compgeek89
Golden Member
- Dec 11, 2004
- 1,860
- 0
- 76
I agree that the r520 will likely be a tad slower than the gtx. The IQ thing is really up in the air at this time though, so may be very interesting. I do not think Nvidia held back the unannounced, unseen by anyone, ultra for this reason ---- because it never existed - as the cost of the extra cooling and power is likely not worth the gain in clockspeed.
Look at it from another perspective and things change around a bit. First off is that the rumors of the Ultra ever being planned are as yet unconfirmed, but let's say the G70 was supposed to be a 32 pipe card or one clocked much higher(makes little difference).
nVidia was obviously very concerned about not paper launching the G70- they wanted to have the board available in quantity when they launched which they were able to do. Given the complexity of the chip the yields on the earlier runs couldn't have been awe inspiring, likely why we are seeing the GT in the configuration it shipped in. It is possible that nV could have been binning the highest yield G70 parts in order to use as an Ultra launch to combat the R520 when it hit- but look at the current timeline.
The G70 hits in June- that would place a likely nV refresh somewhere in Q1 '06- as of this point it looks like ATi will be pushing up against Q4 '05 before they will get minimalistic quantities out the door and it will likely be Q1 '06 before they are hitting with any quantity of any of their high end parts. What would nVidia gain by releasing an Ultra now? They could position themselves better at the highest end against whatever ATi has in terms of mindshare. What would they lose? They would have to drop their margins on all of their existing products and work another product line into the upper echelon of pricing- the 7800GT can't be produced cheaply enough to drop in to the $200-$300 price range yet(in terms of MSRP) so they drop the GT to say $349, have the GTX @$449 and the Ultra @$549(all MSRP). That would reduce the profits considerably on their top tier parts and would introduce another problem for them.
Whatever they have lined up for refresh in early '06 it is likely to be based around a die shrunk G70. By leaving the GTX as their top tier part all they need target is clear lead over it in order to make it a viable refresh solution. With ATi's problems as shocking as they are, I don't think nVidia is feeling any real pressure from them on the high end. With their focus divided between the R500, R520 and R580 all at the high end with the console launch that must be pulled off as their primary concern- it leaves ATi in a bad position to try and play any sort of games in terms of positioning themselves right now. They can't afford to- they need to get product out the door. nVidia's efforts right now seem quite consolidated. Their console contract is for a 90nm G70 based core clocked @550MHZ with 32ALUs- something that would make a fairly nasty refresh part itself. If nVidia were to release an Ultra now wither either 32 pipes or clocked @550MHZ(obviously not both on their build process) the move over to a RSX equal wouldn't look terribly impressive and would leave people less then impressed with their progress. Not a good situation when trying to launch a new top tier part.
We can always say that any IHV should have released part X sooner if they could have- but look at Intel's Extreme Edition processors- is that a market you want to see GPUs headed towards? If the yields are too low to launch something in a reasonable price range- and you have no competitive reason to do so- why spend the additional funds to come up with a quick speed bump when you could instead focus more effort on bringing a much larger improvement to the market?
IF ATi was firing on all cylinders and had their R520 out three or four months ago nV would likely be in a position where they would need the Ultra out the door to bring mindshare in the marketplace their way. As of now, they own the mindshare of the market outside of die hard fans of ATi. I'm not saying they should as we haven't seen the R520 yet, but the fact that we haven't seen the R520 yet is precisely why nVidia can focus solely on what benefits their bottom line the most at the moment.
Yeah ok, let us know when your Ultra ships.
I love how people bash the Inq at every corner just because they don't want to hear what they have to say.
Okay, you've convinced me. I thought I was wrong.
It's probably better for nv in the long run that they picked low clocks and allowed the manufacturers to overclock as they desire.
It's probably better for nv in the long run that they picked low clocks and allowed the manufacturers to overclock as they desire.
