I don't know... my guess is Nvidia is selling them at a loss. Remember, the GPU used in the GTX260/280 is huge. Also, rumors are that the yields are not very good. But, Nvidia would rather sell it at a loss/maybe break even then manufacture the chips and not sell them at all.
Think about it, sell 100% of your chips for a $100 loss, but sell all of them, keep market share and happy customers vs. build the parts, sell only 10% of them at a profit, have lots of inventory that they paid for that they cannot move. That's my guess anyway.
But for me, the reason I plan on using AMD/ATI cards for future builds is that Nvidia kind of rubbed me the wrong way. How come they only release these miracle drivers that enhance performance when they have competition? I feel like they were holding back performance on purpose to have an ace up their sleeve when the 48x0 were released. Same with Physx. Also, how come AMD/ATI, who was lossing money faster then Nvidia was making it can release an official driver release monthly that supports all their cards? Meanwhile, when Nvidia was raking in the money they seemed to just sit on drivers and let people try and find hacks to make them work. Just seems to me that Nvidia doesn't always try to please their customers, they figured they had the performance lead so they could do what ever they wanted.
In a sense that's unfortunately how it works.
And keys, I agree.
Think about there only being one processor manufacturer. They could raise the price of Q6600 from $195 to whatever they wanted. Not many people would be happy.
Nvidia has always kind of had a bad rep when it comes to drivers. ATI's CCC isn't the best either but their driver fixes were always a large improvement.
Nemesis 1
Lifer
- Dec 30, 2006
- 11,366
- 2
- 0
Well I am a ATI user. But Nv has done real good by most you guys in past and thats what U should look at. Is it possiable that NV just got a little stupid when they seen the possiabilities that cuda brings. I mean really if the programms existed this thing would kick ass. This is the reason that NV can't do DX10.1 . NV has to adapt true DX10 or pass on like vodoo.
The problem is the programms don't exist and intel has effectively cut NV off befor NV could reach the pass into cuda valley. Its an X86 world and its doing to remain that way for sometime . Something I am not happy about . EPIC is the way to go . But its seems our higher education are handing out diplomas to people they should not . Its not earn a diploma . Its BUY a diploma. I know I bought 2 . One for daughter and one for wife . Higher education myass.
In the next 3 years there is going to be blood in the streets companies will paass away. But which companies . New tech comes along that changes everthing. Like that new chip that suppose to scale multi gpu's . If it does whats said it can do . Seems to me this changes everthing . Be cause all of the sudden .More than just 2 or 3 players can play . With scaling things can change really fast, So who really knows.
The problem is the programms don't exist and intel has effectively cut NV off befor NV could reach the pass into cuda valley. Its an X86 world and its doing to remain that way for sometime . Something I am not happy about . EPIC is the way to go . But its seems our higher education are handing out diplomas to people they should not . Its not earn a diploma . Its BUY a diploma. I know I bought 2 . One for daughter and one for wife . Higher education myass.
In the next 3 years there is going to be blood in the streets companies will paass away. But which companies . New tech comes along that changes everthing. Like that new chip that suppose to scale multi gpu's . If it does whats said it can do . Seems to me this changes everthing . Be cause all of the sudden .More than just 2 or 3 players can play . With scaling things can change really fast, So who really knows.
The market in the 300+ range of cards is so small there is no need to sell at a loss to retain market share. 300+ is your bread and butter, where you get insane margins to offset the million 80 dollar low end models you sold over the same time period.
Like I said we will see when the 3rd qut numbers come out sometime in Oct.
Actually its quite the opposite. The mid to low range parts are where the vast majority of revenue is made. The high end parts as you said are a very small market and while they have higher profit margins its certainly not enough to outperform the revenue that the low- midrange cards bring in.
Thats one of the reasons ATI decided to shift its strategy. Instead of continuing the sea-saw battle of who has the biggest and best flagship card. They decide, hey lets just crush them in the mid-low range market because thats where the money is anyway.
But pricing isn't done in a vacuum. If vendors tell NV they can't sell cards at $650, NV has to lower the asking price and go to their supplier and renegotiate pricing as well. Everyone eats a little bit less or no one eats at all. That's what happens in a dynamic economy. You'll see it in just about every industry right now as the global economy is down.
