Rumor: Price Cuts on GTX660Ti series coming next week

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
the 4890 was not out yet as this was in 2008. I just looked it up and my card was $230 and had a $40 rebate making it $190. it also came with some Tom Clancy game but newegg ran out of that promo and gave me another $25 credit. so really the card cost me $165 shipped.

:thumbsup: congrats on that awesome deal. The global point I am making is that most people here would agree that outside of these special situations you are describing, AMD offered better price/performance with HD4800/5800/6900 lines and for longer periods of time. Even using Newegg price engine, you can see that GTX260 216 was not $165 on average, not even in Summer-Fall 2009. If you are leading AMD's new strategy, what would you do? It's been 3 years of competing on price/performance and not winning market share. There is another strategy in business:

"First mover advantage". You offer superior technology ahead of your competitor. This allows you to be the leader temporarily and dictate high prices (i.e., high profits). When the competitor arrives, you adjust prices accordingly. This works especially well in the technology industry since early adopters pay a premium for the latest technology and as technology matures, it becomes obsolete, forcing inevitable price drops. IMO HD7000 is the first execution of this new approach as a result of new management. The end result:

- AMD sold their entire line for higher prices most of this generation because NV was 3-8 months late;
- AMD is still selling HD7950/7970 for $299-319/419-449 (this would have normally been their launch prices, or even lower)

You can look at it either way:

1) Why was AMD able to launch at high prices? Because NV was late and ATI used to sell at high prices without problems when they had fast cards.

2) Why is AMD still able to sell cards at what are still high prices vs. their historical GPU price for 3000/4000/5000/6000 series? Because NV under-delivered this generation by launching mid-range Kepler (NV loves this though) and not launching a single 28nm desktop card worth buying under $300 until August. Who is to blame for this? Lack of competition from NV allowed AMD to dictate high prices for HD7750/7770/7850/7870 for 6 months.

3) Why did AMD finally decide to change strategy? Because NV users weren't switching anyway.

Unless NV blows AMD away, it's probably the end of the road for price/performance of the past. I fully expect AMD to reuse the first mover advantage strategy again:

HD8950 for $400-449
HD8970 for $500-549
 
Last edited:

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
Then I'm not sure what this is about:

The numbers I posted are the latest available, Q2-2012.


No. Ya didn't:

AMD 28,155 units 22.8% share, -1.8% from Q1, -1.3% from Q2-2011
nvidia 18,950 units 15.3% share, +0.2% from Q1, -4.5% from Q2-2011
intel 75,80 units 61.4% share, +1.7% from Q1, +6.0% from Q2-2011

discrete
AMD 14,000 units, 42.9% share, -2.1% from Q1, -2.5% from Q2-2011
nvidia 18,600 untis, 57.1% share, +2.1% from Q1, +2.5% from Q2-2011


I'm going to bed. I'll see if you decided to post some honest numbers that reflect our current 28nm sales tomorrow.
Nite good buddy.
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
No. Ya didn't:

Q2-2011 is only used for the last #s in his post to represent a YoY quarterly % change. You can show QoQ or YoY on a quarterly basis. The actual number of units sold in last quarter is what he linked in the post (those are on the left). So it's 100% valid. The YoY quarterly % change is often used by analysts in finance (how many units did you sell in Q2 2011 vs. Q2 2012 to account for that quarter's growth rate from year to year). QoQ (Q2 2012 vs. Q1 2012) is used to measure the growth based on seasonality of a particular business. For example a company selling Apple or Pumpkin pies may do 80-90% of their annual sales in Q4 2012 (Thanksgiving). Then you'd want to compare Q3 2012 vs. Q4 2012 vs. Q1 2013 to see how much the seasonality impacts your business to help you manage your inventories for the following year.
 
Last edited:

AnandThenMan

Diamond Member
Nov 11, 2004
3,991
627
126
No. Ya didn't:

AMD 28,155 units 22.8% share, -1.8% from Q1, -1.3% from Q2-2011
nvidia 18,950 units 15.3% share, +0.2% from Q1, -4.5% from Q2-2011
intel 75,80 units 61.4% share, +1.7% from Q1, +6.0% from Q2-2011

discrete
AMD 14,000 units, 42.9% share, -2.1% from Q1, -2.5% from Q2-2011
nvidia 18,600 untis, 57.1% share, +2.1% from Q1, +2.5% from Q2-2011


I'm going to bed. I'll see if you decided to post some honest numbers that reflect our current 28nm sales tomorrow.
Nite good buddy.
I think you need to understand what you are reading there. But feel free to post the current numbers as you believe them to be.
 

Ferzerp

Diamond Member
Oct 12, 1999
6,438
107
106
You've already been chastised by a mod once before for being intentionally deceptive with that same exact data set no less.

Still at it I see. And still unable to separate the GPU from CPU+igpu markets.
 

