So, do we know anything about RV670 yet?

[AlphaWolf]

http://www.expreview.com/news/hard/2007-10-30/1193719352d6668.html

From FUD-presentation I expect higher wins, so RV670 maybe will goes better than expected.

Zomg the GT is more than twice as fast.

 

[flopper]

It seems in any cases that Nvidia have a better card today.
Its on the shelf at least for a minute until its sold.
No amd card out.

The question seems that ati again plays second fiddle.
dx10.1 what does that bring with today and games for the next 6 months?
I play Loto and bf2 and soon CoD4.
Crysis? maybe.
Double precision, means what?

As a consumer, the last year and still continues seems ati havent got the act togheter to produce a card that I would like to buy.

I used to have amd cpu, got a better one from nvidia.
Still have an old x1950xtx from ati but that also seems to be replaced by a better option.

I dont care about tmus, how the texcture pipes functions but how the card performs and works agaisnt the games and what the competition offers.

So, what do dx10.1 and double precision give in games today and 6 months ahead?

 

[_xxx_]

I've seen a few analysts asking where Nvidia is going to go in the next few years. They are going to have their lunch eaten by AMD on one side and Intel on the other, both offering chipsets, GPUs and CPUs, with economies of scale and complete platform discounts.

Sure, Nvidia may do well in the retail market, but OEMs are where the money is, and Nvidia just can't offer the complete platforms that AMD and Intel are moving towards. Nvidia are going to get stuck in the middle, and everyone wonders how they will get out of between that rock and a hard place.

Wherever I look, I can see next to none ATI cards or chipsets in OEM machines like Dell etc. So: huh?

 

[Mr_Chaos]

http://www.expreview.com/news/hard/2007-10-30/1193719352d6668.html

From FUD-presentation I expect higher wins, so RV670 maybe will goes better than expected.

Love the way they did the graph, but the GT is only between 10 and 40% faster.

 

[Shtal]

Zomg the GT is more than twice as fast.

ATI RV670 clocked @800MHz GPU should outperform/faster then "HD2900XT 740MHz-GPU" and as well possible "8800GTS" in some tests.

I will hold my breath for now before I would says about 8800GT vs. RV670.

 

[satein]

Wherever I look, I can see next to none ATI cards or chipsets in OEM machines like Dell etc. So: huh?

Time will tell

Don't underestimate whatever we don't really take our hand on it.

The way I look on this pull forward launch of NV is that they may not need to stock up a lot of GT for the OEM, so that they could built up stock for the retail market quicker and pull the launch forward (it may be wrong, just my guess). But on the AMD/ATi side, they are now ahead of their schedule on release this RV670 in Q1 2008 to Nov 07 already... so they considerably pull their launch forward to face NV at the NV initial launch.

Only a few weeks we will see then. Anyway, I don't believe that the HD 38xx will slam the 8800GT (G92) but it will be worth for its own.

Edit: typo and spelling

 

[satein]

Ok, this is may be what someone want...

New ATI HD 3800 To Support DX 10.1
Please note: The article is available on Tom's Hardware UK & Ireland!!

Thx you Dave for the head up

Edit: Typo again...

 

[Razor1]

AMD has consistently offered very competetive price performance parts since the days of the K6, and since 2003 has had a much more sophisticated platform than Intel's FSB based fare.

And although AMD is smaller than either nVidia or Intel in graphics and CPUs respectively, it has more diverse assets than either competitor. Integration of components has always been what's made costs low and performance high. It's never changed since transistors got replaced by integrated circuits, new functions might be added from time to time but they are inevitably integrated into a single silicon package

AMD is in a good position to capitalize on this fact. So looking at it the otherway, on the nVidia side nVidia won't have a prayer in terms of CPU design once graphics integration takes off; and on the Intel side, AMD's ATI assets outclass Intel's graphics division by quite a bit. AMD's ultimate plans for integration are also much grander and unifying than Intel's currently announced plans to glue a graphics processor onto Nehalem; AMD ultimately plans on having instruction level GPU integration. Fusion will also benefit from AMD's more mature and robust bus interface. Hypertransport has been on the market for years and is open and CSI won't appear until 2009 and will likely require a license.