Nope, if Nvidia could widen their lead and get the top of the line boys to buy again. They would. If you are saying that Nvidia might reproduce their ultra, ultra thing (exceedingly limited release) to stay at or near the top - maybe. But who really cares about small runs for benchmarking, pr purposes. Whatever problems Ati has, it is not effecting much but prices right now - if Ati drops out of the race, than we can expect a slowing of development money from Nvidia.
ddogg
Golden Member
- May 4, 2005
- 1,864
- 361
- 136
well Nvidia doesnt need to widen their lead as there is no reason to do it. An Ultra part is simply useless if it shows max gains of 10% over the GTX. It would just mean more names, more confusion and availability would be scant. It would be better they release a card 6months down the line (G75 or G77 whatever the heck its called) with gains of atleast 30-40%.
compgeek89
Golden Member
- Dec 11, 2004
- 1,860
- 0
- 76
You are failing to look at the entirety of the situation in this assumption. The pros and cons of such a move need to be weighed when considering such a move. First off is the cost associated with having to special bin the fastest testing chips- but we will assume that is a given. Next you have to line up RAM suppliers to feed slightly higher bandwidth ships to you as you don't want the Ultra finishing in a dead lock with the GTX at any point- so another supply line needs to be opened up or expanded. Then you have to deal with all of the various marketing, offering the additional SKUs to retailers etc(obviously the OEMs would have a big hand in this too). This may not add up to an enormous amount for a company the size of nVidia or their OEMs however it is additional costs incurred. So say all of that is free and takes no time- they can wave a magic supply and distribution wand for their new parts for arguments sake. Where do you launch the Ultra in terms of pricing?
I'll give your line of thought the advantage here and say that we are going to launch the Ultra @$599 MSRP so we can retain very high margins on the part. Of course the only people jumping at this would be the extreme whacko type(aka Rollo
) or those who are simply looking to pick up the highest end part possible. For the former, they are likely already sporting 7800GTXs so they are a good turnaround revenue stream for the OEMs and nVidia.This is where we hit our major problem with the entire scenario. nVidia has just launched a new part that only competes with its prior part that had no competition. This means that nVidia must lower the price on the lower cost higher yield GTXs and GTs to make room for the Ultra. They take their higher volume parts and reduce their yields for what benefit- the extremely small minority that would upgrade from a GTX to an Ultra? If they lost just $20 margin on the GTXs and GTs that would be a loss of millions of dollars- do you honestly think that they would manage to recoupe those costs with the amount of sales that they are going to secure with a marginal speed bump part? Not likely at all.
As a competitor to ATi's flagship the merit may be argued- the Halo effect is very pronounced in the GPU market- but when their only viable competition on this level is themselves it is foolish of them to force themselves to reduce margins on a part before their is need to do so.
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
Originally posted by: BenSkywalker
You are failing to look at the entirety of the situation in this assumption. The pros and cons of such a move need to be weighed when considering such a move. First off is the cost associated with having to special bin the fastest testing chips- but we will assume that is a given. Next you have to line up RAM suppliers to feed slightly higher bandwidth ships to you as you don't want the Ultra finishing in a dead lock with the GTX at any point- so another supply line needs to be opened up or expanded. Then you have to deal with all of the various marketing, offering the additional SKUs to retailers etc(obviously the OEMs would have a big hand in this too). This may not add up to an enormous amount for a company the size of nVidia or their OEMs however it is additional costs incurred. So say all of that is free and takes no time- they can wave a magic supply and distribution wand for their new parts for arguments sake. Where do you launch the Ultra in terms of pricing?
I'll give your line of thought the advantage here and say that we are going to launch the Ultra @$599 MSRP so we can retain very high margins on the part. Of course the only people jumping at this would be the extreme whacko type(aka Rollo
This is where we hit our major problem with the entire scenario. nVidia has just launched a new part that only competes with its prior part that had no competition. This means that nVidia must lower the price on the lower cost higher yield GTXs and GTs to make room for the Ultra. They take their higher volume parts and reduce their yields for what benefit- the extremely small minority that would upgrade from a GTX to an Ultra? If they lost just $20 margin on the GTXs and GTs that would be a loss of millions of dollars- do you honestly think that they would manage to recoupe those costs with the amount of sales that they are going to secure with a marginal speed bump part? Not likely at all.
As a competitor to ATi's flagship the merit may be argued- the Halo effect is very pronounced in the GPU market- but when their only viable competition on this level is themselves it is foolish of them to force themselves to reduce margins on a part before their is need to do so.
QFT.
Anyhow, when the new NV5 silencers for the 7800s are released, not only the temps will reduce by 15 degrees average (basing the temps on the Asus 7800GTX top edition with the built in NV silencer), but reaching 500 will be no sweat. The ultra will just be a waste of a product name, time, money and effort.
Originally posted by: BenSkywalker
You are failing to look at the entirety of the situation in this assumption. The pros and cons of such a move need to be weighed when considering such a move. First off is the cost associated with having to special bin the fastest testing chips- but we will assume that is a given. Next you have to line up RAM suppliers to feed slightly higher bandwidth ships to you as you don't want the Ultra finishing in a dead lock with the GTX at any point- so another supply line needs to be opened up or expanded. Then you have to deal with all of the various marketing, offering the additional SKUs to retailers etc(obviously the OEMs would have a big hand in this too). This may not add up to an enormous amount for a company the size of nVidia or their OEMs however it is additional costs incurred. So say all of that is free and takes no time- they can wave a magic supply and distribution wand for their new parts for arguments sake. Where do you launch the Ultra in terms of pricing?
I'll give your line of thought the advantage here and say that we are going to launch the Ultra @$599 MSRP so we can retain very high margins on the part. Of course the only people jumping at this would be the extreme whacko type(aka Rollo
This is where we hit our major problem with the entire scenario. nVidia has just launched a new part that only competes with its prior part that had no competition. This means that nVidia must lower the price on the lower cost higher yield GTXs and GTs to make room for the Ultra. They take their higher volume parts and reduce their yields for what benefit- the extremely small minority that would upgrade from a GTX to an Ultra? If they lost just $20 margin on the GTXs and GTs that would be a loss of millions of dollars- do you honestly think that they would manage to recoupe those costs with the amount of sales that they are going to secure with a marginal speed bump part? Not likely at all.
As a competitor to ATi's flagship the merit may be argued- the Halo effect is very pronounced in the GPU market- but when their only viable competition on this level is themselves it is foolish of them to force themselves to reduce margins on a part before their is need to do so.
With no competition, Nvidia can charge any price they feel they can get. Nothing there to force them to sell the gtx at a reduced price. Except to some degree the fire sale of all the x800 - x850 stuff has had a lowering effect on the market. These sales have left Nvidia with the choice of dropping the prices of the 6800 line to compete in the upper mid-range or making the gt - gtx a more reasonable jump in cash to intice customers up a notch. My point is that if Nvidia could make anything more than a benchmark special - it would be out the door at a very nice premium. Ati has made a lot of mistakes, but not being able to produce a reasonable run of an ultra is not one of them. I would assume for Nvidia to be able to up clocks they need to go to a 90m process or big time cooling. There is always the rumour that 8 pipes are hidden on the gtx - but who knows maybe yields are bad.
Yes their is Do your realease a card at $700 just because you are the only competitor. They just found out they could charge $600 dollars for a video card and get away with it, but they also found out that a decent supply on the highend video card will lead to the market working it down to $500 or did you forget the MSRP is still $600 on the GTX.
Lets Compare the 6800 Ultra or the 850XT PE. Both were MSRP at $550 both sold almost there whole stock for a year at $700+ but both were very limited yield and supply chips that couldn't be found anywhere. Once Retailers needed to move stock (7800GTX) because of good supply the free market value worked its way from $600 to $500 in almost record time. They in essence found that their isn't a market for decent supply $700 dollar or $600 Video cards, in fact while they may be able to sell it a little more expensive then their previous chip it would be unlikely it would stay at MSRP and that the only thing it would do is cut into the sales of another high margin decent production chip.
We watched a horrible time for computer users everywhere durring the X800, 6800 generation. We saw both companies struggle to produce high end parts in quantity this caused the value of such cards sky rocket. Even with ATI still struggling the sheer numbers of NV high end chips would force prices down. All that would happen with an Ultra realease is it would cause them to reprice every GPU they sell lower.
Nope, totally wrong. Nvidia is now getting around $1000.00 for a top line vid card set up.
This includes sli premium for mb. Also buyer is paying more for ps. They have successfully doubled what the suckers will pay for top line! The competition forcing the price of the gtx down is all those x850 and x800 xt pe going for fire sale prices. Nividia needs to compete in the upper midrange market. If Nvidia or Ati had a card that is competative with sli at $'s per fps at $700.00 - it would have a market.
So what could an ultra or GF7 refresh be? Could this be the list:
1) Possibly unlock 8 hidden pipes.
2) Switch to a 90nm version of nv47 for higher clocks/more pipes
3) Switch to 2-slot cooling/more voltage to clock current GF7 to 500+
On top of this, what is the likely ram spec? 512MB @ 1.4ghz?
1) Possibly unlock 8 hidden pipes.
2) Switch to a 90nm version of nv47 for higher clocks/more pipes
3) Switch to 2-slot cooling/more voltage to clock current GF7 to 500+
On top of this, what is the likely ram spec? 512MB @ 1.4ghz?
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter