nVidia: "We expected more from the 7970"

[Arkadrel]

Let's not go OT, but thankfully I do have water.

31C load @ 900 core, silly silly water coolers

2600k @4.4
7970 @1300

Yes Silly silly water coolers

1300mhz core,... temp says 31 degrees celcius after running 3Dmark.

Whats your 3Dmark 11 score like Bella?

 

[BallaTheFeared]

The reason its only "6%" is because you decided to use a graph that sets the 100% at a much faster card...

If you set the 580 as the 100% mark, the 7970 would have a larger percentage increase. That explains some of the difference.

Basically that graph is not showing what you want it to.

Actually I went by fps for each card... There is a 7% difference on the graph.

For 3Dmark I get around 12,800 for the gpu score, which is faster than most 6970s in CF assuming they didn't turn off tess. Which is one of the reasons I deplore benching against AMD users, people like to take advantage of advantages whenever they can.

 

[LOL_Wut_Axel]

NVIDIA, show your cards or shut up. As it stands right now, AMD has you beat on price/performance, performance/mm^2, performance/watt, and although mostly meaningless, they also have the fastest single-GPU card (HD 7970). With new drivers the GTX 590 is around 5% faster than the HD 6990, though.

Show Kepler and then talk.

 

[lavaheadache]

Yes Silly silly water coolers

1300mhz core,... temp says 31 degrees celcius after running 3Dmark.

Whats your 3Dmark 11 score like Bella?

those temps arent from immediately after. I ran the test while I was making food and When I came back a few minutes after thats where it was. I'm sure it's not too hot though as my card is fairly cool running

 

[CKTurbo128]

I wouldn't put too much faith into nVIDIA's statement about the HD 7970. After all, it's not the first time nVIDIA's made big remarks. Lest we forget, nVIDIA made a similar remarks about the FX 5800 series being superior to the Radeon 9700 Pro, a card that was out on the market for almost five months prior to the release of the FX 5800 Ultra. We all now what happened with that.

Until nVIDIA shows firm numbers (not synthetic benchmarks or useless marketing PR charts) on high-end Kepler, it's just only words.

 

[Vesku]

$300 GTX 580 on 28nm, release it immediately NVIDIA.

 

[Skurge]

Actually I went by fps for each card... There is a 7% difference on the graph.

For 3Dmark I get around 12,800 for the gpu score, which is faster than most 6970s in CF assuming they didn't turn off tess. Which is one of the reasons I deplore benching against AMD users, people like to take advantage of advantages whenever they can.

So anytime a 7970 beats your cards you claim their turn off tessellation? You do know that nobody actually uses the tess setting right? There are many games that don't use tessellation. Even if they do the 7970 has better tessellation performance than a 580 that's why your 470s get beaten by a single 7970 in crysis 2. So there is no reason why anyone would turn it off.

You cards at stock are already slower than a 7970 at stock. The 7970 can get a good 40% OC and it scales better with clocks. Not to mention it has much higher minimums and would slaughter them at 2560x1600.

But keep spreading misinformation. You've already been proven wrong in games YOU picked.

 

[OCGuy]

It is funny to see how defensive the fans can be over one person's statement. Stock 7970s were only slightly less disappointing than Bulldozer. Although in the big picture, there hasn't been a "WOW!" in single GPUs based off pure performance in quite some time. Maybe the 8800GTX? There have been several "WOW!"s in price/performance, however: (4870,gtxX60, etc).

I think with the mutli-gpu improvements allowing us to get a glimpse of next-gen performance 2 years before it actually releases, along with "consolitis" and the stagnation of true PC-torturing games....we will all most likely not be "WOW!"d for a while with a single GPU.

7970 didnt do it. I'm not so sure Kepler will do it either, but I would like to start seeing more credible leaks, and less once-respectable fanboi screaming. (looking at you, Kyle).

 

[chimaxi83]

So anytime a 7970 beats your cards you claim their turn off tessellation? You do know that nobody actually uses the tess setting right? There are many games that don't use tessellation. Even if they do the 7970 has better tessellation performance than a 580 that's why your 470s get beaten by a single 7970 in crysis 2. So there is no reason why anyone would turn it off.

You cards at stock are already slower than a 7970 at stock. The 7970 can get a good 40% OC and it scales better with clocks. Not to mention it has much higher minimums and would slaughter them at 2560x1600.

But keep spreading misinformation. You've already been proven wrong in games YOU picked.

Exactly. Oh, your top end new generation card beats my one step under top of the line SLI setup from 2010? U HAZ CHEATZ. Desperation, like I said.

Anyway, hopefully we see something from NV soon so we have something new to compare against, that or I'll just go for some Lightning 7970's :biggrin:

 

[lavaheadache]

It is funny to see how defensive the fans can be over one person's statement. Stock 7970s were only slightly less disappointing than Bulldozer. Although in the big picture, there hasn't been a "WOW!" in single GPUs based off pure performance in quite some time. Maybe the 8800GTX? There have been several "WOW!"s in price/performance, however: (4870,gtxX60, etc).

I think with the mutli-gpu improvements allowing us to get a glimpse of next-gen performance 2 years before it actually releases, along with "consolitis" and the stagnation of true PC-torturing games....we will all most likely not be "WOW!"d for a while with a single GPU.

7970 didnt do it. I'm not so sure Kepler will do it either, but I would like to start seeing more credible leaks, and less once-respectable fanboi screaming. (looking at you, Kyle).

7970 is only slightly less dissapointing than bulldozer? Welcome to my ever growing list of peoples opinions who I hold very little value to.

 

[Elfear]

It is funny to see how defensive the fans can be over one person's statement. Stock 7970s were only slightly less disappointing than Bulldozer.

Wait, what? Bulldozer loses in decent amount of benchmarks to the 3-yr old Phenom II. The 7970 is 40-50% faster than the 1-yr old 6970. Was their really only a slight difference in your disappointment levels?

 

[Arzachel]

7970 is only slightly less dissapointing than bulldozer? Welcome to my ever growing list of peoples opinions who I hold very little value to.

That would actually be a sensible statement, if his opinion on Bulldozer is "useful for some people, but not the best perf/$" instead of "horrid failure"

 

[Skurge]

It is funny to see how defensive the fans can be over one person's statement. Stock 7970s were only slightly less disappointing than Bulldozer. Although in the big picture, there hasn't been a "WOW!" in single GPUs based off pure performance in quite some time. Maybe the 8800GTX? There have been several "WOW!"s in price/performance, however: (4870,gtxX60, etc).

I think with the mutli-gpu improvements allowing us to get a glimpse of next-gen performance 2 years before it actually releases, along with "consolitis" and the stagnation of true PC-torturing games....we will all most likely not be "WOW!"d for a while with a single GPU.

7970 didnt do it. I'm not so sure Kepler will do it either, but I would like to start seeing more credible leaks, and less once-respectable fanboi screaming. (looking at you, Kyle).

Yeah cause bulldozer was faster than anything intel had at the time. Oh wait...
Bulldozer doesn't even pass as a disappointment, it was a massive failure.

Process tech is reaching diminishing returns and we're hitting power limits as well. So unless there is some breakthrough like unified shaders were, I don't think we'll see anything like 8800GTX type advances anytime soon.

 

[OCGuy]

Wait, what? Bulldozer loses in decent amount of benchmarks to the 3-yr old Phenom II. The 7970 is 40-50% faster than the 1-yr old 6970. Was their really only a slight difference in your disappointment levels?

I had much higher hopes for this oft-delayed node. 6XXX was barely an improvement over 5XXX, which was released on what date? I'll leave your arbitrary percentages out of the conversation.

If you read my whole post instead of just picking one thing that offends your self-worth due to your brand-association, you will see that there was plenty of reason to expect more from the 28nm cards so far. And I am not laying that blame soley on the card makers.

 

[BallaTheFeared]

So anytime a 7970 beats your cards you claim their turn off tessellation? You do know that nobody actually uses the tess setting right? There are many games that don't use tessellation. Even if they do the 7970 has better tessellation performance than a 580 that's why your 470s get beaten by a single 7970 in crysis 2. So there is no reason why anyone would turn it off.

You cards at stock are already slower than a 7970 at stock. The 7970 can get a good 40% OC and it scales better with clocks. Not to mention it has much higher minimums and would slaughter them at 2560x1600.

But keep spreading misinformation. You've already been proven wrong in games YOU picked.

Since you missed it the first time...

175 MHz increase resulted in, lol, get this... Wait let's first start with 900 to 1125, and figure out what kind of scaling we'll see with a 225MHz overclock... So for 225MHz, a 25% clock speed increase we see an increase of: Test 1: 16% Test 2: 16% Test 3: 16% Ohh imagine that. Now let's look at 1125 to 1300, a lesser increase of 175MHz or 15~%. Test 1: 18% increase over 1125 Wait a minute, a greater performance increase, for less of a clock speed increase, how does that work batman? Let's see what happens next! Test 2: 35% increase over 1125 lolwut?!?! Test 3: 48% increase over 1125

Now if you really want to continue to back someone who is claiming a 80% increase in performance over stock that's fine, just don't act like I'm the one spreading misinformation.

Who cares about 2560x1600, those screens cost more than your entire setup.

 

[blastingcap]

I had much higher hopes for this oft-delayed node. 6XXX was barely an improvement over 5XXX, which was released on what date? I'll leave your arbitrary percentages out of the conversation.

If you read my whole post instead of just picking one thing that offends your self-worth due to your brand-association, you will see that there was plenty of reason to expect more from the 28nm cards so far. And I am not laying that blame soley on the card makers.

Let's be fair here, 28nm may be having teething problems, and AMD also had to sacrifice more die space for HPC-related stuff (double precision float, etc.).

If HD 7970 had been a simple die shrink from HD 6970 I suspect that HD 7970 would be significantly faster.

 

[dakU7]

lol desperation, that's funny.

Hard tested what, 4 games at 1600p?

They tested like 15 DX11 titles, also they did 1600p, but I figured I'd post what's relative to me which isn't a $1000 screen.

6% faster on average at 1200p over 15 titles.

So going by that chart, a GTX580 is only 25% faster than a 5850? Lol. :whiste:

 

[96Firebird]

I expected more from the 7970, a lot of people did. Unfortunately, most are too proud to admit it.

So going by that chart, a GTX580 is only 23% faster than a 5850? Lol. :whiste:

How did you come up with that number?

 

[LOL_Wut_Axel]

I had much higher hopes for this oft-delayed node. 6XXX was barely an improvement over 5XXX, which was released on what date? I'll leave your arbitrary percentages out of the conversation.

If you read my whole post instead of just picking one thing that offends your self-worth due to your brand-association, you will see that there was plenty of reason to expect more from the 28nm cards so far. And I am not laying that blame soley on the card makers.

The GTX 580 was only 15% faster than the GTX 480, and the GTX 480 was only 15% faster than the Radeon HD 5870. Your point? The Radeon HD 7970 is only 20% faster than the GTX 580, but it's 40% faster than the HD 6970. That grows with overclocking, and by the time NVIDIA comes out with something AMD will have the HD 8970 ready to launch.

Fail to see how Tahiti and Zambezi/Bulldozer are in any way comparable. Bulldozer is slower than its predecessors in many areas and doesn't have better efficiency.

 

[LOL_Wut_Axel]

Since you missed it the first time...


Now if you really want to continue to back someone who is claiming a 80% increase in performance over stock that's fine, just don't act like I'm the one spreading misinformation.

Who cares about 2560x1600, those screens cost more than your entire setup.

Not a great argument to make when you can get a three display 5760x1080 setup for only $300-500.

 

[Arzachel]

Let's be fair here, 28nm may be having teething problems, and AMD also had to sacrifice more die space for HPC-related stuff (double precision float, etc.).

If HD 7970 had been a simple die shrink from HD 6970 I suspect that HD 7970 would be significantly faster.

Actually, a theoretical vliw4 7970 would be almost identical to the 7970 we got at stock.

 

[dakU7]

I expected more from the 7970, a lot of people did. Unfortunately, most are too proud to admit it.



How did you come up with that number?

Whoops.

ninja'd

 

[BallaTheFeared]

So going by that chart, a GTX580 is only 25% faster than a 5850? Lol. :whiste:

No, it's almost 40% faster according to that chart.

Not a great argument to make when you can get a three display 5760x1080 setup for only $300-500.

If I wanted to see bars I'd go to prison.

 

[BallaTheFeared]

dup

 

[LOL_Wut_Axel]

No, it's almost 40% faster according to that chart.



If I wanted to see bars I'd go to prison.

Wow, what a compelling argument!!!

We were talking about resolution and how it's "prohibitively expensive" to game at a resolution above 1920x1200, BTW.