Nvidia's in trouble - Page 3

I think I'll get my future self to mail me a cryo-stasis chamber, that'll wake me up the day a graphics card is released that'll play Crysis at full smoke.

I wonder if this forum is all populated by kids.
With few exceptions I see childish comments all over without any clue on the subject.

1) The gamer market is only a tiny part of the whole IT market. You may be surprised to know that most of the income of the GFX board makers comes from the cheapest segment of their products. This is because, for each GTX or 3870x2 sold they sell at least 100/1000 computers for the business world. And these do not need directX10.x support to run Word or Excel (at least not yet, seen the latest moves from M$).
The race for having the fastest card is only for market purpose, not for really selling more of those pieces.

2) Notebook market is as profitable as the desktop one, so nVidia is well positioned with their offer of high performances integrated cards. They still probably still offer the fastest gfx cards for notebooks despite the war on chipsets they are suffering.

3) 2 slots, 2 CPUs, 1 lane, 2 fans, n sockets... huh? Fact is that the 3780x2 is a crap try to create the fastest GFX cards (for market and image reasons). The 3870x2 is faster only for games supporting Crossfire, otherwise you get the same performances of the standard 3870 (but wasting double power). When cross fire is supported, this card is only minimal faster than the Ultra, just showing how ridiculous is the try to get the crown of the fastest card at the moment. Yet, price has its importance. So, if you need SLI/Crossfire support enabled, I would buy 2x9600 in SLI for having faster performance at cheaper price than this single board with 2 GPUs . Not that 2 cheaper cards are worse than one more expensive if all of them do the same work and require the same support (SLI/Crossfire).
If you want to make comparisons seriously you will see that the 3870 is a loser in terms of both price and performances. If you want to belive that you have spent you money on the right product, well, for sure you made a better move than going for a much more expensive 8800GTX Ultra.
No, I'm not an nVidia fanboy. Can care less about who is going to provide the best card (I'm out of market with my puny GF6600 AGP8x). But best for me means performance at at a decent price (yet, the Ultra is not decently priced). More professional cards are available if you want to spend few thousands dollars.

Now going a bit technical. Hoping for nVidia to have troubles in future for marketing maneuvers of its competitors is quite dumb. If it will ever be true, it would mean we won't have the best product nonetheless.
Because if nVidia is forced to reduce its research investments due to financial problems that will not make ATi/Intel products better in a absolute point of view. Will just flatten the curve towards the bottom. The loser is the consumer in the end.
I would really hope for ATi (and AMD on CPU front) to come up with better products on their own. Just as it is now nVidia is messing up the market at its will (they could even release a faster card (8800GT 512) cheaper than their previously ultra-priced ones (8800GTS 320/640)) without caring about the confusion that would bring (yet, those cards are still rare on the market probably because they have still to sell some of these now obsoleted GTS laying around).
Meanwhile ATi could just hope to get a performance improvement by going to 55nm and so be able to raise the clock much over the previously version (removing the need for a nuclear power for the 2900XT version). Won't call that a breakthrough in R&D, seen that nVidia chips can obtain better performances at lower clocks and are produced at 90nm (now 65nm, but the fastest is still 90nm).
Still I'm not an nVidia fanboy. I'm a bit to old to be called a boy at all, BTW. I'm describing what I see presently on the market. And I can forsee, as anyone else can if can look at the whole picture, that nVidia still has not expressed it full potential.
New released 65nm chips are just clocked as fast (or even a bit slower) than the old 90nm ones. Now, what is holding nVidia from releasing a 8800GT Ultra card based on the G92 chip? Seen what this chip performs at stock clock with respect to the G90, that would outperform the current ultra a lot. Would also annihilate 3870x2 tiny gain where it has it. The answer is that Ati for now is not a great competitor, and nVidia has not will to outperform/obsolete its own cards that are still selling like cookies.
And this is not taking into account how the new G94 performs in its ultra cut down version (the one mounted on the 9600 has 30% less transistors than the G92 and performs almost as well). Seen the extra numbers of transistors needed for ATi R600 (which is the real record ATi can show) and its successive variants, it is clear that this is not state-of-art design.
More clock, more transistors (which mean production costs, even at 55nm) but less performances.
Other particulars make me think ATi has screwed up this development cycle: one was the try of the 512-bit bus on first R600 models for the memory which clearly show they expected better performance at higher resolutions. With current results, it is clear it is a waste of money.

