[FUD] GT300 is codenamed Fermi

Named after a nuclear reactor


The chip that we ended up calling GT300 has internal codename Fermi. The name might suit it well as Enrico Fermi was the chap that came up with the first nuclear reactor.

The new Nvidia chip is taped out and running, and we know that Nvidia showed it to some important people. The chip should be ready for very late 2009 launch. This GPU will also heavily concentrate on parallel computing and it will have bit elements on chip adjusted for this task. Nvidia plans to earn a lot of money because of that.

The chip supports GDDR5 memory, has billions of transistors and it should be bigger and faster than Radeon HD 5870. The GX2 dual version of card is also in the pipes. It's well worth noting that this is the biggest architectural change since G80 times as Nvidia wanted to add a lot of instructions for better parallel computing.

The gaming part is also going to be fast but we won’t know who will end up faster in DirectX 11 games until we see the games. The clocks will be similar or very close to one we've seen at ATI's DirectX 11 card for both GPU and memory but we still don't know enough about shader count and internal structure to draw any performance conclusions.


Of course, the chip supports DirectX 11 and Open GL 3.1

http://www.fudzilla.com/content/view/15713/1/

So we're looking at high heat output and instability?

Biggest architecture change since G80 sounds good though, sort of the only 'change' since then though (others seem to just be additions)

Hopefully this is an indication of 2009 launch. i.e. stuff starts leaking well before launch. If the net had stayed silent much longer it would have been confirmation of a 2010 card. Still FUD is FUD so not to optimistic

Interesting times ahead.

as long as it's not Chernobyl it's ok

Quote:

---

Originally Posted by rusty

as long as it's not Chernobyl it's ok

---

:rofl:

Quote:

---

Originally Posted by rusty

as long as it's not Chernobyl it's ok

---

Haha, good one!

Quote:

---

Originally Posted by Arkangyl

So we're looking at high heat output and instability?

---

I'd hardly call nuclear fission unstable, 1 failure in 55 years with 100s of reactors isn't exactly a high number.

im fully expecting a NVIDIA limited xmas launch of just one product line at around 400$ ,enough for nV to claim a hard launch title.
a mass nV launch with complete availability of all product lines imo will be coming next year possibly having to compete against an ATI top product line refresh.
the really important thing for the 2 companies is that once again ATI will do better financially and gain further marketshare.
...but yea,nV will have the braggin rights of claiming the "performance crown" once again which in reality is pretty pointless as we have seen with current nV top line and the 4890:shrug:

Quote:

---

Originally Posted by rusty

as long as it's not Chernobyl it's ok

---

Chernobyl was a reactor thats main pupose was to enrich uranium for the military,production of electricity was an afterthought.
Compareable reactor with the same purpose ie mixed military/civilian would be the one in South Carolina which was "officially" closed sometime ago,it was also used for "reenriching" spent nuclear fuel.

Quote:

---

Originally Posted by Helloworld_98

I'd hardly call nuclear fission unstable, 1 failure in 55 years with 100s of reactors isn't exactly a high number.

---

Uh that's what makes it work, instability releases energy. That is high when you consider if you have a meltdown the land is fked for the next 1000+ years.

It sounds like a 5870 with Nvidia branding.

Quote:

---

Originally Posted by Helloworld_98

I'd hardly call nuclear fission unstable, 1 failure in 55 years with 100s of reactors isn't exactly a high number.

---

Three mile island, Windscale, K-19?

Quote:

---

Originally Posted by Helloworld_98

I'd hardly call nuclear fission unstable, 1 failure in 55 years with 100s of reactors isn't exactly a high number.

---

There have been a lot more than ONE failure in 55 years!!!!
The one you are talking about is the biggest known failure in civilian usage...what about Hiroshima and Nagasaki?
What about all the storage facilities in germany with leakages? There's a lot of radiation going into the ground and the atmosphere!
What about Sellafield 1957? Idaho 1961?

Lol those were BOMBS dude, not reactor failures :rofl:

Quote:

---

Originally Posted by trinibwoy

Lol those were BOMBS dude, not reactor failures :rofl:

---

I think those Japanese are still laughing about those bombs :rolleyes:
We were talking about exposition of nuclear radiation - no matter if it were bombs or power plants or some storage facility!
So stop making fun of me!

And yes, I do work for a company who owns more than one powerplant here in germany, and yes, I do have visited a nuclear powerplant several times, so I think I know what I am talking about here... ;)

"has billions of transistors and it should be bigger and faster than Radeon HD 5870. The GX2 dual version of card is also in the pipes."

so this just mean it has high power consumption + cost + heat output... but is going to be damn fast,

surprising there making a gx2 for this "supposedly" massive chip unless is not made from 2 of the top chips and maybe something one step down, think gtx 295 was 2 275's not 285's
we shall see not long to go now it seems :D

Man, that's a toss up right now. Some people are saying we'll know more soon and some people are saying no availability for 6 months. It's a pretty good bet that the truth is somewhere in the middle :)

*sigh* once again another news topic goes wonky and off topic

Do you guys honestly believe NV gt300 will beat a 5870x2 which is right around the corner im asking out of curiosity because the 5870 is a beast.
Nevermind 5870x2 but as the article stated there will be a gx2 version.
Is nvidia smart enough to make it single PCB and not make a dual PCB then a conversion once again like the 295
I hope to see some pricing wars through this competitiong however.
Im all for both companies neck and neck its better for my wallet lol.

It really comes down to how much headroom there is in engineering. Everybody looks at manufacturing process and die size etc, etc. But does anybody ever ask if engineers are making the best use of available transistor budget? If we look at just the usual stuff and assume that both Nvidia and AMD engineers are doing the best they can with available technology, then no it won't beat the X2. If we allow for them finding novel ways to do things then there's a slight possibility.

There just isn't enough technical info available for anybody not under an NDA make any kind of informed guess; sometimes these huge monolithic chips are beasts (G80, G200, R200) other times they are over-hot failures (NV30, R600). There is the rumour from Fuad that it will beat the ATI offering and because that is the only slight hint of a performance comparison available to me [and because it's plausible to think we might get a repeat of the previous generation] I would guess that it might be the case, but it would be a big and not very well founded guess.

Lol, all you guys think of are nukes. Fermi helped kick start atomic knowledge. Work in that field can improve semiconductors as well.

This card has a lot to live up to if they really used that codename. Personally I would like the dual gpu version to be a single pcb, full shader count/speed cores, and come stock with watercooling, lol.

As long as this thing doesnt require the same level of cooling as a nuclear reactor, im good.

Quote:

---

Originally Posted by Solus Corvus

...and come stock with watercooling, lol.

---

mmmmmm +1

Quote:

---

Originally Posted by GrossmeisterB

what about Hiroshima and Nagasaki?

---

its not a failure it is MEANT to explode.

How about we go find someone that is doing nuclear imaging on test circuits if any of his work uses principles discovered by fermi. Sure, nuclear technology has a terrible side. But it has everyday application too. There are nuclear principles operating in your body and in the environment around you right now. What about when you need an X-ray scan or MRI? It's not as if these cards are literally going to have a nuclear meltdown and explode, that's not the first thing you have to think of. Well they could explode, but not like that lol.

I think it takes a lot of balls from Nvidia to use that as a codename if they in fact have. If they think they can live up to the name Fermi, I encourage them to try, lol.

Quote:

---

Originally Posted by Generic user #2

its not a failure it is MEANT to explode.

---

It was the biggest failure ever made ;)

Quote:

---

Originally Posted by GrossmeisterB

It was the biggest failure ever made ;)

---

Definition of failure: an event that does not accomplish its intended purpose

They worked as intended to, so they cannot be a failure. Biggest failure ever made, Hindenburg ranks up there way above nuclear bombs/reactors. That is unless you're a eco-mentalist type.

Interesting choice for a codename though, if it's true. At least it's still able to stir up debate, even if it's hard to tie it down to a meaningful explanation for the choice.

Please nv if you're reading, no sandwich type dual GPU solution again. Let the GTX295v2 be the real-world tryout for a GTX395. What I'm wondering though how you'll be able to cram two 512bit memory buses on that PCB. Would be nice if there where no cut-downs this time, even if it makes the whole thing a 50-100 bucks more expensive due to more layered PCB.

the gtx380 is rumored to be 384 bit... so it would put it at about 1.5 gb of vram... which isn't bad


although... a 3 gb gtx395 or a 4gb 5870x2 would go nicely in my rig, even if it would be bottlenecked

i'm looking forward to the dx11 battles between ati and nvidia :)

i don't have high hopes in gt300 to be my next graphics card though. it'll be a beast, no doubt, but its price will be steep as well - i'm pretty sure about that... i predict the 5870 and gtx380/385whatever to play in different leagues, performance- as well as pricewise :/

"should be faster than 5870"

That doesn't sound very ambitious to me...

So after what happened with 4800 vs gt200, nvidia is still sticking with big chips for their new arc? Can't say if that's a smart move or not. I expected Nvidia to come up with their own small chip like ATi that's very scalable. We'll see.

Quote:

---

Originally Posted by Vit^pr0n

So after what happened with 4800 vs gt200, nvidia is still sticking with big chips for their new arc? Can't say if that's a smart move or not. I expected Nvidia to come up with their own small chip like ATi that's very scalable. We'll see.

---

GPUs take a good ~3-4years of developement, from the start of the design process to product launch. So even if they wanted to change their architecture due to that outcome, it would be atleast another year if not 2-3 until we saw it.

The bombs were intended for destruction. Certainly the collateral damage was not expected--tons of nuclear researchers in the U.S. from the 40s eventually died from long term effects of the radiation. Nobody wants lingering damage from war--one of the reasons pellet-based shotguns are not used in military combat units.

The use of shotguns was brought up after WWI at the Geneva Convention as not being particularly useful in killing, but caused a lot of long-term injuries.

The bombs were intended for destruction. Certainly the collateral damage was not expected--tons of nuclear researchers in the U.S. from the 40s eventually died from long term effects of the radiation. Nobody wants lingering damage from war--one of the reasons pellet-based shotguns are not used in military combat units. The purpose of combat is to kill or force surrender of your opposing, rather than intentionally cause post-war damage.

