NVIDIA Kepler speculation thread

[gongo]

GK104 replaces GF114 in the NV line-up- 299$ perfectly matches a GF114 2GB at launch. Also consider the timeframe. If we believe that GK104 was planed for early Febrauary, but is now delayed until late March, then GK110 could still beready for late April or ealry May. No point in launching the GK104 @ 500$, when the real 500$ chip is right around the corner.

AMD should have no problems at all. Tahiti is just way too expensive for a performance chip at the moment.

Yes..that is what i think too...GTX 460 launched at $229...out matches the GTX 280 in many cases with higher CC specs...GTX 560 launch at $249..beats a GTX 460 on the same process die...if overclocked...on average beats the similarly overclocked HD6950 at 1080p (thanks to TWIMTP games) so a GTX 660 at $299 is not far off...if it has 1.5x the CC of 560..it should be more than a match for GTX 580..expect to see more tessellation hardware rather than tweak caching too...it should overclock just as well as HD7950...meaning i expect it to come toe to toe with HD7970 at 1080p...

You see why i am disappointed at HD7970 pricing....yes it is business but i think it is overpriced at $549 at 30% faster than GTX 580...at that kind of pricing..i expect it to be 60% faster...AMD small die strategy works hand in hand with the kind of pricing for HD4xxx to 6xxx but at $549 it is a little hard to get excited with their new line....

 

[A1xLLcqAgt0qc2RyMz0y]

Hey Arty - bit sensitive are we?

This thread titled: Kepler speculation thread and has been derailed big time.

My post is about the Kepler GK104 & GK110 and is a response to a post on this forum about why nVidia would price the GK104 at $299.

If the performance for the GK104 is as Charlie implies then with the price of $299 as Charlie states AMD will absolutely be hurt as they will have to reduce the price of the 7970 by at least $250 to compete.

If you have a problem with Charlie's statements then take it up with him.

And BTW your attack post is very childish.

 

[Arty]

And BTW your attack post is very childish.

Its not an attack.

Funny how sensitive you are when you get called out. The issue here is that Kepler info is little and hard to come by, so there is occasional tangential discussion but it is still relevant, from a competition/market perspective. So take deep breaths and relax before you start panicking the next time.

 

[Deleted member 13524]

It's downright funny to see people predicting AMD's "Doom of 2012" based on a couple of semiaccurate posts and without knowing what products+prices AMD can offer in mid-2012 within reasonable profit margins.

 

[rpg.314]

In what way is it a doom, precisely?

Even if we were to assume that Semiaccurate is completely accurate this time, there is no indication that GK104 is < 300mm2 so far, so die costs are pretty close. 7970 has 4 more RAM chips, but then that's that. There's a lot of overclocking head room on Tahiti. The BOM differences are much larger in perception and not so much in reality, I'd say.

 

[trinibwoy]

Why is BOM parity with GK104 important? AMD has much bigger things to worry about if it comes anywhere close to the 7970 performance wise.

 

[boxleitnerb]

Like what? Lower the price by 200$ and be done with it.

 

[AlphaWolf]

if the BOM is lower it's profitable at a lower price, meaning being the top performing part isn't important.

 

[no-X]

Why is BOM parity with GK104 important? AMD has much bigger things to worry about if it comes anywhere close to the 7970 performance wise.

AMD can release 1,5GB model anytime and boost price/performance ratio at least by 10-15 % without decreasing their margins. They can release ~20% faster 1100MHz model while keeping current manufacturing costs. They are in a very good position.

Current rumours are in fact saying, that Nvidia's part will be 3 months late and not even as fast as HD 7970. It's quite logical it should be cheaper, because nobody would buy GTX580-like performance for GTX580-like price in Q2/2012

 

[trinibwoy]

Maybe I'm missing something but why is fighting for bragging rights at $300 with your biggest chip a good thing? If I understand correctly you guys are saying that its ok because GK104 may be the same size. Why does that matter? The last time AMD priced their top card that low it was faster than nVidia's second fastest - 4870 vs 260. Do we expect a repeat of that this round?

 

[A1xLLcqAgt0qc2RyMz0y]

It is again you who has posted off topic

Its not an attack.

I posted on Kepler you didn't.

You reposted an old post of mine asking for the AMD only posts to be removed as they were filling up multiple pages that had nothing to do with Kepler as if somehow my reply on Kepler GK104 & GK110 was somehow off topic. It's really funny how you see my Kepler post as off topic on a Kepler thread. My original reply was on Kepler so there can not be any calling out.

So if you have nothing to say about Kepler on a Kepler thread then don't bother posting at all as it is just a waste space much like the original off topic AMD only posts from before.

I will not reply to any further posts from you as I have better things to do that to feed your trolling.

 

[Arty]

You should probably go back and read, there were not pages but a few posts and we were discussing GK104's positioning in the stack. Since we have no new SKUs from Nvidia at this point, the reference was GCN - 7950 to be exact and GTX580 as well. This is still on topic.

What is not on topic is how its doom and gloom for AMD - your post.

Again if you dont like the medicine you want others to take .. Best to take this to PM, that is if you still havent made it to my ignore list yet.

 

[Xenus]

That's what I'm not getting from no-x's argument. I wouldn't say AMD is in a great position if their fastest card is competitive with Nvidia's 2nd fastest chip. (note said that way because I expect there to be a cut down 110K at the 570 like always so 2nd fastest card doesn't really work) It hurts their margins sure it's not a disaster as it's still a good chip and they still would be making money but I wouldn't call it a good position.

 

[rpg.314]

Why is BOM parity with GK104 important? AMD has much bigger things to worry about if it comes anywhere close to the 7970 performance wise.

Like what? In case BOM for the two is quite close.

 

[trinibwoy]

That's what I'm not getting from no-x's argument. I wouldn't say AMD is in a great position if their fastest card is competitive with Nvidia's 2nd fastest chip. (note said that way because I expect there to be a cut down 110K at the 570 like always so 2nd fastest card doesn't really work) It hurts their margins sure it's not a disaster as it's still a good chip and they still would be making money but I wouldn't call it a good position.

I'm glad somebody else sees it that way. It's certainly not a rosy position to be in. All hypothetical at this point of course.

Like what? In case BOM for the two is quite close.

1. No halo part
2. No more price/perf customer friendly aura (this is already slipping)
3. nVidia marketing blitz

 

[no-X]

That's what I'm not getting from no-x's argument. I wouldn't say AMD is in a great position if their fastest card is competitive with Nvidia's 2nd fastest chip.

2nd fastest chip? Depends on definition. Maybe I missed something, but Charlie (nor VR-Zone) didn't mention any faster part, which would be launched during April. I belive the fastest GPU isn't going to be launched anytime soon. In this case the $299 card will be their fastest product on market. It seems to me quite similar to G92 / GT200 launch.

 

[rpg.314]

1. No halo part

They didn't have the single chip GPU crown even during the RV770 era.

2. No more price/perf customer friendly aura (this is already slipping)

Easily and painlessly restored if the BOM is similar.

3. nVidia marketing blitz

They cut a lot of PR side jobs recently due to general financial troubles. A better or worse Tahiti wouldn't have changed this factor either way.

 

[trinibwoy]

They didn't have the single chip GPU crown even during the RV770 era.

Not the crown but they definitely had halo parts. Both the 5870 and 6970 qualify as they were up there with nVidia's best. We're talking about a hypothetical scenario where AMD is left battling nVidia's performance chip. That hasn't happened in a very long time.

Easily and painlessly restored if the BOM is similar.

Why does everyone keep saying they can just lower prices if BOM is similar to GK104? What does nVidia's cost structure have to do with AMD's profit potential? A high margin halo part gives you a lot of flexibility on lower tier pricing - nVidia's costs are irrelevant.

They cut a lot of PR side jobs recently due to general financial troubles. A better or worse Tahiti wouldn't have changed this factor either way.

That only makes the situation worse....

No point in discussing hypothetical scenarios to death though. I'm just baffled as to how a Tahiti vs GK104 @ $300 situation could be peachy for AMD, unlikely as it is.

 

[Arty]

Not peachy and definitely not doom&gloom, what ever happened to middle-ground?

 

[trinibwoy]

Not peachy and definitely not doom&gloom, what ever happened to middle-ground?

I agree. I was just responding to the claims that "Oh they just need to drop the price $200 and everything will be ok".