In the end I really hope for ATi to come back with a new chip (R700?) that is going to further push nVidia to release something new on its turn (I think that without ATi 3xxx series we propably would have never seen nVidia's 9600 release so soon, at least). Or we may remain with 8800GTX Ultra as the top nVidia card for a whole year again. Which is not simply an ATi's market problem. It becomes to be a problem for everyone as the prices won't drop shortly (and they have already be stable for too long, IMHO).

Realistc as I am, I won't expect ATi to overthrown nVidia with next chipset (unless miracles happen) as I am quite sure nVidia is well hiding its wardogs, but I hope it will at least able to move waters a bit again.

If you find this a bit long, don't read it at all. It contains only bullshit of a probably too old guy.

I find your comments interesting, and to a point mostly true, except they lack the whole picture. You never mentioned Intels entry and you didnt address the cpu/gpu design forthcoming. Thats what I see as a roadblock/trouble for nVidia

Who would buy this card and play old games that don't support Crossfire? What games don't support Crossfire/Sli?

I'm not saying they are in financial trouble, I'm saying they're in trouble if they get locked out of a major emerging technology because they do not have an x86 license. I do not believe that Nvidia's enthusiast market lead will save them anymore than it saved 3dfx. I knew Nvidia worked on some kind of processor, but I didn't know they did so with Via. Via has the old x86 license from Cyrix. Perhaps that's not as restrictive a license as the one AMD has?

At any rate, Nvidia's not innovating the way they should in the enthusiast market. That won't stop their profits, and many will buy a dual PCB dual GPU card that the CEO of Nvidia has disdain for, but it doesn't change the facts that he's alienated Intel, AMD and his board partners.

Even more than AMD, Nvidia needs a new CEO.

Im not saying or wishing any demise or decline in current makers of either cards. From what Ive read, in as little as 2 years from now, Larrabee et al will either be out or on the doorstep. This isnt a pipedream, this is the future. As the manufacturing process becomes smaller, the on die solutions become wider, meaning that availibility of say , Larrabee will be there. Read this http://hoho.bafsoft.net/data/konquerort0Lnyb.pdf I know its dated, but with current comments from Intels Otellini, itll be here by 09 or 2010. Pay special attention starting at page 15 on the pdf. I DONT want less, I want more. But as this tech becomes reality, it leaves little space for nVidia, or discrete as we know it

Damn!! Skip skip...

i remember a crysis / nvidia promotion video when the game was released, hyping up how much better performance would be with SLi ....

Benchmarks have since proved otherwise.

I cannot believe that Crytek didnt optimize its game for multi gpu setups.

It sounds like you don't have an understanding of TWIMTBP program. It is not a simple pay off for an advertising deal. They actually help the developers with coding and such so that the games run with more stability. It is win win win situation (yes 3 wins, one for us). NV plugs the game, the game plugs NV, and if we have NV hardware, the game runs better than on other hardware.

NV gets a sneak peak at the game to better optimize drivers for it and since they get the developers to use NV coding, it ensures that NV hardware will run it best. It is actually a genius program that is killing ATI.

For me it's hard to understand the meaning of what you talk about.Who cares if the hd3870x2 it's better then two 8800gts.
Do you think I will change my mobo plus two new video cards to go crossfire instead of SLi. I choose to go SLi, next year when I decide to get new parts I will look at that time what is the better option. All thoses cards will be out there driver will more mature and eveything will go find.
Now it's more a discussion about, mine is bigger then your's. Continue to check yourself to be sure your the fastest the biggest. Me I will wait and decide when I'm ready.

Bah ignore this... I misread some of the info on the ATI site about the x2.

So in that case then ATI does have the advantage back imo.

Hehehe !

Sorry for my mistake

It really hurts these forums when misinformation gets spread like this. I'm sure you really did think that the 3870x2 requires a Crossfire mobo, but you should at least do a quick Google search before posting something as FACT if you aren't 100% sure. Otherwise you will look like a fool and confuse those who come here looking for advice.