AnandThenMan

Diamond Member
Nov 11, 2004
3,991
627
126
ALL I'm trying to say is going by what each vendor is currently shipping (and the numbers don't radically change if you go back 1-2 years) then it is reasonable from said numbers to extrapolate that AMD is eating up more 28nm wafers than Nvidia. One possible exception is if NVDA is having atrocious yields and dumping most of the wafers in the trash. That would explain apparent constraints, although that does not mesh with actual availability.

What I see is JHH cleverly shifting the issue from their lack of rolling out on a new node in a timely manner versus AMD, to stating that things are good except they need more wafers. Sounds much better than saying, yes we are late and have found the new node a challenge, doesn't it?

One thing people do need to understand is Nvidia makes good profits on a consistent basis, so they can afford to lose sales short term, with the plan of making it up over the course of a generation of products. The danger there is letting your competitor consistently beat you to a new gen of product, you can fall behind the curve and never catch up. AMD knows this all too well when it comes to competing with Intel. What saves Nvidia is AMD's lack of brand recognition, and marketing dollars. If AMD was awash in cash Nvidia would be in trouble.
You've already been chastised by a mod once before for being intentionally deceptive with that same exact data set no less.

Still at it I see. And still unable to separate the GPU from CPU+igpu markets.
I was never intentionally deceptive, I've been very clear in what I am saying, and what the numbers mean to the discussions at hand. Plus the numbers are there for all to see and are germane to this discussion about 28nm GPU supply and pricing. Also, the integrated/on-die stuff will be transitioning to 28nm as well which consumes wafers of course.
 
Last edited:

Ferzerp

Diamond Member
Oct 12, 1999
6,438
107
106
ALL I'm trying to say is going by what each vendor is currently shipping (and the numbers don't radically change if you go back 1-2 years) then it is reasonable from said numbers to extrapolate that AMD is eating up more 28nm wafers than Nvidia. One possible exception is if NVDA is having atrocious yields and dumping most of the wafers in the trash. That would explain apparent constraints, although that does not mesh with actual availability.

What I see is JHH cleverly shifting the issue from their lack of rolling out on a new node in a timely manner versus AMD, to stating that things are good except they need more wafers. Sounds much better than saying, yes we are late and have found the new node a challenge, doesn't it?

One thing people do need to understand is Nvidia makes good profits on a consistent basis, so they can afford to lose sales short term, with the plan of making it up over the course of a generation of products. The danger there is letting your competitor consistently beat you to a new gen of product, you can fall behind the curve and never catch up. AMD knows this all too well when it comes to competing with Intel. What saves Nvidia is AMD's lack of brand recognition, and marketing dollars. If AMD was awash in cash Nvidia would be in trouble.


What your data shows is that even when nvidia didn't have a competitive part and was still shipping last gen cards, AMD still sold less.

How does that help your argument exactly?
 

AnandThenMan

Diamond Member
Nov 11, 2004
3,991
627
126
What your data shows is that even when nvidia didn't have a competitive part and was still shipping last gen cards, AMD still sold less.

How does that help your argument exactly?
I've made my position as clear as I can, I respect if you don't agree with it.
Those numbers don't break down the price-points and sales from AIB's to consumers though.
STEAM numbers have always seemed wacky to me, don't know how much they can be trusted honestly.
 

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
I've made my position as clear as I can, I respect if you don't agree with it.

STEAM numbers have always seemed wacky to me, don't know how much they can be trusted honestly.

For me, may offer some insight but the key is official financial info utilizing Jon Peddie Research or Mercury Research.
 

AnandThenMan

Diamond Member
Nov 11, 2004
3,991
627
126
Of course they did because they liked the numbers. :D Press releases that spins things to make the company look good, hardly news.
For me, may offer some insight but the key is official financial info utilizing Jon Peddie Research or Mercury Research.
Agreed. Although I admit I don't always understand or appreciate the gritty details of the financials.
 

Jaydip

Diamond Member
Mar 29, 2010
3,691
21
81
How can AMD outsell NV in discrete space where the later has ~60% market share is beyond me.However if you include APUs the situation will change naturally.
 

AnandThenMan

Diamond Member
Nov 11, 2004
3,991
627
126
How can AMD outsell NV in discrete space where the later has ~60% market share is beyond me. However if you include APUs the situation will change naturally.
They don't, the numbers clearly say so. But the whole point is wafer demands (and total GPUs shipped) are not only for discrete products over the life of a specific node, if they were it may change the discussion. But even if we say that AMD is 40% of 28nm (ignoring any other TSMC customers) and NVDA the rest, would that cause a late launch of new node GPUs?

The real qualm I have with the claim that Nvidia is supply constrained boils down to that one thing. If they were, would that cause them to lag behind their competitor to launch new node products? Or would it just mean said products would be in short supply? I say the latter.

On the price cuts, I predict Nvidia will cut prices on all the 6xx SKUs in the next few weeks.
 

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
$20 price cut is still too small.

Doesn't matter to me, the HD 7850 went into my machine on Friday. It was Nvdia's sale to lose and they did.
 

bunnyfubbles

Lifer
Sep 3, 2001
12,248
3
0
At what price would it make sense to purchase a 660 TI?

