tweaktown gx2 review

[Piuc2020]

People are blind and I guess I should remind you...

The 9800GX2 is a single card multi GPU card featuring dual 8800GTS G92 chips... go look for a review of SLI 8800GTS's, I guarantee the numbers will be higher than the HD 3870 X2. The 9800GX2 performing bad is probably just a driver issue, after all, the card is not supposed to be out for 2 more days anyway.

Instead of spreading FUD, you should try to see the big picture and wait for things to play out before trolling around the forums.

 

[Sylvanas]

Originally posted by: Piuc2020
People are blind and I guess I should remind you...

The 9800GX2 is a single card multi GPU card featuring dual 8800GTS G92 chips... go look for a review of SLI 8800GTS's, I guarantee the numbers will be higher than the HD 3870 X2. The 9800GX2 performing bad is probably just a driver issue, after all, the card is not supposed to be out for 2 more days anyway.

Instead of spreading FUD, you should try to see the big picture and wait for things to play out before trolling around the forums.

Just because it has two GPU's onboard it doesn't mean it will magically be exactly like an SLI setup...more goes into it than that. Case in point the 7950GX2....

I await Anand's review.

 

[dadach]

like i said...he needs to say what he knows or keep his mouth shut until people can see some credible benchmarks...all the hints he is dropping about the card could be misleading, but due to where is coming from, some people might actually believe him, and hold off buying something they wanted right now...so either he states the facts that he knows, which is fair, so we get none of the:"might, could, maybe" stuff



Dadach, feel free to take his comments with a grain of salt in the absence of other information, and everyone else is free to do the same.

You can say something like ?all of the evidence I?ve seen thus-far points to miserable performance on the GX2, and nRollo hasn?t provided any hard evidence to the contrary, so I have good reason to doubt his claims?.

Telling him to "shut the F up" however is not appropriate because he's entitled to post here just like you, plus he hasn't violated any forum rules that I can see.

Thanks for your co-operation in this situation.

Video mod BFG10K.

 

[Sable]

All of the evidence I?ve seen thus-far points to miserable performance on the GX2, and nRollo hasn?t provided any hard evidence to the contrary, so I have good reason to doubt his claims.

That and the fact he's clutching at straws for positives by pointing out the temperature. People buying a ridiculous hog of a card like that are (in my opinion of course) NOT worried about the temperatures.

Don't worry Rollo, I know you're just doing your job.

 

[nRollo]

Originally posted by: dadach
like i said...he needs to say what he knows or keep his mouth shut until people can see some credible benchmarks...all the hints he is dropping about the card could be misleading, but due to where is coming from, some people might actually believe him, and hold off buying something they wanted right now
...so either he states the facts that he knows, which is fair, so we get none of the:"might, could, maybe" stuff

People should hold off buying anything high end right now (NVIDIA cards included) because launch date for the GX2 is rumored to be 3/18.

If I were considering buying a high end card(s) and spending a lot of money, I think I could definitely wait till the day after tomorrow to see why a card with specs that should blow away a 3870X2 is getting beat by it, and two guys that supposedly have the cards and have seen other non-published benches are calling shenanigans on the January 173 series drivers when we're a few revisions past that.

Ever consider pre-release drivers from months in advance of launch just may not enable full product functionality to prevent competitors from knowing how cards perform dadach?

As far as "posting what I know" goes, I could do that, but:
a. I'd be betraying people I consider friends who have stood by me
b. I'd lose access to pre launch and insider info, which I like a whole lot (it's pretty fun being at the web conferences with guys like Anand and Derek, and invited to product launches)
c. I'd lose all the free hardware and software
d. I'd likely be sued by NVIDIA who would only have to hold up my signed NDA in court and win

So, no, all you're going to get from me till NDA lifts are hints. If people want to buy 3870X2s today in hopes of getting the highest performing single slot solution, I'd say "Go for it, and if the rumors are true and nRollo/ChrisRay aren't lying, you 'd better do overnight shipping so you get one day of feeling like you have the best."

