Bookmark and Share


Nvidia Corp. has learnt its lesson when it unveiled code-named Fermi architecture about half a year ahead of actual release and would like to ensure instant availability of next-generation graphics cards after the formal launch, according to unofficial information. At present the company considers to schedule the launch of code-named Kepler products on March or April.

Nvidia, a leading designer of various multimedia processors, did not show any new graphics cards at the Consumer Electronics Show last week and even did not make any comments regarding its next-gen Kepler graphics processing unit (GPU) products. Instead, the company focused on demonstration of products running Nvidia Tegra 3 system-on-chips, including tablets and even cars. Although the market is clearly shifting towards portable and ultra-portable computing, standalone GPUs remain the most important business for Nvidia and the lack of any mention about Kepler was somewhat surprising.

According to information from VR-Zone web-site, Nvidia's new plan dictates launching the GeForce Kepler when the company is able to insure hardware availability and have all of its partners covered. While the approach may not be the best from advertising and marketing points of view, it is clear that instant availability allows Nvidia to sell more hardware since there will be loads of emotional pre-orders online after reading product reviews, whereas waiting kills emotions.

Kepler is Nvidia's next-generation graphics processor architecture that is projected to bring considerable performance improvements and will likely make the GPU more flexible in terms of programmability, which will speed up development of applications that take advantage of GPGPU (general purpose processing on GPU) technologies. Some of the technologies that Nvidia promised to introduce in Kepler and Maxwell (the architecture that will succeed Kepler) include virtual memory space (which will allow CPUs and GPUs to use the "unified" virtual memory), pre-emption, enhance the ability of GPU to autonomously process the data without the help of CPU and so on. Entry-level chips may not get all the features that Kepler architecture will have to often. Production of Kepler chips was supposed to start in Q4 2011.

Nvidia did not comment on the news-story.

Tags: Nvidia, Geforce, 28nm, Kepler, Quadro, Tesla


Comments currently: 11
Discussion started: 01/16/12 01:29:00 PM
Latest comment: 03/05/12 05:53:27 PM
Expand all threads | Collapse all threads


They didnt want to show a wooden card like last year? So in other words they dont even have a working sample yet.
3 2 [Posted by: cashkennedy  | Date: 01/16/12 01:29:00 PM]

1. GTX470/480 were launched around March 26, 2010. That's almost 2 years ago, not last year. Your time frame is way off.

2. Despite launching 6 months late, NV still managed to have a ~ 60% market share on the desktop. This in itself highlights the fact that launching first even by 6 months doesn't guarantee success, especially since the market for $549+ graphics cards is very small (~2%). Unlike HD5870/5850 cards that were launched during Q4 2009 (pre-holiday season), HD7970 launches into a slow Q1 2012, following Q4 2011 that saw PC sales fall 6%:

So add extremely high price of $550 that attracts 2-3% of buyers, add lack of sufficient stock of HD7970 cards on retail, HD7950 delayed more than a month, and the head start this time is even less of an advantage than it was last time because HD5850/5870 were priced very aggressively and launched in a quarter where a lot of gamers bought hardware for Xmas/etc.

3. Launching first is effective only IF you are able to deliver an entire new product line, including lower priced cards such as $99-$349. This is because 85% of desktop discrete GPUs are sold in the < $349 price bracket. Yet, so far AMD only launched 1 high-end HD7000 series card on the desktop. It will take them another 1-2 months to get their entire product line out. At this time, NV still has very competitive product line-up under < $349.

4. Q1 2012 doesn't have a single demanding PC game launching. Therefore, there isn't a catalyst for current gamers with high-end hardware (GTX480/570/580/HD6950/6970) to upgrade since they will have beaten all the demanding games before (Crysis 2, Metro 2033, Dragon Age 2, Witcher 2, etc.).

5. The key games in 2012 are Mass Effect 3, Diablo 3, Starcraft II: Heart of the Swarm, Max Payne 3, etc. None of these games needs anything faster than a $200-250 GTX560 Ti/HD6950. How many people are going want to upgrade for BF3 alone then?

6. On the mobile side, the 7400M, 7500M, and 7600M are all VLIW5 parts, which are “rebadged” 40nm HD6000 GPUs. All of the yet-to-be-announced parts above 7600M (7700M/7800M/7900M) will switch to 28nm GCN (Graphics Core Next), but according to AMD won't launch until April/May 2012 timeframe. So oce again, AMD still needs at least 1 full quarter to complete their "real" HD7000 line-up for laptops.

