to my own opinion without being subject to ridicule for it. i DO tend to become adamant at the bottom of a dogpile - it's my nature and it's self-defense.

Why should i "admit" something just because you made a statement urging me to do so: "admit that nvidia pulled a slick marketing trick"
.. You and the angel are assuming it is a market trick ... and i notice you use the word "slick" .. why is it "slick" ?


it is NO marketing 'trick' imo ... IF i had a link-boost MB and a 9600GT, i'd consider it a TREAT.
===============

You keep talking about the Link Boost. You say it's around since 2006 and you keep saying it's related to the overclocking of GPX's. That's false. It never overclocked a gpu core before.
There is no difference in overclocking with Link Boost and overclocking PCIe on any other MB. The frequency is calculated in the same way on both.

Of course ,,, the SUbject of this topic is "nvidia's "shady Trick" ... Link Boost IS part of it. ... and it isn't that "i" say it ... it is on nvidia's own website since '06. Any you say LB never O/C'd a GPU core BEFORE .... well, maybe so ... but that does not preclude it happening NOW.

You are the one saying that LinkBoost makes no difference ... i do not agree.

 

[taltamir]

Appopin is correct, there is nothing wrong with him giving nvidia the benefit of the doubt.
You assume they are trying to trick reviewers. He assumes that they are activating new features (that have been advertised since 06) as they become available, giving performance boosts with no drawbacks.
Trick or Treat, it's a matter of perspective.

 

[apoppin]

OK, ill update you: i had one of my friends with an old nForce system actually try it out and i will quote the results:

... with a nforce 590 chipset and a 7900GTX in it. Linkboost is an option and it's active. No instability. No core overclock either.

OK there you have it on 7900GTX. i checked for you. But then, so what? G71 and G94 are two different completely GPU's - it appears that we have PROGRESS. Why should nvidia not have improved LinkBoost in the last 2 years? Perhaps it was not ready for PrimeTime in 06.

Perhaps Rollo's contacts at NVIDIA can shed light on this ... or we can wait for a further article.

===============================

and i do want to respond to Lopri

Lopri's paragraph here:

NV chipsets, while not the most popular, are capable of high PCIe clocks thanks in part to their early adoption of HyperTransport (note that Intel's DMI is essentially a PCIe x4 lane), and this gave them the opportunity of LinkBoost ver. 1 in the past, and now the revised ver. 2, which they will argue their way through with, in case they face a criticism. If this theory stands, NV will prove itself once again to be smart in taking advantage of, well, whatever advantage they've got but at the same time they are walking in an extremely dangerous line. Because as easily as can this be seen as LinkBoost Ver. 2, this can also be seen as unfairly crippling competitor's platform. (which could be accompanied by a lawsuit)

How does nvidia - using their own pciE first developed for AMD platforms -Linkboost was introduced on nforce 590 for *AMD CPUs* - can possibly be "unfairly crippling" intels' own chip sets? Unless you want to take his argument to the extreme that intel's chipset engineer are also incompetents regarding PCIe ... i think not

 

[lopri]

Originally posted by: taltamir
they could counter that by allowing anyone's boards to activate a linkboost state... provided that it is capable of sustaining the increased pcie speeds. Which they know their competitors can't.

True that.

After collecting some more information, things don't look as simple (or complex). What I've gathered;

On 780i reference board

1. 1. When PCIe 2.0 card is inserted - both NV and AMD card (!), according to a poster @Guru3D - the top and bottom PCIe 2.0 x16 slots are automatically set @125MHz. (Note that the middle PCIe x16 slot is PCIe 1.0)
   
   2. This doesn't show in the BIOS nor any other proggy that reads clocks from drivers, but only in ClockGen or something that reads clocks off the PLL directly.
   
   3. 9600 GT has a PLL of 27MHz, unlike NV's claim of 25MHz. This, when combined with the PCIe clock 125MHz, results in discrepancy in clock speed reported by drivers and ClockGen.

The above are the only confirmed information so far, it seems. This leaves just too many holes to fill.