 

[nRollo]

Originally posted by: Sable

That and the fact he's clutching at straws for positives by pointing out the temperature. People buying a ridiculous hog of a card like that are (in my opinion of course) NOT worried about the temperatures.

LOL- I'm not clutching at straws, and personally agree with you that heat isn't a deal breaker on high end solutions. However, many people on forums across the web speculated heat would be a deal breaker on these cards due to their design, and it's nice to see NVIDIA has apparently engineered a product that runs much cooler than it's competition.

It's also possible that if fullfunctionality of these cards isn't being used by the 3 month old beta drivers, they will run hotter when fully functional.

Time will tell.

 

[Tempered81]

Originally posted by: Piuc2020
People are blind and I guess I should remind you...

The 9800GX2 is a single card multi GPU card featuring dual 8800GTS G92 chips... go look for a review of SLI 8800GTS's, I guarantee the numbers will be higher than the HD 3870 X2. The 9800GX2 performing bad is probably just a driver issue, after all, the card is not supposed to be out for 2 more days anyway.

Instead of spreading FUD, you should try to see the big picture and wait for things to play out before trolling around the forums.

Who is this directed at?

 

[bryanW1995]

if you want to criticize somebody here it is best to be very circumspect. being funny and/or oblique is a bonus. please review my signature for ideas

back on topic, why are we even talking about this? We don't need rollo to throw out his "I know this stuff but can't tell you ..." crap to know that 9800gx2 is going to stomp all over 3870x2. They have to price that card based upon its performance against its competition. A $200 price premium for 25% performance increase is very typical in high end video card situations, just check out 8800gtx prices from 2007 vs 2900xt and 8800gts 640. Is it going to be worth $600 vs $400 for 3870x2? Not unless you NEED that extra performance RIGHT NOW. 4870x2 is 3 months out and gt200 is 6-8 months out. Anybody taking bets on how well those will do against 9800gx2 or anything out right now?

I'm sure that when r700 rolls around we'll see similar "leaks" that initially show crappy performance, and we'll probably even see rollo imply that the nvidia solution is somehow superior. Let's just sit tight for 2 days and hope that nvidia doesn't resort to tricky overclocking techniques again to boost their performance numbers.

 

[Munky]

The only good thing about the gx2 release is the implication that Nvidia must be working on something really good, and they need more time to finish it. Same thing about the 3800x2 card. I'm waiting for the real next-gen cards; these sandwich card solutions just don't seem that intersting.

 

[surfsatwerk]

Originally posted by: bryanW1995
Let's just sit tight for 2 days and hope that nvidia doesn't resort to tricky overclocking techniques again to boost their performance numbers.

If it increases the cards performance I don't care if Nvidia sacrifices a virgin to Cthulhu. I don't know why people get so bent out of shape over what they did with the 9600 GT. I have yet to see one person who owns one say, "It's a great card for the price, too bad there's some funky overclock mojo going on ".

 

[apoppin]

i believe we need to wait for release day before passing any judgment

... on GX2

too many unknowns .. but the past points to a Video card that will beat the HD3870x2 as decisively as 7800GTX-512M beat x1800xt.
- since we are still guessing at price and performance .. at whether r700 is a factor or GT200 is rumored to be 2x faster
.. what real judgments about it can we make?

and there is *nothing* "shady" ... or "tricky" .. or deceitful or unfair about O/C'ing the PCIe slot to boost 9600GT's performance

the only thing possibly "unfair" would be that AMD hasn't thought of using it for themselves - yet

 

[SlowSpyder]

Originally posted by: nRollo

Originally posted by: Sable

That and the fact he's clutching at straws for positives by pointing out the temperature. People buying a ridiculous hog of a card like that are (in my opinion of course) NOT worried about the temperatures.

LOL- I'm not clutching at straws, and personally agree with you that heat isn't a deal breaker on high end solutions. However, many people on forums across the web speculated heat would be a deal breaker on these cards due to their design, and it's nice to see NVIDIA has apparently engineered a product that runs much cooler than it's competition.

It's also possible that if fullfunctionality of these cards isn't being used by the 3 month old beta drivers, they will run hotter when fully functional.

Time will tell.

So the Tweaktown performance numbers for the 9800GX2 aren't that good so you say to wait for more reviews because something is fishy with the Tweaktown numbers. But, the heat numbers are good in the Tweaktown review so you claim victory over ATI in regards to the thermals measured in the very same review.

 

[surfsatwerk]

Originally posted by: SlowSpyder
So the Tweaktown performance numbers for the 9800GX2 aren't that good so you say to wait for more reviews because something is fishy with the Tweaktown numbers. But, the heat numbers are good in the Tweaktown review so you claim victory over ATI in regards to the thermals measured in the very same review.

Originally posted by: nRollo

It's also possible that if fullfunctionality of these cards isn't being used by the 3 month old beta drivers, they will run hotter when fully functional.

Time will tell.

 

[bryanW1995]

Originally posted by: apoppin
i believe we need to wait for release day before passing any judgment

... on GX2

too many unknowns .. but the past points to a Video card that will beat the HD3870x2 as decisively as 7800GTX-512M beat x1800xt.
- since we are still guessing at price and performance .. at whether r700 is a factor or GT200 is rumored to be 2x faster
.. what real judgments about it can we make?

and there is *nothing* "shady" ... or "tricky" .. or deceitful or unfair about O/C'ing the PCIe slot to boost 9600GT's performance

the only thing possibly "unfair" would be that AMD hasn't thought of using it for themselves - yet

it's shady that most programs, including ntune, don't show the increased clocks. Dynamic overclocking itself is a great idea, however, most people would like to know what's going on.

What is the base core clock on a 9600, 650? With that base core clock it does pretty well against 8800gt. However, with the core at 685 or 700 it does even better. Why wouldn't nvidia have the higher core clock displayed to users if they weren't being sneaky? Was it an "oversight"? If they were just competing with 8800gt then we could call it an oversight, but they are really competing with 3850/3870. That upper midrange level is a very crowded field right now, and many people who buy a card in that range will probably oc. wouldn't it be nice if you were oc'ing to know that nvidia has already introduced an automatic-oc function? Maybe it would even be nice to know how to disable said feature? And you don't think that it's "sneaky"???


edit: by the way, I'm still mad at you for stealing my cheetah...

 

[apoppin]

Originally posted by: bryanW1995

Originally posted by: apoppin
i believe we need to wait for release day before passing any judgment

... on GX2

too many unknowns .. but the past points to a Video card that will beat the HD3870x2 as decisively as 7800GTX-512M beat x1800xt.
- since we are still guessing at price and performance .. at whether r700 is a factor or GT200 is rumored to be 2x faster
.. what real judgments about it can we make?

and there is *nothing* "shady" ... or "tricky" .. or deceitful or unfair about O/C'ing the PCIe slot to boost 9600GT's performance

the only thing possibly "unfair" would be that AMD hasn't thought of using it for themselves - yet

it's shady that most programs, including ntune, don't show the increased clocks. Dynamic overclocking itself is a great idea, however, most people would like to know what's going on.

