[TPU] Nvidia prepares "price cuts across it's entire lineup"

[GaiaHunter]

I never said it was 8%... :hmm:

It is indeed more like 20-25%, depending on resolution and AA/AF. Which is, of course, not 30%.

I wonder how many 7970 actually run for more than a few minutes at 925 and not at 1000+.

 

[96Firebird]

I wonder how many 7970 actually run for more than a few minutes at 925 and not at 1000+.

Ok RS...

AMD should have released the 7970 at 1000+ in December then!

 

[SiliconWars]

I never said it was 8%... :hmm:

It is indeed more like 20-25%, depending on resolution and AA/AF. Which is, of course, not 30%.

The point stands. The 7970 was faster compared to the 580 than the 780 was compared to the 7970 GHz.

All the AMD guys realised that the 7970 wasn't anything special yet the Nvidia guys lost their minds over the 780 at stupid prices. I guess that's what happens when you don't have much to celebrate for years.

 

[raghu78]

I like TPU, I know where to go for the performance summary. And another plus, the site is already in English! My native language! How fun is that?! :awe:

what about hardwarecanucks. they are a reputed site in english. if you want to be in denial suit yourself. the perf gap was 20 - 25% on avg across many reviews. anyway lets not take the thread off topic.

looks like the R9 270X will face off against a GTX 760 SE / GTX 750 Ti , whatever the name nvidia choose.

http://videocardz.com/46347/nvidia-geforce-gtx-750-ti-pictured-detailed

 

[96Firebird]

what about hardwarecanucks. they are a reputed site in english. if you want to be in denial suit yourself. the perf gap was 20 - 25% on avg across many reviews. anyway lets not take the thread off topic.

I've never read their site, but I'm curious as to what makes a site reputable?

And to the bolded part...

It is indeed more like 20-25%, depending on resolution and AA/AF. Which is, of course, not 30%.

We have come full circle! I am proud.

 

[GaiaHunter]

Ok RS...

AMD should have released the 7970 at 1000+ in December then!

Yeah, Intel should also have released the Core2 at higher clocks.
I remember these forums "No, no, Intel OC results don't count only the stock results. On these forums we never overclock!".

850/2000 @ 1.025V (which was stock for my card).

or

http://forums.anandtech.com/archive/index.php/t-2126842.html

Notice how everyone is comparing them at stock... Do you plan to overclock at all? This will play quite a factor in deciding which is better.

Also very fun of you thinking I'm RS.

 

[blackened23]

Ok RS...

So, essentially, you really don't ever have anything useful to say and instead do passive aggressive and condescending posts directed at RS?

Why don't you post something with substance instead of condescending nonsense? You don't really come off as being much different than the extreme AMD fans that populate the CPU forum for whatever reason with your posting style.

Member call outs are not allowed -Shmee

 

[96Firebird]

Hmm, lets look back.

Someone posted something that was false, I call them out on it with evidence. Then everyone else jumps in and some even post links that back me up, while trying to do the opposite.

I love this place!

I also like how GaiaHunter decided to go through my past posts to find quotes which have nothing to do with the discussion at hand. :thumbsup:

 

[blackened23]

It's the internet. Misinformation and off topic posts - It happens. On all sides of every argument that has ever occurred on the internet, not just in the context of video cards. Not saying it's cool or correct, but not a reason to get bent out of shape... That said, I know the feeling, though. Some posts are annoying, been there done that. I can't say I haven't done the same from time to time in terms of getting annoyed and posting as a result of that. I suppose I didn't read the full context of the thread, I just saw a few posts.

Anyway...back to the topic...

puppies.

 

[Erenhardt]

Hmm, lets look back.

Someone posted something that was false, I call them out on it with evidence. Then everyone else jumps in and some even post links that back me up, while trying to do the opposite.

Evidence? EVIDENCE? E V I D E N C E?!
Well... here is an evidence that you are wrong with your evidence and backs up "30%":

There is a term for such specific evidence gathering, Sherlock...

 

[96Firebird]

So I post a performance summary, you post a single game.

I guess the smarter posters of Anandtech can see the difference.

 

[Erenhardt]

So I post a performance summary, you post a single game.

I guess the smarter posters of Anandtech can see the difference.

Ohh.. let me re-phrase:

I like PCLab and, I know where to go for the performance summary based on single source. And another plus, the site is already in Polish! My native language! How fun is that?!

 

[iMacmatician]

I like PCLab and, I know where to go for the performance summary based on single source.

One game is a performance summary?

 

[Erenhardt]

One game is a performance summary?

I find this game to summarize the overall graphics performance well

 

[SiliconWars]

So I post a performance summary, you post a single game.

I guess the smarter posters of Anandtech can see the difference.

You posted the worse case performance summary. I can do that as well with the 780, for example at Hexus where it only beat the 7970 Ghz by 10% - http://hexus.net/tech/reviews/graphics/55445-nvidia-geforce-gtx-780/

Or like the smarter posters of Anandtech, we can use what is the average over most reviews and throw out the outliers.

 

[Jaydip]

So I post a performance summary, you post a single game.

I guess the smarter posters of Anandtech can see the difference.

Everybody can see the difference.The problem is 7970 at launch was underwhelming and people comparing node shrinks should take a deep breath.

 

[SiliconWars]

Everybody can see the difference.The problem is 7970 at launch was underwhelming and people comparing node shrinks should take a deep breath.

Why compare node shrinks? Why not compare die sizes?

 

[Jaydip]

Why compare node shrinks? Why not compare die sizes?

It is stupid that's why.

 

[SiliconWars]

It is stupid that's why.

Why do I get the feeling that "stupid" means anything that makes Nvidia look bad.

Compare transitor count then as that is where the performance comes from. The 580 has 3 billion, the 7970 has 4.31 billion. So the 7970 performed ~25% faster on immature drivers with 44% more transistor budget.

The 780 has 7.1 billion. So the 780 performed ~20% faster on mature drivers with 65% more transistors.

Anyway you look at it - apart from your "node shrink" way - shows that the 7970 was a better upgrade than the 780 was.

 

[Jaydip]

Why do I get the feeling that "stupid" means anything that makes Nvidia look bad.

Compare transitor count then as that is where the performance comes from. The 580 has 3 billion, the 7970 has 4.31 billion. So the 7970 performed ~25% on immature drivers with 44% more transistor budget.

The 780 has 7.1 billion. So the 780 performed ~20% faster on mature drivers with 65% more transistors.

Anyway you look at it - apart from your "node shrink" way - shows that the 7970 was a better upgrade than the 580 was.

You absolutely got it wrong.Consumers don't care about transistor count but they do care about performance and in that regard 7970 was not a huge upgrade over 580.In the same way 680 was not a worthwhile upgrade for 580 users as well.

 

[blackened23]

The 7970 launch was problematic because the launch drivers sucked. I said this before but subsequent drivers fixed that issue 2 months later and it was inexcusable - the 7970 was definitely quite a bit faster than the 580 eventually, but the launch drivers sucked - this caused underwhelming launch benchmarks showing 8-10% above the GTX 580. Hell, eyefinity didn't even work correctly IIRC. Later drivers fixed that and improved the performance over the 580 by a ton, but AMD messed the 7970 launch up in many ways.

So yes, the 7970 wasn't represented well at launch (compared to the 580) because of the pathetic driver situation. That issue was rectified later but it should not have happened at all. The bottom line here is that I hope AMD learned lessons from that because the 7970 is how you do NOT launch a GPU. Software is just as important as hardware - if AMD hasn't learned that yet, I feel bad for them. That said, I think they realize their mistakes from that time frame - I do not think they will replicate that situation with the R9 series.

I also agree with Jaydip that nobody really cares about die size. I only care about noise and performance, I believe most feel this way.

 

[SiliconWars]

You absolutely got it wrong.Consumers don't care about transistor count but they do care about performance and in that regard 7970 was not a huge upgrade over 580.In the same way 680 was not a worthwhile upgrade for 580 users as well.

And yet the 780 was a huge upgrade over the Ghz Edition?

 

[GaiaHunter]

I also like how GaiaHunter decided to go through my past posts to find quotes which have nothing to do with the discussion at hand. :thumbsup:

It is called google - you type 96Firebird and overclock and whatnot.

Fact is hardware is evaluated not only by stock performance but also by the ability to overclock.

If the 7970 required 1300 to be 30% faster than the GTX580, that would be basically pointless most 7970 won't reach that.

But a 7970 only needed 1000 or even less to reach that 30% and most will reach 1100+.

 

[Jaydip]

And yet the 780 was a huge upgrade over the Ghz Edition?

Who told you that?

 

[SiliconWars]

Who told you that?

This is what all this discussion is about - http://forums.anandtech.com/showpost.php?p=35561386&postcount=43