Despite its hiccups now, AMD's foresight in terms of CPU/GPU integration roadmap will pay off handsomely in time.

I highly suggest to you read this...

http://www.fool.com/investing/general/2007/10/25/this-just-in-upgrades-and-downgrades.aspx

Their is alot of merit in Freedmans analyst. AmTech downgrades NV stock as well claiming NV's growth has peaked several months ago and is in a down turn. Freeman a dime in the dozen?

More anaylst down grade NV stock.

Including...

-Lehman chip analyst Tim Luke-



-CRT Capital Group’s Ashok Kumar-



During the past five days, Nvidia has saved off 14%.

Both of you really should go talk with Morgan Stanley analysts, the ones that actually gave AMD the loans AMD needed for the past year and see how they feel about AMD, nV's downgrade has little to do with AMD's future performance or Intel's entry in the graphics market (Intel won't go to gaming for quite some time still). nV's potential to give outperform numbers is capped, the market growth is harder to come by now since they own so much of the market. Stiffer competition means they won't capture as much marketshare or it levels off.

SC stock price means shit when you talk about this kind of stuff, why do you bring that up?

 

[vertex_shader]

http://www.expreview.com/news/hard/2007-10-30/1193719352d6668.html

From FUD-presentation I expect higher wins, so RV670 maybe will goes better than expected.

Reading this slides always fun, slide starting on 90%

 

[vertex_shader]

Fruitzilla

Radeon HD3000 to be cheaper than 8800GT

The fastest of the upcoming RV670XT will be cheaper than Geforce 8800GT. We already told you that this 825MHz+ clocked card is going to lose to a 600MHz default clocked 8800GT but ATI will simply lower its prices.

ATI already did this with the 2600XT/PRO and 2400XT/PRO series, and this put quite a dent in Nvidia sales, so the company plans do it again with both RV670XT and PRO. We still don’t know the real prices but we guess that ATI can easily drop some $30 to $50 from the price and we heard that RV670XT is between eight and ten percent slower than the 8800GT.

Radeon HD38x0 aka RV670XT might be selling for as low as €199 or a few dollars more, which will certainly get Nvidia in trouble. We will let you know when the prices are final and the launch should be fixed at 15th of November, in some two weeks

 

[Sound_Card]

http://www.xbitlabs.com/news/video/display/20071029062106.html

The Analyst I quoted are basing their estimates for the fourth quater going on to the first quarter. Was RV670 in the second and third quater?

 

[Geo]

Let's keep Nvidia financial analysts and stock performance past and future out of this thread, please.

 

[Sound_Card]

Both of you really should go talk with Morgan Stanley analysts, the ones that actually gave AMD the loans AMD needed for the past year and see how they feel about AMD, nV's downgrade has little to do with AMD's future performance or Intel's entry in the graphics market (Intel won't go to gaming for quite some time still). nV's potential to give outperform numbers is capped, the market growth is harder to come by now since they own so much of the market. Stiffer competition means they won't capture as much marketshare or it levels off.

SC stock price means shit when you talk about this kind of stuff, why do you bring that up?

Morgan Stanely Analyst on Intel, AMD, and Nvidia.

Claims Nvidia's growth will trickle down for the next four to six quarters.
AMD will be at full capacity with price pressure.

Although this has nothing to do with RV670 and I can't find Morgan's opinion on Graphics.

 

[Sound_Card]

Let's keep Nvidia financial analysts and stock performance past and future out of this thread, please.

Sorry I was replying already. I will stop now.

 

[Sound_Card]

DX10.1/HD3800 overview

http://www.tomshardware.com/2007/10/29/amd_hd_3800_to_support_dx_10/page3.html
http://www.pcper.com/article.php?aid=472

 

[Arty]

http://www.expreview.com/news/hard/2007-10-30/1193719352d6668.html

From FUD-presentation I expect higher wins, so RV670 maybe will goes better than expected.

If it is 10-35% from Nvidia's PR, then the race is much closer in reality.

 

[Arty]

I was laughing when I read Fuad in the morning today. I referenced the fall midrange (HD3870 v 8800GT) as 2600XT v 8600GTS couple of days back in this very thread

Remember, even if HD3870 v 8800GT is like 2600XT v 8600GTS then it would be a good win for them. Smaller die and also they will be trumpeting DX10.1 all the way ..

http://www.fudzilla.com/index.php?option=com_content&task=view&id=3918&Itemid=1

 

[3dilettante]

AMD has consistently offered very competetive price performance parts since the days of the K6, and since 2003 has had a much more sophisticated platform than Intel's FSB based fare.

The shared FSB was of only moderate hinderance to the desktop and mobile lines, and it greatly simplified the introduction of non-native multicore x86 for Intel.
AMD's use of Hypertransport and IMC helped in the 2-4 socket server space, particularly 4 sockets.

Unfortunately, some of the design decisions that made AMD's offerings great for that space hamstrung it in others.
Notably, the IMC hurt flexibility with regards to memory and memory capacity. Systems that had need of large amounts of memory and lesser processing were better served by Intel.

The demotion of the northbridge also hurt system providers' ability to differentiate their products, and RAS on AMD's chips has never matched the highest-end chipsets, shutting AMD out of that end of the server market.
If AMD had any server marketshare at all when K8 came out, this would have been a bad thing. Since AMD didn't have any, it was not an issue.
Growing out of its niche has now forced these questions to the fore.

Fortunately for AMD, Intel's reliance on FB-DIMMs likely widened the area of interest for Opteron long after Core2 entered the picture.
Intel will eventually move to a similar topology, but one that sounds in some ways more flexible than HT3.
AMD's lackluster processors are now dragging down the better platform.

And although AMD is smaller than either nVidia or Intel in graphics and CPUs respectively, it has more diverse assets than either competitor.

I don't know much about Nvidia's full product spread, but Intel makes processors, IGPs, chipsets (north and south bridges), flash business, and has licensed graphics IP.

If it weren't for the ATI aquisition, AMD would have had processors only as a commercial concern, having spun off flash and mostly relying on third party chipset manufacturers.
Now that it has ATI, AMD is back to having a subset of Intel's range in all but discrete graphics.

AMD is in a good position to capitalize on this fact. So looking at it the otherway, on the nVidia side nVidia won't have a prayer in terms of CPU design once graphics integration takes off; and on the Intel side, AMD's ATI assets outclass Intel's graphics division by quite a bit. AMD's ultimate plans for integration are also much grander and unifying than Intel's currently announced plans to glue a graphics processor onto Nehalem; AMD ultimately plans on having instruction level GPU integration.

The time frame on this plan is so long-term that we don't know what Intel will do.
The first Fusion products won't even have the GPU on the same die.
I the medium term, AMD's going to be happy when the CPU and GPU are on the same chip.
It may be a good 4-5 years before any serious integration begins to take place.
That is too far out to know what Intel might do.

If Fusion does well, Intel might paste Larrabee or parts of it on-die.
Being mostly x86, a somewhat ungainly Larrabee integration would likely provide a lead time over a "native" AMD solution.

Fusion will also benefit from AMD's more mature and robust bus interface. Hypertransport has been on the market for years and is open and CSI won't appear until 2009 and will likely require a license.

Cache-coherent Hypertransport requires a license right now, and will likely always require one.

Despite its hiccups now, AMD's foresight in terms of CPU/GPU integration roadmap will pay off handsomely in time.

We shall see how large the payoff might be. Too much is up in the air, and AMD has said little of substance about actual product or the necessary software and tool support needed.

There is no physical constraint that keeps competitors from going down the same path, and one large competitor is likely privately doing the same thing.

edit:
And I didn't get into the point about which company has issues with time.

 

[ShaidarHaran]

I've have seen many analysts ask about AMD and where its going to be in the next year, not even a few years. Analysts are a dime a dozen, and the market only goes by best performance/price/cost, and AMD doesn't have that with Barcelona, nor will they have that with K11. nV only has to worry about the ATi division, Intel will take care of AMD.

"nor will they have that with K11"

Really?

What is K11 and when's it due out?

 

[Geo]

CPUs also OT!