More bad news rumors for Fermi

[Hauk]

Damn if Charlie's not looking less wacky by the day. Shaders, clock speeds, supply, all still very hot buzz, and now more delays?

The more days that tick by, the more uncomfortable I am about even considering Fermi. The more rumors that fly, the more I believe nV has dropped the ball. Squash the rumors (if you have the ability)?

ATI, why hold back on the 2GB 5870? You've had it ready since November right? Man I hate this crap..

 

[Genx87]

Settle down, it is just a piece of hardware. When Fermi is finally released the sun will rise in the east and settle in the west.

 

[tviceman]

So for an extra 1 billion transistors the performance we get vs a 5870 is.....wait for it....about the same. I'll reserve judgement till launch day but it's not looking good for Nv.

So you buy your video cards based on how many frames per second per transistor it can provide? Personally I don't care if fermi has 8 transistors or 15 billion so long as it performs as well or better than other products within it's price range.

 

[cbn]

So for an extra 1 billion transistors the performance we get vs a 5870 is.....wait for it....about the same. I'll reserve judgement till launch day but it's not looking good for Nv.

I look at the xtor/gaming ratio this way: If Nvidia is able to make profitable HPC cards and sell the same product as a "gaming card" I am happy provided this money saving strategy results in lower prices to the consumer.

 

[SlowSpyder]

So you buy your video cards based on how many frames per second per transistor it can provide? Personally I don't care if fermi has 8 transistors or 15 billion so long as it performs as well or better than other products within it's price range.

I don't think that's what he was getting at. Fermi is coming 6 months later, has many more xtors, so a lot of people were probably waiting to see what kind of performance it would offer. Most expected it to be faster by a good amount than the 5870. That would give consumers who held off some more choices, a better performing Fermi based part, and hopefully a cheaper AMD part that still performs well. Instead, if you've been waiting all this time, it looks like you're going to end up disappointed that you've missed out for the last ~6 months.

The Nvidia guys (one in perticular) had mentioned how it makes sense to wait for Fermi. Now it's looking more like waiting for Fermi may have been little more than a waste of time.

 

[Lonyo]

I don't think that's what he was getting at. Fermi is coming 6 months later, has many more xtors, so a lot of people were probably waiting to see what kind of performance it would offer. Most expected it to be faster by a good amount than the 5870. That would give consumers who held off some more choices, a better performing Fermi based part, and hopefully a cheaper AMD part that still performs well. Instead, if you've been waiting all this time, it looks like you're going to end up disappointed that you've missed out for the last ~6 months.

The Nvidia guys (one in perticular) had mentioned how it makes sense to wait for Fermi. Now it's looking more like waiting for Fermi may have been little more than a waste of time.

There's only one way the wait will be worth it, and that's if AMD prices drop to below initial MSRP as it was 6 months ago.
If a 6 month wait ends up with products which are just as fast as they were 6 months ago, and at the same price or higher, then it truly will have been a waste of time 100%. If prices end up somewhat lower, the wait won't have been quite so terrible, but will by no means have been worth it, except for the die hard NV fanboys who will finally have a new product.

It's all about pricing, and with 10 days until launch, it's still an unknown for both parties.

 

[OCGuy]

Why are some people talking like they 100% know how the performance stacks up?

 

[Lonyo]

Why are some people talking like they 100% know how the performance stacks up?

Well, we already have the HD5970, so performance will be lower than the top AMD/ATI card, so... it doesn't really matter exactly where performance ends up, it might/should be faster than some AMD cards and will be typically slower than other cards, that means it's all about how these cards impact on pricing.

Unless you are going to tell me that you have some evidence of the GTX480 beating the HD5970. Because from all the rumours and leaks, that's never looked likely.

 

[OCGuy]

Well, we already have the HD5970, so performance will be lower than the top AMD/ATI card, so... it doesn't really matter exactly where performance ends up, it might/should be faster than some AMD cards and will be typically slower than other cards, that means it's all about how these cards impact on pricing.

Unless you are going to tell me that you have some evidence of the GTX480 beating the HD5970. Because from all the rumours and leaks, that's never looked likely.

480 does not compete with 5970. 5970 will be more expensive, is hard to find, and you need a monster case to fit it.

Besides, it is a dual-GPU solution. The same reason you would recommend the 5870 over the 295, is the same reason the 480 will be much more attractive than the 5970.

The only reason I am even considering 480, even though I wont gain that much performance likely in games that scale well, I am over the dual-GPU thing for now. If I can get it down to 1 card I will be happy.

 

[SolMiester]

nvidia knows ati is kicking there butt , this is what the hole delay is about ,they know that there two cards can not compete with ati ,,
come on would'nt you be advertising you kick arse product getting people interested in it ,
so why arent they because its crap ,why would'nt you put some reviews out there so people can go wow man look at this thing go ,ive got to have one of these ,
you no why they have'nt done any of this ,why why why ,because its
crap crap crap ,,,,
just my view, i ordered my ati gigabyte 5970 card yesterday being monday here in Australia


the best way to understand what they are doing and why is to ask yourself what would you do ,,think about it

Didnt see leaked cypress benchies before launch, stands to reason NV wouldnt either.

Good luck with profiles on the 5970

 

[cbn]

Fermi is coming 6 months later, has many more xtors, so a lot of people were probably waiting to see what kind of performance it would offer.

I think a lot of people prefer a larger single GPU over two smaller GPUs in Crossfire for gaming purposes.

But what if Fermi isn't that much faster than Cypress? (Due to high double precision xtor count)

Will Nvidia abandon "Big Die" approaches in the future?

Is there an HPC equivalent to "microstutter"? If not, why not just create all Future HPC cards from the ground up as 300 watt TDP dual GPU models?

 

[Schmide]

Good luck with profiles on the 5970

They work as of 10.2, next.

 

[SlowSpyder]

Why are some people talking like they 100% know how the performance stacks up?

It should be fairly obvious that any one speaking about the performance of Fermi are basing what they say on rumors floating around the net. Nvidia hasn't exactly been forthcoming with performance or pricing data (or other information, such as TDP and power consumption) so when a rumor is seen repeated on various news/rumor sites a few times over it tends to gain some traction on the enthusiast forums.

Obviously none of us know exactly what to expect with Fermi. If the limited bits of info/rumors we have are wrong, well Nvidia could have at any time released some proper information if they chose too, but have not. So it should really go without saying that any conversation about Fermi's price/performance are based on rumor and not 100% fact.

 

[SlowSpyder]

Didnt see leaked cypress benchies before launch, stands to reason NV wouldnt either.

Good luck with profiles on the 5970

Completely different situations in my opinion. AMD risked cannibalizing their 4xxx series high end parts by leaking Cypress information. Nvidia has very little high end parts on the market to worry about cannibalizing... the GTX2xx parts are all EOL. Nvidia has everything to gain by building momentum I would think. By releasing Fermi information, assuming it was faster, they would probably stop a number of people from buying high end AMD parts.

 

[Puffnstuff]

I personally hope fermi owns all but that remains to be seen.

 

[OCGuy]

Completely different situations in my opinion. AMD risked cannibalizing their 4xxx series high end parts by leaking Cypress information. Nvidia has very little high end parts on the market to worry about cannibalizing... the GTX2xx parts are all EOL. Nvidia has everything to gain by building momentum I would think. By releasing Fermi information, assuming it was faster, they would probably stop a number of people from buying high end AMD parts.

I personally know 2 people this week that have purchased OEM machines with a GTX260 and a 275. They are still on newegg.......

EOL doesnt mean they arent on the market. IIRC nV had a great Q4 partly because of 260 sales.

We have to remember 5XXX and 4XXX are an extremely small part of the market, even though it is all the people on forums care about.

 

[SHAQ]

With all the screwups that have happened so far I don't see how it could be much faster than ATI's offerings. Nvidia is late with its product and they are trying hard to prevent leaks, yields are bad, your shader target is missed etc etc. Now we hear clocks still aren't finalized. That doesn't add up too well. By just being late and preventing leaks it says to me they screwed up. If Nvidia wants people to wait and buy their product they would show us benchmarks.

Now having said that I am still buying 2 470's if they are under $349.

 

[GaiaHunter]

I personally know 2 people this week that have purchased OEM machines with a GTX260 and a 275. They are still on newegg.......

EOL doesnt mean they arent on the market. IIRC nV had a great Q4 partly because of 260 sales.

We have to remember 5XXX and 4XXX are an extremely small part of the market, even though it is all the people on forums care about.

Because the people that are on the market for the GtX260/275 are the same that will be on the market for the GTX470/480?

Or is the GTX470/480 the same as the 58xx and 5970 market?

 

[cbn]

deleted

 

[SolMiester]

GTX470 pre order here in NZ is $720NZD, which works out at around $500USD!...I know we pay premium over here and freight etc, but $200 USD over estimated prices?!

 

[SlowSpyder]

GTX470 pre order here in NZ is $720NZD, which works out at around $500USD!...I know we pay premium over here and freight etc, but $200 USD over estimated prices?!

I wonder if $500USD is the correct price that it'll sell at when released, or if it's just insanely priced right now?

 

[Keysplayr]

Wow, that's up there. Might be a gouge.

 

[Hauk]

I wonder if $500USD is the correct price that it'll sell at when released, or if it's just insanely priced right now?

That would mesh with initial rumors, and would put 480 in the $600+ range?
Or (hopefully), it's an inflated price. 480 in that price range sounds much better..

 

[tviceman]

I don't think that's what he was getting at. Fermi is coming 6 months later, has many more xtors, so a lot of people were probably waiting to see what kind of performance it would offer. Most expected it to be faster by a good amount than the 5870. That would give consumers who held off some more choices, a better performing Fermi based part, and hopefully a cheaper AMD part that still performs well. Instead, if you've been waiting all this time, it looks like you're going to end up disappointed that you've missed out for the last ~6 months.

The Nvidia guys (one in perticular) had mentioned how it makes sense to wait for Fermi. Now it's looking more like waiting for Fermi may have been little more than a waste of time.

I agree with your reasoning - it is late, it is big, it was expected to perform amazing, and so all of this is adding up to a possible disappointment. Trust me, I'm right there. But I also try to keep things in perspective. A disappointment doesn't outright mean it sucks. Although it could still suck. I'm just waiting for a full, honest, unbiased evaluation.

Theoretically speaking,

If the gtx470 trades blows (even losing more often than winning) with the hd5870 in DX10 but in DX11 outclasses the hd5870 by 10-20% then is Fermi considered "good" or "worth the wait"?

Honestly - except for Crysis running at 1920x1080 or higher with 4X AA or greater, just about every other DX9/10 can run completely maxed out with AA and get incredibly smooth frame rates. I'm more interested in performance of just released/future titles that will be utilizing the new features these graphics cards can deliver.

IMO, if the gtx470 is better vs. 5870 in DX11 games, then Fermi is a success. I'll be most interested in seeing the benchmarks of Stalker Pripyat, the new Metro 2033, AvP, Battlefield Bad Company 2, Dirt 2, and perhaps Battleforge as well.

 

[Apocalypse23]

One thing I know for sure, and I'll take the heat for saying this:

Metro2033 will play better on Fermi.
Crysis will play better on Fermi.