Post in thread '10 years of Nvidia Video cards. Ultra high end ,high end, mid range, lower end, and

Toggle sidebar Toggle sidebar
happy medium

happy medium

Lifer
Jun 8, 2003
14,387
480
126
 
Last edited by a moderator:
  • Like
Reactions: Sweepr
I

iiiankiii

Senior member
Apr 4, 2008
759
47
91
You forgot the halo card for the Kepler generation; the Titan Z; clocking in at a modest $3000.
 
E

Erenhardt

Diamond Member
Dec 1, 2012
3,251
105
101
What is the difference on arch level between gtx 680 and 780ti that you put it into two separate "generations"?
And like iiiankiii noticed, TitanZ is missing.
 
R

Ranulf

Platinum Member
Jul 18, 2001
2,348
1,165
136
"One thing I noticed is # 4, the mid range ,stayed right around $250."

Heh. Yeah, midrange in price and thats about it for many of their cards. Certainly everything after the 560 or 660.
 
P

Piroko

Senior member
Jan 10, 2013
905
79
91
First off, good job!

happy medium said:
Now we can compare cards and their price/performance tier, for the past 10 years.
Are we paying significantly more now than in the past, for the same tier of performance?
Click to expand...
Counter question, is the same tier of performance still giving you the same experience in the product stack? Case in point:

One thing I noticed is # 4, the mid range ,stayed right around $250.
Click to expand...
In the 7000 Series #4 had ~60% of the shader performance and 80% of the memory bandwith of #2, with half the memory capacity.
In the current Geforce 10 series #4 has no more than 50% of the shader performance and only 60% of the bandwith of #2, albeit with 75% of its capacity.

So, in my opinion #4 of this generation looks weaker than the #4 SKU from 10 years ago. I did cheat by not looking at #1, but that tier had really inconsistent performance and pricing over the past 10 years. I also didn't look at the stellar 8000 series either.
 
N

nurturedhate

Golden Member
Aug 27, 2011
1,742
673
136
Piroko said:
So, in my opinion #4 of this generation looks weaker than the #4 SKU from 10 years ago. I did cheat by not looking at #1, but that tier had really inconsistent performance and pricing over the past 10 years. I also didn't look at the stellar 8000 series either.
Click to expand...

Exactly, though looking at #1 would have only exasperated the issue further because it itself is a cutdown version of a part (GP102) that isn't even at the top of the stack (GP100). With that known #4 slides even further down when compared to prior generations within their own stack.
 
N

nurturedhate

Golden Member
Aug 27, 2011
1,742
673
136
happy medium said:
the reviews gave me the impression that is was a compute card not really a gaming card.
I also left out some other cards for example Asus Mars cards, and other high price dual gpu cards that were more like specialty cards.
Click to expand...
So you omitted data points to validate a preconceived conclusion instead of allowing the data to speak for itself.
 
  • Like
Reactions: [DHT]Osiris, Final8ty, Grooveriding and 3 others
N

nurturedhate

Golden Member
Aug 27, 2011
1,742
673
136
+
happy medium said:
say again?
and your welcome, this took hours to do.

From the review, this is what made me leave the Titan Z out.

"That said it’s clear from NVIDIA’s presentations and discussions with the company that they intend it to be a compute product first and foremost"
Click to expand...

The you need to omit the rest of the Titan line especially Titan Black, but you didn't. You also omitted the original Titan but included Titan Black and the maxwell/pascal Titans. You are manipulating data to validate a preconceived conclusion.
 
N

nurturedhate

Golden Member
Aug 27, 2011
1,742
673
136
happy medium said:
fixed. I don't see why the gtx700 series is so important to you, there are many other series also.
I didn't reach a conclusion. its just a list showing performance top to bottom.

