Go Premiums!
Ignore the fact that this is the better buy in the price bracket for some buyers. Nope, we're AMD loyalist. Gotcha.
Shill on brothers!
Go Premiums!
Sure, if you overload the 8800GTX/9800GTX memory limits, the GTX280 crushes it.
Sure, but you could also use the example of 65nm->40nm as well, after all it is more recent.
If you want to make the argument that AMD's architecture isn't as refined as possible on 28nm, OK, but more proof than an assumption is needed.
There you go. I didn't buy any 7970 or GTX580 or any GPU for over >$350 in my all life.
So, the people you need to convince is the ones that buy or are considering buying these cards.
No one is defending price premiums, people are just explaining reality - when there is competition, supply and demand rule the prices, unless there is price fixing which is illegal (dumping is illegal at well)
If you have any legal document, where AMD guaranteed you they would never price their cards over a certain value, I encourage you to take legal action.
Otherwise is the price of the 7970 justified considering market conditions?
If it is, what is wrong?
Are you hopping people refuse to buy the 7970 and buy the GTX580 instead as a protest? The GTX580 is even more overpriced.
And people already buy cheaper cards in higher volumes.
You want to explain how those physics work when some people aren't even breaking 1050MHz on their 7970's? :whiste: Again, you're using assumptions as fact, and you need more proof than that if you're going to argue a point.Grab a 7970 and overclock it. Physics have a funny way or working, they don't always align with the whims of wishful thinking
My point is the product they released isn't as refined as they could have released it at. That exact product back up my stance. If you are arguing against what I am saying right now, you need to explain why the 7970s are insane overclockers. Funny thing happens when you overclock them to their potential- performance gains fall in line with where I am saying they should be. Hmmm, imagine that
I can't argue with anything here. Defending value against a heavily premium priced, year old, 40nm SKu.
Since Techpowerup is the choice it seems to showcase not only relative performance for the HD 7970 here is the performance per dollar:
http://www.techpowerup.com/reviews/AMD/HD_7970/30.html
I don't agree with their metrics based on it uses relative performance, which is watered down. I think the HD 7970 offers more premium value than a GTX 580, specifically with their OC scaling.
You want to explain how those physics work when some people aren't even breaking 1050MHz on their 7970's?
Again, you're using assumptions as fact,
I'm also curious, what did you think of NVIDIA's GTX 400 cards in regards to their refinement?
Actually, that's basic math. 7970 is already 35% faster, add 12% to that level and it breaks 50%(I know some people are going to question it so I'll write it out)-
60 FPS- Baseline
81 FPS- 35% faster
90.7 FPS- 12% faster then that-
Cumulative- 51% faster
Are you hopping people refuse to buy the 7970 and buy the GTX580 instead as a protest? The GTX580 is even more overpriced.
Actually, that's basic math. 7970 is already 35% faster, add 12% to that level and it breaks 50%(I know some people are going to question it so I'll write it out)-
60 FPS- Baseline
81 FPS- 35% faster
90.7 FPS- 12% faster then that-
Cumulative- 51% faster
Your worst case example falls into my range. I could also point out that a chip that didn't clock higher coudl have been binned as a lower end part if they were pushing hard as nV and AMD always do, but the worst case numbers you gave still hits the 50% mark.
What do you mean? In what area? In terms of performance they had no issues.
The 280 has a 50% edge @1280x1024 in Crysis which is the page I linked to(COD4 50% @16x12, Oblivion almost 60% at 16x12, almost 70% @16x12 in Bioshock- lowest resolutions tested still holds my point), I'm going to go out on a limb and say that the 8800GTX wasn't close to memory limited at that setting.
You could have used the GTX280, or considering the 45nm was canned, I guess the 55nm->40 nm also represents a full node hump so you could have used the GTX480, GTX285 and the 4890 and 5870.Honestly I tried, AT didn't include the 8800 parts in their 480 review. That actually would have proved my point even better.
Considering 28nm is a substantial and significant node change, one may expect to see much larger performance/value and to redefine performance/value instead of using a year old, premium priced sku as an objective barometer of pricing. Virtually nothing has been redefined and using the 40nm status quo performance/value.
You're making a different point without completing the thought on the other. Please stick to one point before we move on. You had said:Actually, that's basic math. 7970 is already 35% faster, add 12% to that level and it breaks 50%(I know some people are going to question it so I'll write it out)-
60 FPS- Baseline
81 FPS- 35% faster
90.7 FPS- 12% faster then that-
Cumulative- 51% faster
Your worst case example falls into my range. I could also point out that a chip that didn't clock higher coudl have been binned as a lower end part if they were pushing hard as nV and AMD always do, but the worst case numbers you gave still hits the 50% mark.
To which I replied:My point is the product they released isn't as refined as they could have released it at. That exact product back up my stance. If you are arguing against what I am saying right now, you need to explain why the 7970s are insane overclockers. Funny thing happens when you overclock them to their potential- performance gains fall in line with where I am saying they should be. Hmmm, imagine that
12% overclocking is not considered "insane," mediocre is more accurate. Also, I'm not sure where your seemingly arbitrary 50% comes from, but that's besides the point. If you want to correlate lack of refinement with the ability to overclock, then there has to be more consistency in the results than the wide spread we currently see. I'm not arguing that the 7970 is as refined as it could be (it's a new architecture on a new process, no engineering team is that good), but your proof that it isn't is inaccurate to say the least.You want to explain how those physics work when some people aren't even breaking 1050MHz on their 7970's?
Well, using the same standards, would you consider Fermi "refined," which you assert the 7970 is not.What do you mean? In what area? In terms of performance they had no issues.
How do you know that is the case?
At the moment, I believe that is just an assumption, or do you have data about other current products on the market?
It's just my vocal view. Will see how accurate it is.
But if I told you X GPU has 33% more shaders, 33% more texture units, same rops, 50 % more bandwidth and it was clocked 5% faster than GPU Y, and then told you GPU X was 40% faster than GPU Y, you wouldn't be surprised, would you?
In my view the most surprising is that the chip is almost the same size and not much smaller than the 6970. Is that due to extra GPGPU capabilities, simply limitation on shrinking stuff or the 7970 does indeed pack extra CU but they are disabled for yields?
And if is a GPGPU toll, will NVIDIA be able to release a chip that is smaller than the 7970 but be as fast/almost as fast? And will NVIDIA sell it for less to undercut the 7970 or will NVIDIA price it according to its relative performance to the 7970 current price?
Neither Kepler nor Tesla has been released, and in that second quotation he compared unreleased GPU's against other unreleased GPU's. What's so difficult about that? Again, if you have a quotation that actually proves what you're accusing IonusX of, post it. Otherwise stop derailing the thread and wasting forum space. End of discussion.
The benches I've seen show the 7970 getting a 74% increase over it's stock performance once overclocked. At 1125 it's getting between a 12 and 16 percent increase in performance assuming cpu limitations aren't occurring.
The baseline percent increase over the 580 varies with reviews and resolution though so it's hard to say in absolutes what the difference really is unless you're talking specifically about a single resolution.
Price: http://semiaccurate.com/2012/01/23/exclusive-and-the-nvidia-keplergk104-price-is/
$299
Definitely nVidia's RV770.
I wonder if they will let this new mid-range monster do TRI-SLI or limit it to 2 card SLI.
Which also saves driver team time and money in chasing those elusive scaling driver tweaks as well as incentive for the power chasers to get the next series up.