7. Current generations now last 18-24 months. So if NV launches within 1 Q, it's not a big deal, especially if it handily beats HD7970. It would be unfortunate for early adopters to buy a $600 card only to be 2nd best in 3-6 months, especially when Q1 is "console ported" PC games.

8. NV launching late isn't amazing for anyone. It is in our best interests that they deliver something competitive.

All in all, in my opinion, what makes a generation exciting is utilizing higher performance hardware in next generation games. And so far the 1st half of 2012 doesn't have a single "killer" demanding game like Doom 3, Far Cry, Crysis 1, Metro 2033, etc. that calls for a $550 GPU upgrade from NV or AMD. On the AMD side, given that HD7970 is only 40% faster than HD6970, it doesn't look like there will be a card much faster than HD6970 for $299 any time soon. This is unfortunate. So really, only people who jump from GTX480-->580-->7970 (i.e., drop $500+ on GPUs) are excited about 7970/GTX680 style cards.

What we need are MUCH faster mid-high end cards such as HD6950/GTX560Ti and we might have to wait until 2013 before that happens.
3 3 [Posted by: BestJinjo  | Date: 01/16/12 04:25:49 PM]
- collapse thread

A very well thought out post. As an average PC gamer, I can't see buying an incrementally better card when my current 560ti card just played Metro 2033 at highest graphics settings. Waiting until 2013 for a new mid-high end card fits into my upgrade needs just fine.
2 0 [Posted by: ralphrainwater  | Date: 01/17/12 03:52:42 AM]
Very nice post, I can't understand who gave you thumb down.

2 2 [Posted by: nitro912gr  | Date: 01/17/12 07:41:32 AM]
GTX470/480 were launched around March 26, 2010. That's almost 2 years ago, not last year. Your time frame is way off.

In fact the author was talking about Fermi architecture announcement (autumn 2009) and Fermi release (spring 2010). It's literally: "half a year ahead of actual release".

Launching first is effective only IF you are able to deliver an entire new product line

What do you mean with "effective"? It's always effective to release as soon as you can.
1 0 [Posted by: Marburg U  | Date: 01/17/12 11:38:31 AM]
What I mean is if you release a niche product at $450-550, then you are only going to capture a small fraction of the market share since most gamers do not buy those products. Normally, a new generation brings in better performing low and mid-range offerings, but AMD has delayed these cards by a full quarter.

What I mean is that from a business perspective, HD7000 launching first would have had a tremendous competitive advantage for AMD if they launched a full line-up. However, their current strategy will hardly hurt NV initially because AMD only decided to release $450-550 cards in the form of 7970 and 7950. For example, the best selling card on Amazon during the holidays was GTX550Ti.

I disagree that it's always effective to release as soon as you can because it depends on the product and the cost per wafer vs. die yields, etc. AMD put all their efforts into launching HD7970 in small volumes due to poor yields on 28nm at the moment. Instead they should have focused on low-end, mid-range and mid-high end discrete GPU and mobile parts, which are MIA.
0 1 [Posted by: BestJinjo  | Date: 01/17/12 02:06:58 PM]
Do you know of any chip (worth of billions in R&D) which was ready to be released but was put on hold because it was not yet its time?
Also, which are the profits AMD makes from the low-end to mid-high segment?

Really, i cannot see any logic in waiting to release a high profit margin product, that is ready, and did cost you a lot of money.
Gosh, firms release even bad products, because it's, nevertheless, better than have nothing to release.
1 0 [Posted by: Marburg U  | Date: 01/18/12 12:34:51 AM]
"i cannot see any logic in waiting to release a high profit margin product"

1) Wait until you can achieve higher supply volume / better availability.

2) Giving 1 full quarter at TSMC will improve yields, so AMD might have launched 1075 or 1125 mhz cards in the first place. To get higher yields, they neutered the card with only 925mhz clocks.

3) Since other companies are competing for same 28nm wafers, the price per wafer is MUCH higher in the 1st quarter than it will be later in 2012. AMD is paying the highest possible price per wafer at the moment and poor availability shows they are supply constrained.

4) The programs/drivers that allow for hybrid HD video encoding aren't even ready! That means the entire launch was rushed as some key features are unavailable.

But most importantly, my point is that they should have launched low and mid range HD7000 series cards first because there are no new demanding PC games coming out in Q1 (Max Payne 3 just got delayed to May 29th). So what was the point of launching a $550 GPU first? 85-90% of gamers don't buy $400+ videocards either. Right now NV has an extremely competitive lineup in the <$349 level and below. AMD had a huge opportunity to capitalize on the bulk of that market but instead NONE of their HD7600/7700/7800 cards is ready. Finally, they'll probably need to respin HD7970 to have any chance of competing with Kepler (or perhaps AIBs will release much faster clocked cards).