I want one, honestly, but I can't justify it no matter how hard I try because my eyes keep going over to the 7950 cards.

~$250, although I'd say ~$220-230 is the true sweetspot.

however if you're looking forward to Borderlands 2 and have yet to preorder a copy, the 660TI is already a fair deal
 
Feb 19, 2009
10,457
10
76
Too close to the 670? The 670 costs 33% more for what, 10% more performance? If anything the 660 Ti is the sweet spot in NV's lineup right now.

It's $300 vs 400, although you can find custom 660ti and 670 to be even closer. It's 25% slower with high AA/1440p+. I don't care about no AA results in this segment.

A mid-range should be much better value than high end, not identical value, being the case, there's simply NO value. Remember the HALO tax? Here comparing a 660ti to a 670 there seems to be none, thus, the 670 is great value.
 
Last edited:

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
It's because its priced too close to the 670, which is IMO the best card this generation for non-overclockers.

IMO, it's because it's priced too close to the 7950 and too far from the 7870. The 192bit mem bus just limits the card too much for the $300 price point.

Considering most 660 ti's are aftermarket, not reference, I'm not too sure if this isn't just more than lip service if aftermarket cards are going to be ~$320. At that price I would definitely by the 7950. :\
 

Jaydip

Diamond Member
Mar 29, 2010
3,691
21
81
They don't, the numbers clearly say so. But the whole point is wafer demands (and total GPUs shipped) are not only for discrete products over the life of a specific node, if they were it may change the discussion. But even if we say that AMD is 40% of 28nm (ignoring any other TSMC customers) and NVDA the rest, would that cause a late launch of new node GPUs?

The real qualm I have with the claim that Nvidia is supply constrained boils down to that one thing. If they were, would that cause them to lag behind their competitor to launch new node products? Or would it just mean said products would be in short supply? I say the latter.

On the price cuts, I predict Nvidia will cut prices on all the 6xx SKUs in the next few weeks.
Difficult to say as NV also has a strong professional market and I imagine that take precedence over consumer gpus.For example Tesla K20 will probably take a few iterations to make it perfect.Also if you sell more gpus, naturally you will need more gpus for warranty purposes.
 

sontin

Diamond Member
Sep 12, 2011
3,273
149
106
You can look at it either way:

1) Why was AMD able to launch at high prices? Because NV was late and ATI used to sell at high prices without problems when they had fast cards.

nVidia was not late. They needed time to produce enough chips - thx to AMD. Their overpriced strategy was one side of the success of Kepler.

2) Why is AMD still able to sell cards at what are still high prices vs. their historical GPU price for 3000/4000/5000/6000 series? Because NV under-delivered this generation by launching mid-range Kepler (NV loves this though) and not launching a single 28nm desktop card worth buying under $300 until August. Who is to blame for this? Lack of competition from NV allowed AMD to dictate high prices for HD7750/7770/7850/7870 for 6 months.

AMD is only able to sell at these high price points because nVidia is waiting for supply from TSMC. Everytime nVidia launched a new Chip or new SKUs they adjusted their prices. There are rumors that the price of the GTX660 would be $229. So i guess we can expect new prices for 78xx and maybe 7950 cards.

It was really a great strategy to "go premium". :sneaky:
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
well to be fair the gtx660 ti launched at 20% more than the gtx560 ti launched at but is 50% faster. the 7870 and 7850 increased their price at the same level of increasing their performance. so at least with nvidia you got 30% more performance for your dollar where with AMD you got basically zero.

comparing the $300 gtx660 ti to the $229 gtx460 seems odd but even doing that the gtx660 ti offers 90% more performance for 30% more money. so if anything that's a pretty huge increase in performance per dollar at launch prices.

bottom line is that no matter how bad the Nvidia prices look, they still gave an increase in performance for your money. it was AMD's launch prices that were a freaking joke and made no increase at all in performance per dollar.

GTX570 was launched in December 2010 at $349.00, GTX660Ti has the same performance on average for $299 after a year and a half. Not to mention you could find GTX570 at $299 5-6 months ago.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
I've said it before, but you can't compare pricing on cards released 9 months after the node launched with cards that were earliest out of the gate. As the process matures prices go down. Even a month or two later with a smaller chip isn't directly comparable.

Smaller chip + more mature process = lower costs.
 

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
nVidia was not late. They needed time to produce enough chips - thx to AMD. Their overpriced strategy was one side of the success of Kepler.



AMD is only able to sell at these high price points because nVidia is waiting for supply from TSMC. Everytime nVidia launched a new Chip or new SKUs they adjusted their prices. There are rumors that the price of the GTX660 would be $229. So i guess we can expect new prices for 78xx and maybe 7950 cards.

It was really a great strategy to "go premium". :sneaky:

It also doesn't help that AMD dropped GloFo in favor of TSMC to manufacture their APU's at the last possible minute, taking up even more 28nm die space. AMD claimed that GloFo wouldn't be able to supply the demand on 28nm.
http://www.pcr-online.biz/news/read/amd-dump-former-chipmaker-for-new-apus/027566
 
Last edited: