GF100 (Fermi) previews and discussion

[Mousemonkey]

188 was the rated TDP of the 5870, according to AMD, but that number seems consistent with multiple review sites. I realize they measure system power draw at the socket, but you can infer the number is at least fairly accurate based on teh draw of the other cards in an identical system.

I know you love nVidia, but that number certainly wasn't pulled out of any asses.

I didn't say asses I said air and the reason for that is because once the GSOD issues are finally sorted I wonder if that number is still going to be the same but that's assuming that the issue is ever sorted.

Don't forget this is also a 50% larger chip so "winning" by a few percent isn't impressive at all.

Which is exactly what people were saying about the 5870 in relation to the 295 wasn't it?

 

[Guest]

No the 5870 is much smaller than the 295 in size. This is what is scary about nVidia at the minute, they are slower with bigger chips and that is bad bad news.

If the 470 scores are true that means a bigger chip, using more power, being slower than the 5870. That is really bad for what was supposed to be a great new architecture with fermi.

 

[crosko42]

http://tweakers.net/nieuws/65933/cebit-nvidia-gtx-480-op-de-gevoelige-plaat-gevangen.html

new pictures of 480

 

[notty22]

LOL , insert something here and this bad news, thats worse than expected, OH KNOW , the sky is falling ! thats been repeated about 4 times in the last two pages. As far as power, MSI is putting TWO 8 pin 150 watt pci-e connectors on its 5870, add 75 watts minimum available from the slot, I think that adds up to more than 190. I wonder why?mmmh
Fermi has got the natives restless !

 

[roofus]

I generally but not always have a Nvidia preference but if rumors are even partially true, I would consider this launch a failure given pricing, perf, thermal, tardiness. That isn't to say they cannot evolve from that arch because some of our best tech is evolved from not so impressive launch products but even a fanboy has to acknowledge this will most likely be a failure by comparison of the 5xxx series launch. It isnt the end of the world or the beginning of the end but it should serve as a wake up call for green team that they got fat and lazy for a few years and I cant help but think Fermi is a knee jerk reaction that didn't go through development and testing like previous products. I don't see myself buying one of these for any purpose until 2nd or 3rd revision which looks to me like around December 2010. The problem with that is I wont spite myself and buy an overpriced and inferior product if the 6xxx series beats it soundly and priced better. I think my 295 will withstand another year of crappy console ports passed off as PC games though.

 

[RealityRush]

No the 5870 is much smaller than the 295 in size. This is what is scary about nVidia at the minute, they are slower with bigger chips and that is bad bad news.

If the 470 scores are true that means a bigger chip, using more power, being slower than the 5870. That is really bad for what was supposed to be a great new architecture with fermi.

Considering the 480 is going to be competing with the 5870, and the 470 is competing with the 5850, if the 470 can match the 5870, I'm sure the 480 will smash it.

So in this case, bigger chip does in fact = > perf. And I'd hazard that a 480 will beat the 5870 by more than 5%, but we'll see in a few weeks now wont we?

*queue more rampant speculation*

 

[Mousemonkey]

No the 5870 is much smaller than the 295 in size. This is what is scary about nVidia at the minute, they are slower with bigger chips and that is bad bad news.

If the 470 scores are true that means a bigger chip, using more power, being slower than the 5870. That is really bad for what was supposed to be a great new architecture with fermi.

The key word there being 'new', with room to grow and improve and maybe even take GPU's in a new direction. I don't care if it uses 300w or 300kw as I don't buy into first gen graphic hardware but I do know that when ATi cards are 1 or 2fps faster then it's shouted from the rooftops about how much better they are and I'll never fail to be amused at how defensive those same fans get if it looks like the tables might be turned.

 

[Guest]

What is competing with the 5970 then?

 

[Mousemonkey]

What is competing with the 5970 then?

Who's to say that one of the AIB's isn't going to do a dual 480 card that takes no notice of the 300w PCIe spec, a bit like Asus have done with the Ares card?

 

[Guest]

Who's to say that Asus won't increase the clocks further or add another chip while taking no notice of the 300w pcie spec? That pci-e spec exists for a good reason.

 

[sabot00]

Who says ASUS won't remake the MARS card?

 

[eaclou]

I didn't say asses I said air and the reason for that is because once the GSOD issues are finally sorted I wonder if that number is still going to be the same but that's assuming that the issue is ever sorted.

Wow you are slippery. Go right for the irrelevant part of my post (HUGE difference between those two figures of speech) and sidestep with the GSOD issue.

whether you want to admit it or not, the GSOD is not the norm for the 5xxx series cards. It is a problem, certainly, and should not be discounted, but none of us really have a good idea of the scope of the issue except that it is not huge. (if it affected a quarter or more of the cards shipped, it would be all anyone would hear about, and reviews of the product on online retailers would reveal it, especially since people are more inclined to leave reviews when the product sucks and they're angry than when it worked as expected)

I own both nVidia cards and ATI cards, so don't sidestep with a 'fanboy' comment - i just wanted to point out that the 188W TDP is a real number, and to dismiss it as such makes you just as guilty as all the ATI fanboys as far as blind bias goes.

 

[Mousemonkey]

Who's to say that Asus won't increase the clocks further or add another chip while taking no notice of the 300w pcie spec? That pci-e spec exists for a good reason.

What? , the Ares is a dual GPU card that requires about 400w already!

 

[Guest]

http://bbs.chiphell.com/viewthread.php?tid=73452&extra=page%3D1

Wow thats a terrible 3dmark score! I used to get that with my old 8800gtx!

 

[Guest]

What? , the Ares is a dual GPU card that requires about 400w already!

ye i meant a 3rd chip? if you are allowing nVidia to do whatever they want in order to get a faster card out, why can't ati do it too?

when is nvidia going to get the industry-standard fastest card again? i'm betting never.

 

[Mousemonkey]

I didn't say asses I said air and the reason for that is because once the GSOD issues are finally sorted I wonder if that number is still going to be the same but that's assuming that the issue is ever sorted.

Wow you are slippery. Go right for the irrelevant part of my post (HUGE difference between those two figures of speech) and sidestep with the GSOD issue.

whether you want to admit it or not, the GSOD is not the norm for the 5xxx series cards. It is a problem, certainly, and should not be discounted, but none of us really have a good idea of the scope of the issue except that it is not huge. (if it affected a quarter or more of the cards shipped, it would be all anyone would hear about, and reviews of the product on online retailers would reveal it, especially since people are more inclined to leave reviews when the product sucks and they're angry than when it worked as expected)

I own both nVidia cards and ATI cards, so don't sidestep with a 'fanboy' comment - i just wanted to point out that the 188W TDP is a real number, and to dismiss it as such makes you just as guilty as all the ATI fanboys as far as blind bias goes.

If by your own admission your posts are irrelevant, I bother responding to them in future then.

 

[JAYDEEJOHN]

Considering the 480 is going to be competing with the 5870, and the 470 is competing with the 5850, if the 470 can match the 5870, I'm sure the 480 will smash it.

So in this case, bigger chip does in fact = > perf. And I'd hazard that a 480 will beat the 5870 by more than 5%, but we'll see in a few weeks now wont we?

*queue more rampant speculation*

This will only apply if the pricing is competitive, otherwise, its meaningless.
If the 480 wins by more than 15%, it may end up the single fastest gpu, as new 5870s roll out with higher clocks and better drivers.
Yes, time will tell, and from what Im seeing, Fermi needs to come in cheap cheap cheap

 

[TheGreatGrapeApe]

^^ But...that would mean Charlie was wrong...that can't be right...

Really?
Which Charlie would that be, cause it certainly isn't this one;
http://www.semiaccurate.com/2010/02/20/semiaccurate-gets-some-gtx480-scores/

"The fans were reported to be running at 70 percent of maximum when idling, a number that is far, far too high for normal use. Lets hope that this is just a BIOS tweaking issue, and the fans don't need to be run that fast. It would mean GF100 basically can't downvolt at all on idle. On the upside, if it's any comfort, the noise from the fans at that speed was said to be noticeable, but not annoying.

Doesn't seem like he was saying they were loud, but maybe it reads differently for others.

I don't know who would think the cooler was massive, since it was always touted as a GTX280 type cooler, so it's far from the ASUS ROG cooler. But perhaps that's why the fans are running @ 70%?

 

[JAYDEEJOHN]

Another misconception floating around is, the 480 will be sooo much better than the 470.
If the 470 struggles with a plain Jane 5870, the 480 isnt going to destroy the 5870 anyways.
Itll only be 12-15% faster at most than the 470, so if the 470 doesnt beat the 5870, thats cutting into the 12-15%, which a 5870 with clocks of 950 may beat, just a few things to ponder here, and is why if this is so, Fermi needs to be cheap cheap cheap

 

[notty22]

Another misconception floating around is, the 480 will be sooo much better than the 470.
If the 470 struggles with a plain Jane 5870, the 480 isnt going to destroy the 5870 anyways.
Itll only be 12-15% faster at most than the 470, so if the 470 doesnt beat the 5870, thats cutting into the 12-15%, which a 5870 with clocks of 950 may beat, just a few things to ponder here, and is why if this is so, Fermi needs to be cheap cheap cheap

Come on, now your calling a 5870 , "plain jane". Last time I checked it exploded on the market and became the fastest single gpu card by about 20-25%. We know ATI-Nvidia's architecture never match up tick for tock. Can you ever hit the submit button on a post that isn't total anti-doom-gloom Nvidia ?

 

[JAYDEEJOHN]

By plain Jane as opposed to ones thatll have higher clocks. I guess I should have used stock, since many here are sensitive sheeesh

 

[IzzyCraft]

Come on, now your calling a 5870 , "plain jane". Last time I checked it exploded on the market and became the fastest single gpu card by about 20-25%. We know ATI-Nvidia's architecture never match up tick for tock. Can you ever hit the submit button on a post that isn't total anti-doom-gloom Nvidia ?

20%? is exploded on the market and so what did nvidia's 8800gtx do nuke the market?(8800gtx up from the 7900gtx and 1900XTX was nearly 70-90% fps improvements and had better image quality over all along with bringing better AA to the market etc) Frankly this whole gen has been a disappointment in terms of theoretical performance and actually produced. People seem to choose to forget the R870 was suppose to be a ton better then what it actually came to be but now that's its been here people use it as a ruler to judge and dismiss the GF100 either way we should at least wait till some people who don't need to put tape or blur parts of photos run some test on the cards.

 

[TheGreatGrapeApe]

http://www.heise.de/newsticker/meldung/Nvidias-Fermi-Leistung-der-GeForce-GTX-470-enthuellt-946411.html
Unigine tesselation, 8x AA

RadeonHD5850 19 fps
GeForceGTX470 20 fps
RadeonHD5870 23 fps
http://www.realworldtech.com/forums/?action=detail&id=107906&threadid=107906&roomid=2

Hope Fermi is better than this

The thing that I find interesting about that, is that @ 8X it's not that the GTX470 is Bandwidth limited (it should still be about 10-15% faster more bandwidth than the HD5870 thanks to the 320bit memory (although 100mhz slower mem clock [supposedly]), nor should it be cache starved either, so it's actually processing power starved at 8X, which could be down to drivers, but since this seems to be their marquee demo tool I doubt it has gone want for optimizations.
It'll be interesting to see what happens with the drop-off. ATi has always had more efficient algos and memory techniques than nV, but it's interesting that that is the one area it dropped off, and it was the one area I would've expected it to pull away further.

It's still very early though, so can't really trust much of anything out there right now.

 

[JAYDEEJOHN]

All Im saying is, I was expecting Fermi to own the single core gpu field, now Im not so sure, and thats NOT good.
So, now, I should have to have your permission to say what I just said?
nVidia NEEDS Fermi to kick some tale, and it should, but theres rumors floating that it wont, and that isnt good, and since Ive actually added some content here, I guess it is relevent whether its gloom or boom, bust or must have

 

[sabot00]

Come on, now your calling a 5870 , "plain jane". Last time I checked it exploded on the market and became the fastest single gpu card by about 20-25%. We know ATI-Nvidia's architecture never match up tick for tock. Can you ever hit the submit button on a post that isn't total anti-doom-gloom Nvidia ?

The 5870 is very "plain Jane" relative to the supposed performance of the GTX 480 & the Unigine benchmarks, so if nVidia is right, then the 5870 would be plain Jane.