Please correct your post and learn from this mistake

Excuse me sir but what planet do you live on? What ever planet that is I would love to buy property there. The early 7 series drivers and the current 8 series drivers are the best example of how bad a driver can get. I cant with good intent install an Nvidia card in any of my builds. I refuse to deal with the reports. I can however install a 3800 ATI in any build as I know its supported well now and is in the most future proof family of graphics cards out there.

Yeah, but my dad's a bigger alcoholic than yours!

Ryan.

We do need new architectures for sure. What alot of people missed about AMD is they banked on the improved performance of dx10.1. I think the DX improvement alone is going to put the 3800s over the 8 series on its own. "If" things go half as well as projected for dx10.1 it "should" put the 3800s above the 8800s in performance as long as they can improve on the high end x2 and keep it a viable option. We had a nice even field for a while. If these companies keep fighting it out we stand to see much better prices. If Nvidia does not jump on the dx10.1 fight yesterday nobody makes out.

Drink in moderation kids.

Ryan.

P.s - where is this Moderation place??

i just have 1 thing to say to mister game_boy.
u seem to know crap about anything ur saying.

Nvidia's g80 chipset as u see in almost every 8000 series from nvidia (except 8800 gt and 8800 gts 512 since both are g92) is at LEAST 9 months old.

u r freeking comparing a brand new card been placed on market from ati less than 1 month ago to gtx? gtx has been placed on market 8 months ago man.

and still gtx is almost as fast as the new ati card lolol. do u really think ati will be there to compete in performance against nvidia when 9800gx2 and 9800 gtx will come out? and no, if ur expecting 9800 gtx version to be ONLY 15% faster than the GTS version ur COMPLETLY wrong... and proves once more that u dont know what ur freeking talking about

and 1 more thing. HD3870X2 has 2 chipsets in 1 pcb and that makes it pci express 1 compilant. 1 more puch in the stomach of ati....

thats bullshit man
do ur homework before saying anything at all

Wait and see. Nvidia could pull a 9 series dx10.1 card out of there behinds who knows. The new ATI line is very promising as well. We are faced with 2 lines of cards. One is has slightly better performance all around but DX10.1 compatability reduced future proofing a great deal because of current dx10 performance. The other option from ATI is slightly slower and thus cheaper aside from the X2 who currently holds the high end benchmarks but all the 3800s are future proofed in comparison to the Nvidia line. Its realy apples and oranges here. Neither camp is offering us anything realy special atm causing me to look to the future aside from the X2. Now when the X2 is revamped I will give it another look untill Nvidia releases a dx10.1 card.

This must be sarcasm.

It wasnt actually.... nVidia, since the 6 series has always release high end cards that have around double the performance of the mid-range card.

If this 9800GTX is only just pipping the 9600GT, then why not call it the 9600GTX??

After all, same bit-bus, same memory amount... infact why bother with a 9800GTX at all if it's only superior specs are double the stream processors and a few mhz here and there?

Only for those who buy into the prepaid advantage Nvidia buys in a very few games. The only game I'm playing right now that's part of Nvidia's program is LOTR online, and I'm doing quite well with my 3870x2. In The Witcher, ATI's doing quite well because both companies are on the same playing field.

Nvidia has fans who prefer a few extra fps in a FPS, but ATI has fans who prefer image quality and don't want drivers fudged for FPS demos to be the reason we buy our cards. I will freely admit, I might have gotten just a 3870 instead of a 3870x2 if I'd had known that 4870 and 4870x2 was coming out in June, but I still don't regret it because reviews of The Witcher showed that it stuttered on a 3870 and an 8800gt, so the dual GPU card makes all the differences.

I'm opposed to any game being pre released optimized for either company. It's a loss for consumers. The only game that's in Nvidia's mishegoss program that I play is LOTR online and I'm still getting decent CRPG performance at full quality except very sharp instead of 8x AF.

If I played mostly FPS like Crysis, then perhaps I'd be more of an Nvidia fan, but I saw their blurring of image with the 7xxx series, and remember the early FX series as a waste of anyone's time to set up in a PC. Nvidia takes the low road while ATI tries for genuinely new tech, which eventually works when new drivers or second generation cards are released (ie the 3870 vs. the 2900xt).

Not that I normally go out of my way to defend Nvidia, but there are sources from game companies to news sites that they will skip DX10.1 because DX10 won't be fully utilized for quite some time. They will wait for DX11. Though DX10.1 is Microsoft's baby and not an ATI version like DX9.0b, I see it getting shafted from game companies waiting for the DX right afterwards.

Still, if any company does go DX10.1, then I'll benefit but current Nvidia cards will not. As far as it goes, if an RPG or RTS comes out with DX10.1, then it's for me. It will probably be the FPS companies that ditch DX10.1 because it's just too much work.

Ok, time for my points:

@Game_boy
1. Out of the 9 posts you made, 7 (or 8, i probably missed one) of your posts are in this thread wherein you predict things based on rumors. The other post you have is on another thread wherein you recomment the 3870x2.

2. On your original post, 5 out of 7 of your points are SOLELY BASED ON RUMOURS AND HEARSAY. The other two points are the only ones that have some factual basis. On the fifth part:

And probably on the 6th part.

Let me ask you, how could you criticise ideas based upon rumours and unreleased, unannounced products? These are basically just predictions, nothing more nothing less. What you said are as factual as any $2 fortune teller off the street.




Since I'm being so generous today, I'll criticise each of your points:

On your opening line, you did say that nvidia had 3x the market cap. With a ton of money you could do a ton of things, classic example, intel's prescott processor. It was hot, it was energy inefficient, it was slow, but it did sell. With help from intel's money and marketing they were able to sell millions of those things. (Don't talk to me about the morality or correctness of intel's methods, I'm stating what intel was able to do) With all that money, they could just buy off a lot of other companies, and still pounce on AMD.

Point 1: Nvidia's 7xx chipset is not yet released. A lot of things could be changed from beta stages and the retail product.

Point 2: I love the way that you confidently used phrases such as "is rumoured" and "a high chance". Well again, you assumed based on rumours that the 9xxx series would be a rebadge. Also why would nvidia even try to release a new high end product when ATi's flagship model has trouble challenging even nvidia's mainstream.

Point 3: I love this statement, "you haven't seen AMD's real high-end yet," my question to you is that "Have you seen their real high-end?" Unless you're an industry insider, that's a different story. The scores you gave could also be the final scores, or could have only marginal difference with the final score. We could never say until AMD does release it first!

Point 4: Gotta love these predicting lines: "plenty of independent rumours", "some potential model numbers", "there have been very few rumours", "we hear that", "almost ruling out", "I think". I couldn't even begin to comprehend how you could make this a point, as there is no factual point here.

Point 5: My question, who uses DX10.1 now? Yup, nada. Ahh, you say in the future they'd use them. But wouldn't it be also coincidental that probably by the time DX10.1 is used and made a minimum requirement, a line of new and faster 10.1 cards would be out? So I guess we could praise AMD for making a mid-end (or probably low-end) DX10.1 card.

Point 6: Probably the only point that is actually based on news and fact. Though this is not an indicator of the future of nvidia, as they could only make a faster card. Also how could their market share fall? When their money comes a lot from the mainstream GeForce cards. Sure workstation cards have high profits, but sale volumes in non-workstation cards does also make a difference.

Point 7: Larabee and Fusion haven't been released, all we have is marketing and technical white paper. How sure are you that nvidia wouldn't release a unified gpu and cpu before or alongside the big two. The best case release dates of Larabee and Fusion are somewhere along the lines of 2009 all the way up to 2012, so you just shot yourself in the foot with the statement: "I can't make solid predictions beyond 2008 though." Though I favor open-source programs, the users of open-source OS' against Windows or Mac users are still outnumbered by at least 3 to 1. Also the quality of software would be different between 2 paid employees against the work of 10 self-motivated, unpaid coders.


"I think they will try and offer better price/performance as AMD did in 2007" - Uhhh, nvidia and intel did have the better price/performance ratio througout the whole of 2007. Way to shoot yourself again in the foot.


Point 6:

Points? You come into this ignorance fueled flame-fest and offer us points? Release you inner troll

I got nothing better to do. And I'm bored.