The use of shotguns was brought up after WWI at the Geneva Convention as not being particularly useful in killing, but caused a lot of long-term injuries. Thus they are no longer wielded in combat.

Quote:

---

Originally Posted by Vit^pr0n

So after what happened with 4800 vs gt200, nvidia is still sticking with big chips for their new arc? Can't say if that's a smart move or not. I expected Nvidia to come up with their own small chip like ATi that's very scalable. We'll see.

---

By 'very scalable' you mean AFR? ;) I wouldn't call 'em 'very scalable', if you're referring to the X2 versions.

I'm glad the companies are doing different strategies, might make for better and more interesting choices.

What's with the nuclear weapons debate? It's just the name of a researcher, get over yourselves.

lol, what will they name the cut down mainstream part? tschernobyl? ^^

Quote:

---

This GPU will also heavily concentrate on parallel computing and it will have bit elements on chip adjusted for this task

---

huh? what does he mean with that?

It sounds very much like "Hey we are releasing the best GPGPU's by far! (oh btw they should be alright in games too)"

They might be competitive with the 5870 performance wise, but i'm sure they won't be nearly as power efficient.

They will either have to whore out the cards at low prices or rely on the goodwill of their regular customers from the past when nvidia was on top.

Quote:

---

Originally Posted by DeathReborn

Definition of failure: an event that does not accomplish its intended purpose

They worked as intended to, so they cannot be a failure. Biggest failure ever made, Hindenburg ranks up there way above nuclear bombs/reactors. That is unless you're a eco-mentalist type.

Interesting choice for a codename though, if it's true. At least it's still able to stir up debate, even if it's hard to tie it down to a meaningful explanation for the choice.

---

In that moment, it worked as intended, but the failures have been made before - in developing such weapons of mass destruction - and the consequences who are still present to the people - 64 years later - and will be present for the next few thousand years!!

I don't blame nVidia for chosing the name, I was just commenting about the guy, who said there has been only one "failure" in using nuclear power - which is by far not true! The developing of nuclear bombs is one of the biggest mistakes ever made on earth! There has been, there are, and there will be a lot of conflicts concerning nuclear weapons.

Beside of the bombs, the usage of nuclear power to produce energy is a thing from the past, it's not safe at all, the storage of the used fuel assemblys is a very very big problem, not only for us, but for the next few thousand years...

And now you have to explain to me, what HINDENBURG has to do with nuclear bombs??

The GTX295 single-PCB was an absolute nightmare with regards to design.. I can't wait to see them attempt to cram two of these onto the same sized PCB. Does anyone know what GT200 was codenamed or is this the first GPU named after something besides an incremental number?

Quote:

---

Originally Posted by All you people talking about stuff not related to GT300

blah blah off topic blah blah

---

Guys, chill out and get back on topic. I'm too tired to deal with deleting half the posts in this thread in order to focus this thread, so just cut it out with nuclear weapons, nuclear reactors, hindenburg/hindenpeter, the latest family guy episode, multiverse and France. I may have thrown in a few things for good measure but you get the point.

A dual chip GT300 should be easier to build than a GTX 295, assuming the rumors are right and the chip is smaller using GDDR5.

Ooops, I guess I was outdated on rumors, I was still thinking of 512bit interface. 384bit should be okay then, I guess.

Quote:

---

Originally Posted by [XC] gomeler

The GTX295 single-PCB was an absolute nightmare with regards to design

---

You know this how?

Quote:

---

Originally Posted by tdream

Uh that's what makes it work, instability releases energy. That is high when you consider if you have a meltdown the land is fked for the next 1000+ years.

---

The instability of the nucleous is what makes it radiate, splitting the atom is not the same thing. A nuclear reactor is stable as long as you are able to absorb the neutrons to prevent the china syndrome.

Quote:

---

Originally Posted by GrossmeisterB

I think those Japanese are still laughing about those bombs :rolleyes:
We were talking about exposition of nuclear radiation - no matter if it were bombs or power plants or some storage facility!
So stop making fun of me!

And yes, I do work for a company who owns more than one powerplant here in germany, and yes, I do have visited a nuclear powerplant several times, so I think I know what I am talking about here... ;)

---

If you don't understand the difference between a run away chain-reaction and very controled fission event then i'd very much say you have no idea what you are talking about.

Quote:

---

Originally Posted by 003

You know this how?

---

he means pcb components layout, not pcb trace design...
single pcb 295 cards are far from perfect for enthusiasts when it comes to component layout, location of pciE power connectors, location of gpus and nvio and nf200 chips, location of pwms, mounting hole locations...

Quote:

---

Originally Posted by astrallite

one of the reasons pellet-based shotguns are not used in military combat units.

The use of shotguns was brought up after WWI at the Geneva Convention as not being particularly useful in killing, but caused a lot of long-term injuries. Thus they are no longer wielded in combat.

---

Really? Is that why the Benelli M4 is the standard issue shotgun for the U.S. Marine Corps? Typically loaded with 00 buckshot. They're carried for close combat, structure clearing, and door breaching.

http://www.spareammo.com/posters/ben...es_440x352.jpg

PS: I own one. :)

Quote:

---

Originally Posted by [XC] gomeler

Guys, chill out and get back on topic. I'm too tired to deal with deleting half the posts in this thread in order to focus this thread, so just cut it out with nuclear weapons, nuclear reactors, hindenburg/hindenpeter, the latest family guy episode, multiverse and France. I may have thrown in a few things for good measure but you get the point.

---

+1
Guys, come on, there is Wampeteers, Foma, & Granfaloons section for this!
Thanks. :up:

And talking about the single PCB GTX395 design... If the chip is really that hot and so big, single PCB card isn't really an option because it'd have to be really really long.
I mean, look at 5870x2. Huge already... And now imagine a board with 2x GT300...

Meh, my 800D can take a long card, if you are gona buy a card like the gtx395 you should not be putting it in a silly little case

Quote:

---

Originally Posted by Bodkin

Meh, my 800D can take a long card, if you are gona buy a card like the gtx395 you should not be putting it in a silly little case

---

I think single PCB 395 can beat the length of B|tchin' fast! 3D 2000, so I wouldn't be so sure. :ROTF:

Quote:

---

Originally Posted by zalbard

+1
Guys, come on, there is Wampeteers, Foma, & Granfaloons section for this!
Thanks. :up:

And talking about the single PCB GTX395 design... If the chip is really that hot and so big, single PCB card isn't really an option because it'd have to be really really long.
I mean, look at 5870x2. Huge already... And now imagine a board with 2x GT300...

---

yeah it seems like nvidia is not done with dual pcb GX2s

lol.. I use to work @ Fermi

Quote:

---

Originally Posted by oohms

It sounds very much like "Hey we are releasing the best GPGPU's by far! (oh btw they should be alright in games too)"

They might be competitive with the 5870 performance wise, but i'm sure they won't be nearly as power efficient.

They will either have to whore out the cards at low prices or rely on the goodwill of their regular customers from the past when nvidia was on top.

---

Some pretty big assumptions there to reach that conclusion!

So when ATi had its flagship selling for ~$150 AMIR, was that considered whoring out?

did anybody else notice this?
the whole news article is written in very nvidia friendly pleasing language, but then theres this:

Quote:

---

The gaming part is also going to be fast but we won’t know who will end up faster in DirectX 11 games

---

which essentially says gt300 might NOT beat 5870? :o
idk if its fudo stumbling over his own words like he does sometimes... or if this is a hint...

its the first time somebody is questioning that gt300 will outperform every other single gpu card for sure.

Quote:

---

Originally Posted by Ozzfest05

Do you guys honestly believe NV gt300 will beat a 5870x2 which is right around the corner

---

Yes.

Personally, having owned a pair of 4850's since they first came out, and seeing nVidia's continual underhanded renaming practices and the sheer bloat of the GTX280, AMD have my money again this time. Sorry, but the 5870 is far too tempting not to take advantage of. I dont care if the GTX gets 110fps vs 90fps in a particular game - it's high enough for me, i buy the cards to play GAMES with, not to be a benchmark king. Heat plays a large part of my decision, as does energy efficiency, as i don't want to be forced into buying ANOTHER power supply, just because my SEVEN HUNDRED WATT one is suddenly deemed to be 'weak', and also lacking in 8-pin PCIe power connectors (just having 2x 6pin).

nVidia may be the king of the brute force hill, but it's taken too many tradeoffs to get there. They've sacrificed practicality (heat, power consumption, noise, size & affordability) for a win. AMD's approach is far better thought out, in my opinion.

I'm not a fanboy however, if the GT300 comes out and surprises me, then i'll consider it - but i wont buy a card that costs the earth just because 'IT ARE TEH FASTEST!!!!1111ELEVENTY!!11'...

Quote:

---

Originally Posted by Kai Robinson

... noise, size

---

Please don't put noise as a sacrifice for using nvidia; ever.

and wtf size? they are almost the same.

nVidia's history is replete with such examples...does the FX5800 ring any bells? I'm not singling out the 5870 vs the GTX280, but the companies approach as a whole. nVidia will do anything it takes to win, regardless of the sacrifices it makes. AMD's approach is more considered.

Why *shouldn't* i put down noise as a consideration? I leave my rig up 24/7 and sleep in the same room, i dont want a loud ass fan whirring away. My 4850's are almost silent. The last nVidia card i had sounded like a 747 at takeoff almost all the time, and could still barely maintain 50*c idle temps.

Quote:

---

Originally Posted by Kai Robinson

They've sacrificed practicality (heat, power consumption, noise, size & affordability) for a win.

---

Ha. The 4890 is noisier than the GTX285 and uses more power at idle. At load the 285 uses just slightly more power, but it's within the range of 10W, so negligible.

Quote:

---

Originally Posted by Kai Robinson

nVidia's history is replete with such examples...does the FX5800 ring any bells? I'm not singling out the 5870 vs the GTX280, but the companies approach as a whole. nVidia will do anything it takes to win, regardless of the sacrifices it makes. AMD's approach is more considered.

---

I've always gotten a laugh out of this whole theory that the "big" company is always evil and will do anything to win... (nVidia, Intel, Microsoft, etc)... and somehow the little guy... (ATi, AMD, etc) is more noble, honest, and has a far better strategy.

Quote:

---

Originally Posted by Andrew LB

I've always gotten a laugh out of this whole theory that the "big" company is always evil and will do anything to win... (nVidia, Intel, Microsoft, etc)... and somehow the little guy... (ATi, AMD, etc) is more noble, honest, and has a far better strategy.

---

Someone invoke Godwin's law :P

Quote:

---

Originally Posted by Kai Robinson

nVidia's history is replete with such examples...does the FX5800 ring any bells? I'm not singling out the 5870 vs the GTX280, but the companies approach as a whole. nVidia will do anything it takes to win, regardless of the sacrifices it makes. AMD's approach is more considered.

Why *shouldn't* i put down noise as a consideration? I leave my rig up 24/7 and sleep in the same room, i dont want a loud ass fan whirring away. My 4850's are almost silent. The last nVidia card i had sounded like a 747 at takeoff almost all the time, and could still barely maintain 50*c idle temps.

---

What are you talking about? Have you even seen the HD4870/4890s in action? They use more power and are nosier. :rolleyes:

Not to mention HD4000 series fails Furmark so you can be sure their build quality is inferior.

Quote:

---

Originally Posted by Clairvoyant129

What are you talking about? Have you even seen the HD4870/4890s in action? They use more power and are nosier. :rolleyes:

Not to mention HD4000 series fails Furmark so you can be sure their build quality is inferior.

---

Dont get me started about my bloody 4870x2, was like having a space heater with a jam in it. Made my dyson sound wispier quite. I hate all reference cooling (8800gt is the exception) but 48xx relly sucked in this regard. I dont know what the 58xx reference cooler is like but I doubt it is much better

I mentioned a 4850....and you came up with a 4890 being noisier than a GTX285....

How also, is a stupidly unrealistic 'benchmark' that bears NO relation to real world usage indicative of 'poor quality'? Go read Anandtech's review, and find out why furmark is a steaming pile of useless doo-doo.

I've never stated that nVidia is the evil company either, nor that AMD is some plucky underdog. I've had cards from both companies over the years, but all i've seen lately out of nVidia, is renaming G92's over and over, the top of the line products being bloated and overpriced, noisy (having HEARD a GTX285, i know it's a card i'd never want to buy for that reason alone). AMD's offerings since the 4800 series was introduced have been far more appealing to me (price v performance being the prime reason), hence my decision to choose the 5870 this time around, and stick with the AMD camp.

All we've heard for ages is how Jen-Hsun Huang is going to 'open a can of whoop-ass', and yet, i've seen no can's being opened, just refreshes of 18-20 month old products with new badges. I know i'm not the only person to be thinking that nVidia has dropped the ball again. They did it with the FX lineup (vs the 9x00 series), and ATi did it with the X1800 and the 2x00 series.

Bodkin - if you bothered to READ any reviews of the 5870 - you'd see that actually there's nothing wrong with the reference design as far as cooling goes. The only criticism i have of my 4850's, is that the heat is dumped back into the case, rather than vented out the back (just like the 8800GT i had before it), but considering i run an open case, it's a moot point. And a 4870X2 WILL be hot - its two GPU's on one PCB ffs. The GTX295, in comparison, isn't going to be magically cooler just because it comes from the nVidia camp.

Quote:

---

Originally Posted by Kai Robinson

Bodkin - if you bothered to READ any reviews of the 5870 - you'd see that actually there's nothing wrong with the reference design as far as cooling goes. The only criticism i have of my 4850's, is that the heat is dumped back into the case, rather than vented out the back (just like the 8800GT i had before it), but considering i run an open case, it's a moot point. And a 4870X2 WILL be hot - its two GPU's on one PCB ffs. The GTX295, in comparison, isn't going to be magically cooler just because it comes from the nVidia camp.

---

I did say I had no experience with 58xx so fair play to you, but I think you are wrong about gtx285 being loud compared to the 48xx

I have TWO 4850's putting out a mere 39dB of noise between them, currently sat at idle. I'll fire up TF2 later on tomorrow and measure them under load.

Quote:

---

Originally Posted by Kai Robinson

I have TWO 4850's putting out a mere 39dB of noise between them, currently sat at idle. I'll fire up TF2 later on tomorrow and measure them under load.

---

Cool, and I will fire up batman and enjoy the AA (runs away :P)


Im only kidding. Like I say I was just burnt with the 4870x2, whatever setup works best for you is the one you should run :up: No hard feelings?

Play away chum, i never really liked superheroes :p:

So is there going to be a dual GPU GT300 card? Hell, I hope not. If NVIDIA achieves to revert the situation from the last generation (and I hope they do, otherwise, bad for them and bad for everybody) their >500mm^2 giganormous chip should be direct competitor to the HD5870 X2 (the higher end uncut version, of course). And if that happens, then a dual card with >500mm^2 chips has no point. Last generation they released one because the single chips from both companies were too close despite of their different goals so they needed it to compete with the dual solution of ATi. Let's hope that this time around the same thing doesn't happen, because that means huge losses to NVIDIA (repricing and all). And that is not going to affect the consumer on the short term, but probably will in the long term.

Quote:

---

Originally Posted by Kai Robinson

Personally, having owned a pair of 4850's since they first came out, and seeing nVidia's continual underhanded renaming practices and the sheer bloat of the GTX280, AMD have my money again this time. Sorry, but the 5870 is far too tempting not to take advantage of. I dont care if the GTX gets 110fps vs 90fps in a particular game - it's high enough for me, i buy the cards to play GAMES with, not to be a benchmark king. Heat plays a large part of my decision, as does energy efficiency, as i don't want to be forced into buying ANOTHER power supply, just because my SEVEN HUNDRED WATT one is suddenly deemed to be 'weak', and also lacking in 8-pin PCIe power connectors (just having 2x 6pin).

nVidia may be the king of the brute force hill, but it's taken too many tradeoffs to get there. They've sacrificed practicality (heat, power consumption, noise, size & affordability) for a win. AMD's approach is far better thought out, in my opinion.

I'm not a fanboy however, if the GT300 comes out and surprises me, then i'll consider it - but i wont buy a card that costs the earth just because 'IT ARE TEH FASTEST!!!!1111ELEVENTY!!11'...

---

Quote:

---

Originally Posted by Kai Robinson

nVidia's history is replete with such examples...does the FX5800 ring any bells? I'm not singling out the 5870 vs the GTX280, but the companies approach as a whole. nVidia will do anything it takes to win, regardless of the sacrifices it makes. AMD's approach is more considered.

Why *shouldn't* i put down noise as a consideration? I leave my rig up 24/7 and sleep in the same room, i dont want a loud ass fan whirring away. My 4850's are almost silent. The last nVidia card i had sounded like a 747 at takeoff almost all the time, and could still barely maintain 50*c idle temps.

---

Quote:

---

Originally Posted by Kai Robinson

I mentioned a 4850....and you came up with a 4890 being noisier than a GTX285....

How also, is a stupidly unrealistic 'benchmark' that bears NO relation to real world usage indicative of 'poor quality'? Go read Anandtech's review, and find out why furmark is a steaming pile of useless doo-doo.

I've never stated that nVidia is the evil company either, nor that AMD is some plucky underdog. I've had cards from both companies over the years, but all i've seen lately out of nVidia, is renaming G92's over and over, the top of the line products being bloated and overpriced, noisy (having HEARD a GTX285, i know it's a card i'd never want to buy for that reason alone). AMD's offerings since the 4800 series was introduced have been far more appealing to me (price v performance being the prime reason), hence my decision to choose the 5870 this time around, and stick with the AMD camp.

All we've heard for ages is how Jen-Hsun Huang is going to 'open a can of whoop-ass', and yet, i've seen no can's being opened, just refreshes of 18-20 month old products with new badges. I know i'm not the only person to be thinking that nVidia has dropped the ball again. They did it with the FX lineup (vs the 9x00 series), and ATi did it with the X1800 and the 2x00 series.

Bodkin - if you bothered to READ any reviews of the 5870 - you'd see that actually there's nothing wrong with the reference design as far as cooling goes. The only criticism i have of my 4850's, is that the heat is dumped back into the case, rather than vented out the back (just like the 8800GT i had before it), but considering i run an open case, it's a moot point. And a 4870X2 WILL be hot - its two GPU's on one PCB ffs. The GTX295, in comparison, isn't going to be magically cooler just because it comes from the nVidia camp.

---

I don't know what to make of this?

You are buying the 5870 because its quiet and your basing this claim off of your ownership of a 4850 crossfire. Yet when people tell you a 4870 and 4890 are load, you try to say your claim is based on your ownership of a 4850. Yet you are buying a 5870 which is the replacement for the 4870/4890. Don't you think you should disregard your own evidence since the 4870/4890 are more similar to the 5870.

Also your basing the loudness of NV cards in general based on the fx 5800 which is a 6 plus year old videocard?

[Personal insults are not allowed on the XS Forums. - Mods] Seriously you have to see your reasoning why AMD cards are quiet is pretty bad and evidence against NV for making loud cards is pretty bad.

If you look at anandtech review.
http://www.anandtech.com/video/showdoc.aspx?i=3643&p=26

A single 5870 is louder than SLI gtx 285. Also you mention your last card NV card was a 8800gt and you said it sounded like a 747 24/7. Hmm, look at those charts, according to them the 4850 is significantly louder than the 8800gt. Even taking into account experimental variation, I don't see how your first hand accounts can have the 8800gt sounding like a 747, while you can't hear your 4850 in crossfire. Hmmm, are you lying for the sake of your own argument?

Sure the idle of 5870 is quiet, but pretty much every companies videocard is quiet.

Even if your not an AMD fanboy as you claim, its obvious you have bias against NV, which is really no better in making a good argument.

Also don't you think your making too many assumption about the performance, price, power usage and loudness of card not out yet?(the gtx 380).

jeez gpu release season just brings out the worst in all of us doesn't it? :down:

i expect the fermi core to have equally inovative power saving features to the hd5xxx series. power/performance ratios for the gt200b were awesome compared to the rv770 (gtx295 vs hd4870x2). i wouldn't be supprised to see fermi using power gating or serious down clocking (like evergreen) to achieve further energy savings.

Quote:

---

Originally Posted by Kai Robinson

I mentioned a 4850....and you came up with a 4890 being noisier than a GTX285.... blah blah blah.