What is the base core clock on a 9600, 650? With that base core clock it does pretty well against 8800gt. However, with the core at 685 or 700 it does even better. Why wouldn't nvidia have the higher core clock displayed to users if they weren't being sneaky? Was it an "oversight"? If they were just competing with 8800gt then we could call it an oversight, but they are really competing with 3850/3870. That upper midrange level is a very crowded field right now, and many people who buy a card in that range will probably oc. wouldn't it be nice if you were oc'ing to know that nvidia has already introduced an automatic-oc function? Maybe it would even be nice to know how to disable said feature? And you don't think that it's "sneaky"???


edit: by the way, I'm still mad at you for stealing my cheetah...

i'm still mad at you for not sending me a AMD T-shirt. :|
--i get crap-all from them and their fans
-so there

and actually, i'd have picked a Jackal .. or a Leopard ... but cheetah is close enough [if he misses he gets 'leftovers' ]
:Q

Sure, it'd be "nice" to have an extra really helpful feature disclosed before you buy it, but i tend to look at it as a "plus" ... and i am pretty sure that later drivers will allow nTune to report it properly

as to "competing" with AMD's graphics cards, it isn't like they are using "voodoo" - there is nothing UNFAIR about it ... it doesn't affect anyone negatively .. SOME get MORE than what is advertised. No one *loses* anything. and i will BET you that r700 does the same thing

except they will crow about it

 

[bryanW1995]

sorry about the t-shirt thing. I just wear polo's to work every day so I need lots of white t-shirts. My white amd shirts are good b/c they can be easily converted into white flags when intel/nvidia/etc whip up on them too badly. oops, did I say "them"? uh, I meant "us".

amd needs to roll out r700 asap to take advantage of this pathetic 9xxx crap. they can acquire some serious market share for a while if they play their cards right. or introduce their cards right, for that matter

 

[apoppin]

Originally posted by: bryanW1995
sorry about the t-shirt thing. I just wear polo's to work every day so I need lots of white t-shirts. My white amd shirts are good b/c they can be easily converted into white flags when intel/nvidia/etc whip up on them too badly. oops, did I say "them"? uh, I meant "us".

amd needs to roll out r700 asap to take advantage of this pathetic 9xxx crap. they can acquire some serious market share for a while if they play their cards right. or introduce their cards right, for that matter

but if they do that it will draw out GT200 which is "ready", remember?

i think both AMD and NVIDIA are conspiring together to rid themselves of all the old junk first ... THEN we will see r700/GT200

just remember it will be Launched on Day 91- immediately after your EVGA step up expired
:Q

how's is that for "conspiracy"?

 

[bryanW1995]

that evga step-up argument is always fun to examine, but they aren't going to let you step up to a 10800gtx or whatever anyway. AMD hasn't done anything new other than 3870x2 since november, so they're due...if they make it in may it'll be exactly a year since r600. Funny how much difference a year can make...nobody is expecting them to destroy nvidia with r700 but it seems highly likely that they'll at least retake the performance crown for a while. depending on who you are they'll have the crown for something between 1 day and 6 months...we shall see...

 

[Extelleron]

Originally posted by: apoppin

Originally posted by: bryanW1995
sorry about the t-shirt thing. I just wear polo's to work every day so I need lots of white t-shirts. My white amd shirts are good b/c they can be easily converted into white flags when intel/nvidia/etc whip up on them too badly. oops, did I say "them"? uh, I meant "us".

amd needs to roll out r700 asap to take advantage of this pathetic 9xxx crap. they can acquire some serious market share for a while if they play their cards right. or introduce their cards right, for that matter

but if they do that it will draw out GT200 which is "ready", remember?

i think both AMD and NVIDIA are conspiring together to rid themselves of all the old junk first ... THEN we will see r700/GT200

just remember it will be Launched on Day 91- immediately after your EVGA step up expired
:Q

how's is that for "conspiracy"?

I really do not believe that GT200 is just ready, waiting for whenever nVidia needs it...

Just because a chip tapes out, that does not mean it is anywhere close to being ready for release. New chips, especially a huge chip like GT200, will go through multiple revisions before being launch-ready. Look at R600... R600 taped out in the summer of 2006, and we didn't see product shipped until Q2 2007. Intel's Nehalem taped out in September of last year IIRC, and it won't be launched till H2 2008.

nVidia is going to have to compete against R700 with G92b, the 55nm shrink of G92, and I highly doubt that it will be able to match R700 performance. I certainly hope it won't.. if nVidia can match R700 with a 55nm G80, then I'll have lost all hope in AMD's graphics division.

If things go right for AMD, then R700 should be the performance leader for at least one quarter assuming R700 will be on track for late Q2 2008 and GT200 will be on schedule for late Q3-Q4 2008. It would not seem too unlikely that by that time, AMD would be launching a 45nm shrink of R700, considering TSMC's 45nm process should be ready by that time and AMD did a similar thing with R600, launching @ 80nm in Q2 and then shrinking to 55nm by Q4.

 

[bryanW1995]

when did nvidia start planning a 55nm g92 chip? I figured that they would jump straight to 45nm. If AMD goes down to 45nm before nvidia goes below 65nm then things could get mighty interesting...

 

[Rusin]

Well 45nm chips aren't coming soon. UMC and TSMC just started to update their lines to 45nm so where they would have those 45nm chips done ?. According to Nordic Hardware GT200 is pretty much ready and they are preparing it to launch around same time as R700

 

[Extelleron]

Originally posted by: bryanW1995
when did nvidia start planning a 55nm g92 chip? I figured that they would jump straight to 45nm. If AMD goes down to 45nm before nvidia goes below 65nm then things could get mighty interesting...

http://www.techreport.com/discussions.x/14328

Still unconfirmed but it definately seems true, and logical... G92 is a pretty huge chip, 334mm^2, and yields are supposedly not so good and nVidia wants a smaller chip so they can better compete against RV670.

GT200 has always been said to be 65nm, I don't know if nVidia would move it to 55nm. However, since GT200 will be so large, it may be better to have it on a more established 65nm process than moving to a newer 55nm process.

AMD/ATI has always been ahead of nVidia in moving to smaller processes. Notably ATI was the first to 90nm with the X1800 then nearly a 1/2 year later nVidia launched 90nm G71.

Originally posted by: Rusin
Well 45nm chips aren't coming soon. UMC and TSMC just started to update their lines to 45nm so where they would have those 45nm chips done ?. According to Nordic Hardware GT200 is pretty much ready and they are preparing it to launch around same time as R700

TSMC began low-power 45nm production in September 2007 and will certainly be ready for volume production of 45nm GPUs by Q4 2008. In the past TSMC/other foundries, and thus the GPU market, have been far behind the CPU market in process nodes... but this is changing, TSMC is aggressively moving to new processes to close the gap w/ Intel and the CPU market.

 

[bryanW1995]

that 55nm shrink sounds like pure fud. nvidia barely has time to do it before introducing gt200. maybe they'll keep 9xxx around as midrange parts and shrink them to 55nm.

 

[Extelleron]

Originally posted by: bryanW1995
that 55nm shrink sounds like pure fud. nvidia barely has time to do it before introducing gt200. maybe they'll keep 9xxx around as midrange parts and shrink them to 55nm.

It's completely logical if GT200 is a late Q3-early Q4 '08 product. 65nm G92 launched in early Q4 '07, 55nm G92 launches in Q2 to soften the blow from R700, then GT200 launches in the H2 timeframe to compete against (maybe) 45nm R700.

 

[andrei3333]

Thats it, i have decided, i am going to wait out this storm of half arsed 3rd remake g92 cards and wait for the GT200 or whatever (PS i also feel a name change coming as 10800GTX does sound too --prehistoric).. i have a feeling it will be released in September/November just before the shopping begins at thanksgiving for you American guys and girls.... that makes the most sense for business, regardless if the R700 comes out a few months before that...

Besides Software is what sells hardware, looks at the jump to 8800 series after Crysis came out, it was insane, and that was during the holiday season....there are no graphics demanding games coming out yet that require truly next gen h/w so there will not be anyone looking to "improve performance" until such a game comes out.. and THATs when Nvidiots will point their lightning rods towards my wallet and strike.... Agreed ?