Plus the board will probably be another $100 slapped on top of that, being 10.5" long 512bit with 1gb of ram will be costly
then theres the packaging, accessories, tax and shipping, sure Nv wont be making a loss probably break even to a small profit on the actual sale of the card but it wont nearly pay for the maintainence costs of a company of nvidias size
The cost for NV does not change. It was $100+ in the beginning. When yields improve then it will go down.
Now when I think about that the cost can not be less then wafer/number of chips. It's 5K/100 chips, about $50 per chip at 100% yield. I doubt they will achieve 100% yield.
NV needs the shrink to lower the cost as soon as possible.
taltamir
Lifer
- Mar 21, 2004
- 13,576
- 6
- 76
it is IMPOSSIBLE to discuss market share without taking price into account.
Price is the number 1 factor when a person makes a purchase.
People even go without getting certain medical treatments because they are too expensive (which they COULD scrounge up enough money to pay for it with extreme cost cutting elsewhere, or liquidating assets).
I would have to disagree. The number 1 would be value, which is different than price.
If price were the #1 factor, we'd all have 8600GTs or 3850s.
What the hell does any of that have to do why NewEgg's review #s are skewed and not an accurate measurement of market share? Way to completely miss the point of my post. I wasn't even making any comment about 4850s, and how or why they're popular. If I were, then price would become relevant.
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
I think branding "price" as the number 1 factor for a purchase decision is misleading. I think its a combination of various factors for instance the products performance.
And im kind of confused by the thread heading. I think declaring a "win" is abit premature unless it meant they won this round. So shouldn't the thread be titled "why AMD won this round?". Plus ATi technically "lost" ever since 2006 since they got outplayed when it came to business and hence ended up being acquired.
And im kind of confused by the thread heading. I think declaring a "win" is abit premature unless it meant they won this round. So shouldn't the thread be titled "why AMD won this round?". Plus ATi technically "lost" ever since 2006 since they got outplayed when it came to business and hence ended up being acquired.
taltamir
Lifer
- Mar 21, 2004
- 13,576
- 6
- 76
true... ATI lost in 06, now its AMD vs nVidia vs Intel (soon). ATI is dead and gone.
Performance is a very important factor indeed, it goes right there at second place, after price. Often performance goes hand in hand with price. And you could always bring in compound values, like performance / price.
But when an item that performs twice as well sells for the same price, it is winning in price, not just in performance.
If an item that performs twice as well costs twice as much, then it has performance alone without price.
If an item that performs twice as well costs 5 times as much, it looses in price and performance / price, but it wins in performance. It will appeal to some, but not to the majority of people.
Performance is a very important factor indeed, it goes right there at second place, after price. Often performance goes hand in hand with price. And you could always bring in compound values, like performance / price.
But when an item that performs twice as well sells for the same price, it is winning in price, not just in performance.
If an item that performs twice as well costs twice as much, then it has performance alone without price.
If an item that performs twice as well costs 5 times as much, it looses in price and performance / price, but it wins in performance. It will appeal to some, but not to the majority of people.
SolMiester
Diamond Member
- Dec 19, 2004
- 5,331
- 17
- 76
SolMiester
Diamond Member
- Dec 19, 2004
- 5,331
- 17
- 76
Originally posted by: dadach
someone said something about price gauging...im sorry, but what nvidia tried to do was pathetic...it actually tried to screw their most loyal customers, with starting prices of new gtx series...after they saw that ati has much better value, they rediculously dropped the price...ofc i could be wrong, and nvidia is losing money on every sale
Isnt that what people do when they sell their home?, ask for the most 1st...duh...not pathetic, just trying to realise as much profit as possible,
I don't see any losers for GPUs b/c both companies have decent GPUs out right now, and Nvidia will make a profit again once the die shrnks come along. Nvidia's real problem is the notebook chipset issue.
so you do agree nvidia tried to screw their customers
how much did the price actually drop percentagewise...does it relate in figures to selling of the houses?as for other points discussed, one thing i think you forget is that when the price drops, it drops only in US market...all the other markets stay unaffected for the long time...here in croatia 4870 is almost 100 USD cheaper than GTX260, and im reading that you can buy both for the same price in the US...ok, croatia is a little extreme, so id be intersted to see how fast did the price dopped in other EU countries...
and btw GTX280 is 720 USD overhere :S
Originally posted by: dadach
so you do agree nvidia tried to screw their customers
as for other points discussed, one thing i think you forget is that when the price drops, it drops only in US market...all the other markets stay unaffected for the long time...here in croatia 4870 is almost 100 USD cheaper than GTX260, and im reading that you can buy both for the same price in the US...ok, croatia is a little extreme, so id be intersted to see how fast did the price dopped in other EU countries...
and btw GTX280 is 720 USD overhere :S
Sol is on the right track here IMHO
And it's not a case of any one company/person/entity "screwing" their customers Any commodity, whether a home, vehicle, video card, foodstuff, anything that one person has and another wants, is worth what?
It's only worth what someone is willing to pay for it! Basic, basic stuff
You can say that your house is worth $500,000, but if you want/need to sell it and the most you are offered is $400,000, then it's not "worth" $500,000
and if you are offered $600,000, you are not "screwing" the buyer, that's what it is worth to him/her.
That's absurd to think NV's costs are static, you think those SUVs selling for 40% less than a year ago aren't being sold for a profit still? There's just much, much less profit going around and even at break even or a slight loss, its better than not selling anything.
Similarly with chip pricing, if NV approaches TSMC with 2 scenarios, $100 a chip based on $649 pricing or $60 a chip based on $449 pricing but selling 5x as many units, which do you think they're going to choose? The difference in cost to them is some glass.
People have been saying these monolithic chips from NV can't possibly be selling for a profit due to low yields and # per wafer, but that clearly has not been the case with G80 and with GT200 we're seeing just how low NV can go on pricing while still managing to break-even.
apoppin
Lifer
AMD does not "win" and Nvidia "lose"
- this is ongoing. imo Nvidia's problem is *perception*
You guys [simply] perceive they are losing
they are not .. they are reevaluating their situation and basically following their original plan with Tesla ... they are a solid company with a few slips
- one of these "slips" was to alienate their former fans - "us"
i don't see AMD going away either. We have tough competitors that have already survived extreme times - and ultimately that is very good for us
the extreme views in this thread are silly imo; and the tech yellow press is playing it up for page hits
- i call it a tempest in a teapot
The costs are still $110-$125 per GPU. And Nvidia needs the shrink to 55nm to make profitable Tesla GPUS again; they are patient and can afford to hold on. That is why their partners are sticking with them.
- this is ongoing. imo Nvidia's problem is *perception*
You guys [simply] perceive they are losing
they are not .. they are reevaluating their situation and basically following their original plan with Tesla ... they are a solid company with a few slips
- one of these "slips" was to alienate their former fans - "us"
i don't see AMD going away either. We have tough competitors that have already survived extreme times - and ultimately that is very good for us
the extreme views in this thread are silly imo; and the tech yellow press is playing it up for page hits
- i call it a tempest in a teapot
These people are right. Nvidia is not making money on Tesla; however, they can easily afford to take a loss on their new chips as they have a HUGE margin with g80/g92.
The costs are still $110-$125 per GPU. And Nvidia needs the shrink to 55nm to make profitable Tesla GPUS again; they are patient and can afford to hold on. That is why their partners are sticking with them.
They're not selling GT200 Tesla yet, but they would certainly be making a profit on them if they sold for MSRP of $8k for a 1U rack of 4xGT200. This isn't much different from Quadro though, where there is clearly a huge premium and high margin on very low volume.
But people keep basing wafer price on some obscure pricing sheet on the internet and claiming there's no way NV can be getting high enough yields or enough chips per wafer to make a profit. I don't know exactly how much they're paying per chip, but I can guarantee you NV has substantial leverage with pricing as the global leader in discrete GPU sales.
Even in G80's prime the same $100-$150 price per chip and shortages were being thrown about, yet a few months later, NV somehow released a sub-$300 G80 part with the GTS 320MB aimed squarely at the mid-range mainstream enthusiast. Despite similar doom and gloom over G80's costs, NV only managed to post record profits in the YTD from G80's release, prior to G92's launch. Clearly people overestimated chip pricing and yields then, just as they do now. The only difference is now with GT200 is we are truly seeing the lower bounds of NV's pricing/profit model.
WB btw
TRENDING THREADS
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 9K
-
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 7K
-
Facebook
Twitter