The 6800 is overpriced junk!

[stardust]

I fully endorse bios overclocking of components with full support by the manufacturer.. warranty baby!

However, I would be more excited to see some voltage adjustment capabilities with the overclock so the enthusiasts don't even have to void their warrenty for a TRULY satisfying overclock. What companies should really be jealous of is OCZ's 3.1V mem capability and start crying, "OCZ is cheating because at 3.1V that mem has the potential to overclock much further than our 2.8V mem sticks!"

Sorry DAPUNISHER, I still think the new PEG function is a sub-enthusiast level feature like their AI overclock... and thus warrants the flame type thread you purposely created lol

On the 6800 matter, in the words of Rollo, "the 6800 is looking better every day"

 

[DAPUNISHER]

Sorry DAPUNISHER, I still think the new PEG function is a sub-enthusiast level feature like their AI overclock...

Do we have enough data to safely conclude that? I'd say probably not, but only further investigation by hardware sites *hint hint* will provide the answers we seek.

 

[stardust]

Well, what I personally think is the limiting factor to these sort of functions is the fact that you could only push hardware so much before you are in the warrenty voiding territory.. I would truely be interested if certain components like the BFG OC (arbitrary) series were made to not only come factory overclocked, but have the ability, like OCZ mem sticks, to increase the tolerance of the cards for an even higher overclock.

And if ASUS geared their PEG to have additional support for this new "BFG OC series" kind of hardware and have more choices in terms of overclocking features, then it truely makes sense for other companies like gigabyte to become jealous..

In summary, its like having the motherboard automatically seek out and identify the highest clock the video card can handle safely (like coolbit's autodetect) and adjust its settings accordingly.

 

[TStep]

Here is a little info from XBit. Seems as though it is adjusted through the bios before windows startup.

 

[DAPUNISHER]

Originally posted by: TStep
Here is a little info from XBit. Seems as though it is adjusted through the bios before windows startup.

Thanks for the link That makes it sound like a small boost for fudging benchies cheat instead of a serious overclocking feature.

 

[TStep]

Benchie fudging is my take too. Overclocking by 5% is nothing around here.

 

[Xentropy]

Originally posted by: DAPUNISHER

Originally posted by: TStep
Here is a little info from XBit. Seems as though it is adjusted through the bios before windows startup.

Thanks for the link That makes it sound like a small boost for fudging benchies cheat instead of a serious overclocking feature.

My worry is how this is implemented. Is it a change in the clock signal supplied to the card? If so, would adjusting clocks in software with this feature unknowingly activated by default in a given user's system cause real values to differ from the values the software reports?

For instance, if the default clocks of a card are 400/1000, and the BIOS setting is supplying a 6.25% faster base clock signal causing the default multipliers to result in a 425/1062 clocked card, and then you open up a software OC program, does it report 400/1000 or 425/1062? Is it reading the actual frequencies or just the multipliers and assuming a "normal" incoming clock signal?

If the latter, pushing the sliders to 450/1100 could actually be setting your card to 478/1168. An overclock that doesn't seem dangerous could be dangerous due to EXTRA multipliers you didn't know were there. Add to this things like, assuming that overclock was stable, someone coming to the forum and posting their benchmarks "@450/1100" would be misleading, since unbeknownst to anyone, those results were actually "@478/1168".

Overall, I give this "feature" a thumbs down, especially if it doesn't default to off even when loading BIOS turbo settings. At the very least, it needs to be made very clear when it is active, and, if they don't already, software OCing programs need to be able to see the true clocks, not assume the signals provided to the video card by the motherboard are correct.

 

[AnitaPeterson]

I have a serious question here... How long until the next generation of cards? I seee the 6800GT come highly recommended, but I need to know if it'll last as long as my current trusted v7700 (GF2 GTS)...

 

[VirtualLarry]

Originally posted by: DAPUNISHER
LOL Thanks brudda. Gigabyte is bent because they fear it will make the Asus offering based on those chipsets look far superior to their own in 3D benchies, but given it appears to be user controlled reviewers can just disable it I should think and avoid the controversy. If anything, Gigabyte should be ashamed that they can't refute Asus's statement about being the only one who can juice that chipset for more fsb overclocking IMO

From reading the last bit of that article on The Inq, I came to the conclusion that it's not so much that the feature exists, but that it wasn't disclosed, either in the review by the reviewer, or by Asus to the customer. (Is it a visible BIOS option somewhere? Or is it some sort of auto-overclock, like some of MSI's earliest overclocking P4 boards were. I'm sure that you remember the TH article about that snafu, it was just before MSI annnounced their marketing push for "D.O.T.".)

I wonder though, was this all Asus's invention, and if so, why isn't the feature implemented in such a way that it also overclocks NV cards? Or was it a stealth marketing tact by ATI, to try to sell more PCI-E graphics cards to enthusists, after the word leaked out? Or possibly it was only a way to get Asus higher scores in review benchmarks, knowing that ATI is the only one with "native" PCI-E graphics boards on the market right now.

I would compare this much akin to some mobo mfg's, that (intentionally) tweak the base clock that their system/FSB clocks are derived from, in order to look better on the (altogether stupid in the first place) benchmarks that reviewers sometimes do. (Motherboards should be benchmarked on stability and features, not performance, per se. If anything, points should be deducted, IMHO, if frequencies are too far off from what they are supposed to be, plus or minus. Reasoning being, the frequencies might be due to using cheaper components, with looser tolerances, and running more out of spec. So mobo makers that intentionally overclock by a few mhz, should likewise be punished for this.)

 

[VirtualLarry]

Originally posted by: SickBeast
A few things re: the original topic.

1. I would never purchase a Prescott motherboard unless intel does something drastic to the chip. It's interesting that they're doing this on the intel boards when the A64s are clearly better for gaming.

2. I'm surprised ATi is allowing this sort of feature to be implemented. It wasn't long ago that they clock-locked many of their BIOSes.

Very good point. My Radeon 9200's BIOS is locked. That makes you wonder, if the mobo OC's the video card, and the video card shuts down and fails to boot when OC'ed, then what happens next? Who will the user blame for the boot failure incompatibility between Asus's mobo and ATI's video cards?

Originally posted by: SickBeast
3. This would be a nice feature as it would eliminate the need for overclocking software once the max core/mem clocks have been determined. ATItool will still be best for this IMO. Hopefully they've even added settings for GPU and memory voltages in the BIOS; that would be amazing.

IMO, ATITool is shite, or it has some severe incompatibilities with my system or system chipset. Just simply opening the program, can lock my machine. Letting it run for quite some time, seems to somehow make my system unstable. This is all at *stock* card speeds/voltages, btw. I don't know, I suppose it could be the ATI drivers too. Perhaps that "large system cache" memory-corruption issue that WinXP users have experienced with ATI cards really are the fault of the ATI drivers, and W2K systems suffer from them as well, except that due to a slightly different address-space layout between W2K and XP, they don't notice it as much.

Edit: Actually, what would also be a really great feature, would be if the system BIOS re-programmed the PCI config space of the card, to make it think that it was a FireGL board, or enable the extra pipelines, etc. (BIOS-level "softmod"). That would be interesting, moreso than just straight-up overclocking to me.

 

[VirtualLarry]

Originally posted by: TStep
To clarifiy:
Is the bios of the motherboard clock manupulating the bios of the video card at start up prior to starting windows?
or
Is it using an Asus developed overclocking tool similar to ATItool on windows startup, just in a cloaked fashion?

I know from experiments with my "locked" Radeon 9200, using the stock ATI drivers - it appears that there are software-configurable clock freq/divisor settings in the GPU. The stock ATI drivers, either check or set the freq (or let the card's BIOS do it during initialization), and then "lock" those settings in, basically making those registers read-only for the remaining duration, until power-cycled. There were some registers on the old IBM VGA that were similar, I think. (The PSX's hardware copy-protection scheme uses something similar, too.) In order to overclock using ATITool, I had to first install the "unlocked" (hacked) Cats from ocfaq.com. I guess they NOP out the "lock" code in the low-level drivers. I noticed that they missed a spot though, because if "VPU recover" ever kicks in, that also resets the clocks to BIOS defaults, and then "locks" the clock registers too.

So the system BIOS, during bootup, has the ideal opportunity to change those clock speed register settings, after video card BIOS's initialization, but before the low-level Windows' video-drivers load to "lock" those settings in. It's pretty clever, really, but I hope that the end-user has full control over it. Otherwise, I don't think that it's a good idea at all, since it probably does also technically invalidate your warranty on your nice shiny new expensive ATI video card.

 

[DAPUNISHER]

I would compare this much akin to some mobo mfg's, that (intentionally) tweak the base clock that their system/FSB clocks are derived from, in order to look better on the (altogether stupid in the first place) benchmarks that reviewers sometimes do.

It certainly is looking that way, and it's definitely cheating to gain an unfair advantage in benchmarks.

 

[Regs]

You should submit this for news on the Anandtech page. Anand loves to see the competition between hardware companies. And, yeah, we all do too! This is exciting news.

 

[DAPUNISHER]

+

Originally posted by: AnitaPeterson
I have a serious question here... How long until the next generation of cards? I seee the 6800GT come highly recommended, but I need to know if it'll last as long as my current trusted v7700 (GF2 GTS)...

There's no way to answer that for certain because games are continuously pushing the envelope. It seems reasonable to expect that it should offer longevity similar to your current card given that the GTS has seriously dated feature set and performance now. If you are referring to it's reliability, get the BFG and it's covered for life, or as long as they are in biz anyways

 

[jamesbond007]

Who is going to warrant the card if it is fried through these 'options/abilities' of the BIOS? Does this work only with ASUS ATi-based cards? All ATi cards? I can understand it as being a great ability if it works exclusively with ASUS' own ATi cards...

I've read most of the posts in this thread, but have not come up with an answer to this question.

Thanks!

 

[DAPUNISHER]

Originally posted by: jamesbond007
Who is going to warrant the card if it is fried through these 'options/abilities' of the BIOS? Does this work only with ASUS ATi-based cards? All ATi cards? I can understand it as being a great ability if it works exclusively with ASUS' own ATi cards...

I've read most of the posts in this thread, but have not come up with an answer to this question.

Thanks!

We are just speculating, there isn't enough info to draw any definite conclusions from. Unfortunately, it looks like a cheat to enhance benchmark scores instead of a real overclocking feature for enthusiasts so far. I don't know if ATi worked with them on this or if it's something they came up with on their own. Asus is touting it as an enthusiast feature so unless the ability to boost speeds far in excess of 5% are included they are full of shiat

 

[jamesbond007]

ATi could have or could have not worked with ASUS on this, since ASUS does market ATi-based videocards. They could probably just test it with their own hardware that they are manufacturering themselves.

It's quite ironic that ASUS would do this though. Perhaps ATi slipped ASUS a few extra bucks to come up with this 'feature'? "Hey, make our cards perform faster than nVIDIA's. We'll look better against the competitor and your motherboards, coupled with our graphics chips, will have higher sales than Abit."

Maybe? Maybe not? Two powerful companies, such as ASUS and ATi, can easily help each other out to squish the sales of competitors.

 

[Cerb]

And this is news? BFG has already been warrantying overclocks for NV cards, now Asus just gives the options for it in ATI ones. Woah.
OK. I don't see these fantastic or dire implications. Am I just missing the black choppers, or what?
Granted, a BIOS is better as an interface than Rage3DTweak to get the job done, but still...eh.

6800 being 30% faster: The Geforce4 4200 was only about that much faster than a GF2 Ultra, as well, in games of the time at no more than 800x600. Funny thing, we started pushing the hardware and using higher res, and what do you know, it started to pull away madly. Doom3 is one of those games--it just arrived early, because Carmack is a living crystal ball

(note: I have a 5900XT, no 6x00 in sight for me)

 

[DAPUNISHER]

No black choppers, but it's not cool if this is the cheat it looks like. It is entirely different than pre-overclocking a vid card and promoting it as such. It appears Asus did it in an underhanded fashion, and just enough, to make their board look like a 3D performance champ compared to the competition. Then they started touting the board's performance while failing to mention that the vid card was being overclocked and subsequently responsible for the performance gains.

Gigabyte calls them on it, and they respond saying they're jealous of this new enthusiast feature. However, unless some serious overclocking percentages are permitted it's nothing of the sort.

Good points about the 6800, potential is the real difference between the 6800 and the 9800p. If this 16 pipe soft mod turns out to be solid and the percentages of successful mods high, it'll just absolutely own for the $275-300 pricepoint.