Enthusiasts know that current generations last 1.5-2 years now since both AMD and NV switched to longer generational cycles. Not many people are going to want to drop $550 on a GPU and not see what the competitor is bringing, especially during slow Q1 with no new demanding PC games. If Kepler flops, worst case scenario you can pick up a higher clocked HD7970 with better aftermarket cooling in Q2 (since stock cooling is loud) and hopefully there might actually be some demanding PC games by Q2. If Kepler is faster, HD7970 might drop in price, or AMD does a respin and gives us an even faster HD7980.

AMD launched HD7000 series but really it was either all rebadged HD6000 series for laptops OR HD7970 series that's almost out of stock everywhere. So in other words, it was a poorly executed launch. Compare this to HD4850/4870 or HD5850/5870 launches...

NV may have purposely delayed their launch by 1 quarter to release a 20% faster GPU and have much better availability.

So basically AMD launched the fastest card they have in low volumes during a quarter with 0 demanding games. Why in the world would any enthusiast spend $$ on that and not wait 3 months to see what NV has? On the other hand, people who actually want to upgrade their older HD5750/5770 cards have nothing to upgrade to since there are no low end or mid-range HD7000 series cards.

It appears the real motivation of releasing first was to please investors and to say they launched first 28nm GPU generation, first DX11.1, first PCIe 3.0, etc. But in terms of market effectiveness/market penetration, it was one of the worst executed launches ever. HD7970 was announced December 22nd, but not available for sale until January 9th (and then it sold out). HD7950 was delayed until end of January. The low-end and mid-range cards are not going to come out until February at the earliest. Overall, they are going to take 1 full quarter to actually get the entire lineup out.

On the mobile side, we probably won't see any good HD7000 chips until April (so Q2 2012).
0 1 [Posted by: BestJinjo  | Date: 01/18/12 08:33:12 AM]
I can't agree 100% here, As someone who replaces every 2 cycles, I tend to buy up at the top when I do. I am sure that there are more of us like me who jumped on the HD5870 bandwagon when they dropped it seemed like a great product with an amazing price, and have regretted it since.

I swapped out my 9800 Gx2 (which kept overheating) for an HD5870 when they first came out, and although I saw a big difference in raw number performance, I have been plagued by awful driver support for windows 7 x64, and tons of random crashing in a bunch of games.

Saving some cash by going AMD was a lesson in frustration, one I am not going to repeat. I'm sad that the software behind the AMD stuff sucked so bad, because the hardware was kick-ass, and frankly the Southern Islands gear looks like it's going to follow in the footsteps of the thanks I say...not at any price.

0 0 [Posted by: cronos1013  | Date: 03/05/12 05:53:27 PM]

What I need is a 790GTX dual card,Wether or not I need the power for 2012 games is another matter, I just want one card to replace my two nuclear power plants, uhhh I mean 480GTX's
1 1 [Posted by: ozegamer  | Date: 01/16/12 11:34:47 PM]

GK 104 is already back from fabrication and boots Windows 7 Benchmarking in progress
0 1 [Posted by: jewie27  | Date: 01/17/12 01:40:49 AM]


Add your Comment

Related news

Latest News

Wednesday, November 5, 2014

10:48 pm | LG’s Unique Ultra-Wide Curved 34” Display Finally Hits the Market. LG 34UC97 Available in the U.S. and the U.K.

Wednesday, October 8, 2014

12:52 pm | Lisa Su Appointed as New CEO of Advanced Micro Devices. Rory Read Steps Down, Lisa Su Becomes New CEO of AMD

Thursday, August 28, 2014

4:22 am | AMD Has No Plans to Reconsider Recommended Prices of Radeon R9 Graphics Cards. AMD Will Not Lower Recommended Prices of Radeon R9 Graphics Solutions

Wednesday, August 27, 2014

1:09 pm | Samsung Begins to Produce 2.13GHz 64GB DDR4 Memory Modules. Samsung Uses TSV DRAMs for 64GB DDR4 RDIMMs

Tuesday, August 26, 2014

10:41 am | AMD Quietly Reveals Third Iteration of GCN Architecture with Tonga GPU. AMD Unleashes Radeon R9 285 Graphics Cards, Tonga GPU, GCN 1.2 Architecture