---

I'm not even going to bother since your argument is obviously flawed and retarded.

Maybe you should take a look at tajoh's post,

Quote:

---

Originally Posted by tajoh111

I don't know what to make of this?

You are buying the 5870 because its quiet and your basing this claim off of your ownership of a 4850 crossfire. Yet when people tell you a 4870 and 4890 are load, you try to say your claim is based on your ownership of a 4850. Yet you are buying a 5870 which is the replacement for the 4870/4890. Don't you think you should disregard your own evidence since the 4870/4890 are more similar to the 5870.

Also your basing the loudness of NV cards in general based on the fx 5800 which is a 6 plus year old videocard?

<snipped out flame> Seriously you have to see your reasoning why AMD cards are quiet is pretty bad and evidence against NV for making loud cards is pretty bad.

If you look at anandtech review.
http://www.anandtech.com/video/showdoc.aspx?i=3643&p=26

A single 5870 is louder than SLI gtx 285. Also you mention your last card NV card was a 8800gt and you said it sounded like a 747 24/7. Hmm, look at those charts, according to them the 4850 is significantly louder than the 8800gt. Even taking into account experimental variation, I don't see how your first hand accounts can have the 8800gt sounding like a 747, while you can't hear your 4850 in crossfire. Hmmm, are you lying for the sake of your own argument?

Sure the idle of 5870 is quiet, but pretty much every companies videocard is quiet.

Even if your not an AMD fanboy as you claim, its obvious you have bias against NV, which is really no better in making a good argument.

Also don't you think your making too many assumption about the performance, price, power usage and loudness of card not out yet?(the gtx 380).

---

+1

Let's all take a 10 minute break and regroup here guys. We do have a Zero Tolerance policy on the news section ATM so I would advise you that if you even just have a questionably toned post you give a serious rethink before hitting "submit".

Other mods aren't nearly as lenient and willing to just publicly embarrass people by modding posts as I am (and you don't even want to know what the admins will do to you).

Just to provide some first hand info I've used SLI / CF 4850/70s and 260/280s in my machine and noise / power wise, the Core 216s took the cake in both single and dual configs. The 280 is louder but still reasonable. The 4870 is a tad louder than the 260s but not too much (somewhere in between the 260/280 at load ). I haven't used a 4890 in my machine personally but I've toyed around with a friends and they are the loudest single gpu dual slot at the moment ( all of this information is based on my experience using these products with reference designs on auto fan profiles and clock speeds )

I've personally been more impressed with Nvidia's reference design over the last 5 years more so than ATI ( and until the 8800gtx, I exclusively used ATI cards from the 9700-x1900xt ) I am glad that ATI did finally use a solid sink on the 5870 ( longer pcb and backplate ect ) I know many people really liked the fact that the 4870s were short in comparison to the GTX200s ( and for the average case I'd agree as well ; when I first installed the 260s first thing I thought was WHOA and this isn't even the high end card :p ) but I'm all for more surface area for a beefier HS/F. Please respect all of the above as the subjective opinion of another user and nothing more. Just thought I could help add some first hand experience to the discussion.

As far as what I'm expecting for GT300 my guess is it won't beat Cypress at performance per watt but I still expect it to be a fair bit faster on average (%10-25 single gpu) so those willing to shell out the extra money for the sake of performance if true, they'd be the card to get but performance for dollar, will remain in the ATI camp.

Quote:

---

Originally Posted by Serra

Let's all take a 10 minute break and regroup here guys. We do have a Zero Tolerance policy on the news section ATM so I would advise you that if you even just have a questionably toned post you give a serious rethink before hitting "submit".

Other mods aren't nearly as lenient and willing to just publicly embarrass people by modding posts as I am (and you don't even want to know what the admins will do to you).

---

Tarring and Feathering should be instituted as a punishment on the XS News forum.

I see the personal insults have started flying.

I'm an AMD fanboy am i?

Well that would explain why i've owned a GeForce 3 Ti, GeForce 6600GT, GeForce 7600GT and a GeForce 8800GT then, would it?

My basis for the 5870 choice at the moment, is 1) performance far better than both of my 4850's with a single gpu solution (and therefore fewer driver issues) 2) driver performance is due to improve 3) the price is appealing

Currently, idle power is far lower than any current offering than i can see which can only translate into lower temperatures. As for my 4850CF setup generating noise, they're both sat at 56-58*c and 35% fan speed. They're pretty quiet and definitely quieter than my evga 8800GT 512mb, which, at idle, would sound loud enough to irritate my ears.

If you *read* my posts, you'd see that i choose cards on more than just the sticker basis. And as for calling me a retard - who the hell do you think you are?

Quote:

---

Originally Posted by Kai Robinson

They're pretty quiet and definitely quieter than my evga 8800GT 512mb, which, at idle, would sound loud enough to irritate my ears.

---

If it's a reference cooler, I don't know what you are talking about. The 8800GT cannot be heard at around 30% and it never goes above 49% at load. (at least mine don't.)

Yeah, my refernece 8800gt were great, no noice noticible over watercooling pump

If one guy tells you his foot has a knife in it and it bleeds, are you telling him he dont have a knife in his foot and it dosnt bleed?
I love videocards, especially new ones, and they all even when same manufacture are different, some fans has pitching noise, some people have really good hearing for subsonic sounds.
Its basically impossible to use ones own preference to a card and how it affect people around it, kai tells people his 8800GT sounds worse than his 4850 crossfire, I rather belive him any day than people who dont sleep at his house and room.
;)

didnt a lot of 8800GT broke down due to cheap third manufacture parts?

Quote:

---

Originally Posted by Kai Robinson

I'm an AMD fanboy am i?
Well that would explain why i've owned a GeForce 3 Ti, GeForce 6600GT, GeForce 7600GT and a GeForce 8800GT then, would it?

---

ahhh come on admit it, you only bought those to trash them and rma them over and over, trying to hurt nvidia as much as you can ^^

seriously, can we get back on topic?
what does 8800s fanspeed and noise have to do with gt300?
i dont see any connection at all :p:

so the latest launch date rumor i heard is november 27?
that means 58 days to go for gt300... supposedly... :D
yet still no hint of a leak...

November 27, sounds like they will try and get at lest one batch out for Christmas. Any source on that Saaya?

Fuad claims Jen Hsun will discuss Fermi at today's keynote at 1pm. Maybe even give a demo.

http://www.fudzilla.com/content/view/15741/1/

Quote:

---

We expect that Nvidia CEO Jensen Huang will demonstrate the card at the GTC keynote that should start at 1.00 PM Pacific time or roughly 10 PM Central European time.

---

Theo posts specs: http://www.brightsideofnews.com/news...6gb-gddr5.aspx

Quote:

---

3.0 billion transistors
40nm TSMC
384-bit memory interface
512 shader cores [renamed into CUDA Cores]
32 CUDA cores per Shader Cluster
1MB L1 cache memory [divided into 16KB Cache - Shared Memory]
768KB L2 unified cache memory
Up to 6GB GDDR5 memory
Half Speed IEEE 754 Double Precision

---

AFAIK:

Code:

---

512 "SPs"
384 Bit
40nm
GDDR5
~245Gbps
Lots of architectural changes
Native C++ execution
Codename: GF100

---

Quote:

---

Originally Posted by trinibwoy

Theo posts specs: http://www.brightsideofnews.com/news...6gb-gddr5.aspx

---

How I wish this was accurate! :p: There's much sense in going into such direction at a point like this though, as it really seems the traditional GPU won't live that many more years in it's current state and NVIDIA might start preparing towards a more hybrid CPU/GPU solution now. Of course the first generations won't be very fast excecuting CPU tasks and still get easily dominated by highend CPUs but it would be a start.

http://www.anandtech.com/video/showdoc.aspx?i=3651

Love how Nvidia's engineers reply to the reason for the set-back, "because designing GPUs this big is ":banana::banana::banana::banana:ing hard". Wonder if they'll be heat issues like the 280GTX.

Quote:

---

Originally Posted by BenchZowner

Native C++ execution

---

What's this supposed to mean? C++ is a compiled language not an interpreted language. So are they supplying a C++ compiler that makes native G300 binary code or developed a way to interpret uncompiled C++ code on the fly?

Then timing is just as valid, because while Fermi currently exists on paper, it's not a product yet. Fermi is late. Clock speeds, configurations and price points have yet to be finalized. NVIDIA just recently got working chips back and it's going to be at least two months before I see the first samples. Widespread availability won't be until at least Q1 2010.

I asked two people at NVIDIA why Fermi is late; NVIDIA's VP of Product Marketing, Ujesh Desai and NVIDIA's VP of GPU Engineering, Jonah Alben. Ujesh responded: because designing GPUs this big is ":banana::banana::banana::banana:ing hard".

http://www.anandtech.com/video/showdoc.aspx?i=3651

EDIT: sorry already posted

Quote:

---

Originally Posted by BenchZowner

AFAIK:

Code:

---

512 "SPs"
384 Bit
40nm
GDDR5
~245Gbps
Lots of architectural changes
Native C++ execution
Codename: GF100

---

---

http://www.nvidia.com/content/PDF/fe...Whitepaper.pdf

Quote:

---

Originally Posted by Solus Corvus

What's this supposed to mean? C++ is a compiled language not an interpreted language. So are they supplying a C++ compiler that makes native G300 binary code or developed a way to interpret uncompiled C++ code on the fly?

---

http://www.nvidia.com/content/PDF/fe...Whitepaper.pdf

Quote:

---

Originally Posted by BenchZowner

http://www.nvidia.com/content/PDF/fe...Whitepaper.pdf

---

Thanks. That clears it up.

Quote:

---

Originally Posted by Solus Corvus

What's this supposed to mean? C++ is a compiled language not an interpreted language. So are they supplying a C++ compiler that makes native G300 binary code or developed a way to interpret uncompiled C++ code on the fly?

---

Yep, I was wondering about exactly the same thing...

EDIT: Hell, I didn't see the BenchZowner's link. I'll read it now...

RE-EDIT: So a few changes to the compute instruction set to allow compiling programs in C++ to that IS.

http://www.hwmania.org/gallery/file.php?n=1133&w=l

HwMania - Visualizza messaggio singolo - [Thread Ufficiale] nVIDIA Fermi