It's a list to refresh Anandtech members memory's. Good to stop the spread of misinformation.
Click to expand...
It has nothing to do with the 700 series and everything to do with omitting data to validate a preconceived conclusion. And you have reached a conclusion. We both know, along with most of this forum, that the whole point of this thread is your posts in this thread https://forums.anandtech.com/threads/nvidia-q3-revenue-up-53.2491579/ and you clearly and repeatedly stated your opinion there. Again, you are attempting to hide or omit.
 
  • Like
Reactions: Final8ty, Det0x and linkgoron
AtenRa

AtenRa

Lifer
Feb 2, 2009
14,001
3,357
136
1. High-End Single Chip (500-600mm2) Video Cards price increased from $500-650 in the Fermi era 2010 (GTX 480/580 529mm2) to $1200 with Titan X Pascal (471mm2)

2. Middle-End Single Chip (300-350mm2) Video Cards price increased from $200-250 in the Fermi era 2010 (GTX 460/560Ti 332mm2) to $600-700 with GTX 1080 (314 mm2)

3. Low-End Single Chip (200-250mm2) Video Cards price increased from $130-150 in the Fermi era 2010 (GTS 450 238mm2) to $250-300 with GTX 1060 (200 mm2)

4. Entry-Level Single Chip (100-120mm2) Video Cards price increased from $100 in the Fermi era 2011 (GT 440 116mm2) to $139 with GTX 1050 (132 mm2)

What NVIDIA has manage to do is astonishing, they managed to increase the price of the Midle-End cards ($200-250 price point) to $400-700 and also increase the volume by 2-3X of those cards sold.




The following is from Q2 2010 when Fermi (GF100) was released.

Pay attention to the table bellow the yellow arrow, its the Desktop segment by price.

The $300+ segment was the High-End, it represented only 7,2% of the entire volume of both ATI(7,9%) + NVIDIA (5,2%) sales.

I dont have the numbers for 2016 but im sure the $300+ segment now constitutes more than 30-40% (or even more ??) of NVIDIAs sales volume.
Add the ASP increase of that segment and we reach record revenue in Q3 2016 (FY Q3 2017)

20ixkhs.jpg
 
  • Like
Reactions: Head1985
I

iiiankiii

Senior member
Apr 4, 2008
759
47
91
happy medium said:
the reviews gave me the impression that is was a compute card not really a gaming card.
I also left out some other cards for example Asus Mars cards, and other high price dual gpu cards that were more like specialty cards.
Click to expand...

ok. Then leave out the Titan series. Those, too, are "compute" cards. You can't have it both ways.
 
H

Head1985

Golden Member
Jul 8, 2014
1,864
688
136
AtenRa said:
1. High-End Single Chip (500-600mm2) Video Cards price increased from $500-650 in the Fermi era 2010 (GTX 480/580 529mm2) to $1200 with Titan X Pascal (471mm2)

2. Middle-End Single Chip (300-350mm2) Video Cards price increased from $200-250 in the Fermi era 2010 (GTX 460/560Ti 332mm2) to $600-700 with GTX 1080 (314 mm2)

3. Low-End Single Chip (200-250mm2) Video Cards price increased from $130-150 in the Fermi era 2010 (GTS 450 238mm2) to $250-300 with GTX 1060 (200 mm2)

4. Entry-Level Single Chip (100-120mm2) Video Cards price increased from $100 in the Fermi era 2011 (GT 440 116mm2) to $139 with GTX 1050 (132 mm2)

What NVIDIA has manage to do is astonishing, they managed to increase the price of the Midle-End cards ($200-250 price point) to $400-700 and also increase the volume by 2-3X of those cards sold.




The following is from Q2 2010 when Fermi (GF100) was released.

Pay attention to the table bellow the yellow arrow, its the Desktop segment by price.

The $300+ segment was the High-End, it represented only 7,2% of the entire volume of both ATI(7,9%) + NVIDIA (5,2%) sales.

I dont have the numbers for 2016 but im sure the $300+ segment now constitutes more than 30-40% (or even more ??) of NVIDIAs sales volume.
Add the ASP increase of that segment and we reach record revenue in Q3 2016 (FY Q3 2017)

20ixkhs.jpg
Click to expand...
I think its because they didnt focus on Gx106 part much.GTX670/970 was so much better than GTX660 and GTX960 so nobody buys them thats why people buy in 300+ segment.GTX670 was best x70 ever made and GTX660 was crap compare to it.
GTX960 was even more crap than GTX660 soo....

GTX1060 6GB is first good x60 card after kepler twist.
 
  • Like
Reactions: Final8ty
L

linkgoron

Platinum Member
Mar 9, 2005
2,293
814
136
That's nice. You average the 780ti and the Titan. You ignore the Titan Z. That's a great way to reach a conclusion.

The nvidia website calls the Titan Z:
GeForce® GTX TITAN Z is a gaming monster, the fastest card we’ve ever built to power the most extreme PC gaming rigs on the planet. Stacked with 5760 cores and 12 GB of memory, this dual GPU gives you the power to drive even the most insane multi-monitor displays and 4K hyper PC machine
Click to expand...
Sounds like a gaming card.

Anyway, you include the 690 and 590 even though people have repeatedly mentioned the rise of the price of single cards. You ignore the Titan Z even though you included the 590 and 690. You have reached a conclusion and then removed and ignored evidence that does not align with your conclusion. You keep ignoring what people are telling you.

The new 80 cards are usually smaller (or significantly cut) cards. The 680, 980, 1080 and all had a x4 markup, which was in the past used for the midrange cards. They also have a significantly smaller die than previous x80 cards, as mentioned by AT in the 680 review. The 780 was a significantly cut chip, only 80% of the full die GK110.

Your numbers align with what people are saying - nvidia has pushed the lower end cards up in the stack. That's how the top single cards have become 1000$+ Titans. You're showing it yourself - the x80 has turned into a middle of the stack single card, while still increasing in price. It used to be the top of the single card stack, and is now way way behind the Ti and Titan single card lines, while still increasing in price.
 
S

Samwell

Senior member
May 10, 2015
225
47
101
What people doesn't seem to look at are the production cost. I just made a list according to wafer prices from here http://img.deusm.com/eetimes/2016/06/1329887/Handel1.png

For 16 nm i assumed the same price as 28nm, some reports see higher cost, others a bit lower so i took the middle
Fermi 3bil 58,2 $
Gf104 1,95 38 $
Gk110 7,1 92,3 $
Gk104 3,5 46$
Gm204 5,1 68 $
Gm200 8 Bil 104 $
Gp104 7,2 94$
Gp102 12 156$
These cost don't take into account yield and probably they are meant at the start of the node as 16nm should get better over time. So don't take them as absolute numbers. But you clearly see the problem we have nowadays. A gp104 just costs way more to produce than a fermi back then. Of course nvidia also took higher margins, but the cost are really a tough problem at this and smaller nodes.That's also the reason you won't see amd with really low prices, the good times are over.
 
Last edited:
L

linkgoron

Platinum Member
Mar 9, 2005
2,293
814
136
happy medium said:
the 590 and 690 are single cards.
looks like a single card to me.
Click to expand...

You know what I mean. The 690 is dual-gpu, as are the 590 and Titan Z. At this point, it's obvious that you are just trolling. You obviously understand now that you've been played by marketing (which is their job), you're just too proud to admit it.

Good luck with the thread.
 
  • Like
Reactions: HiroThreading
Keysplayr

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
nurturedhate said:
It has nothing to do with the 700 series and everything to do with omitting data to validate a preconceived conclusion. And you have reached a conclusion. We both know, along with most of this forum, that the whole point of this thread is your posts in this thread https://forums.anandtech.com/threads/nvidia-q3-revenue-up-53.2491579/ and you clearly and repeatedly stated your opinion there. Again, you are attempting to hide or omit.
Click to expand...
Wow, nurtured hate. You aren't kidding.
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom