AMD Kaveri OC On Planet Neptune

[Vesku]

Yes, if they could have at least tread water on clocks vs the A10-6800K it would go from "minor" improvement to "moderate" improvement. At least GF 28nm did get Kaveri a bit more GPU oomph, but that impact has also been softened by less than desired clocks.

PPB your 2nd image doesn't show up for me. But I remember from the R10 and R11.5 that most "real world" renderers do leverage the integer units better than Cinebench does. In terms of compiler and such there doesn't seem to be much if any impact on R11.5 and R15 scores for AMD CPUs. Whether that will hold true in the future when they start using AVX, AVX2 and such that should be for another future thread.

 

[Homeles]

Remember that when you come here with your proof, use respectable sources, cheers!

Yes, I agree. This image is so enlightening. Quality source.

Considering how often hardware.fr is linked by the AMD-biased of this forum, their credibility is rather suspect...

Do you even know how to measure IPC ??? same clock for both CPUs. How Kaveri at 4GHz is in the same clock as A10-5800K clocked to 4.2GHz ???

He was speaking about IPC, not products. Learn the difference.

Yes, I do. Keep failing to miss the point.

 

[AtenRa]

Yes, I agree. This image is so enlightening. Quality source.

Considering how often hardware.fr is linked by the AMD-biased of this forum, their credibility is rather suspect...

So if ShintaiDK that is an Intel-biased links from Anandtech you will not consider AT a credible source ???

Yes, I do. Keep failing to miss the point.

And you keep missing the point as well.

 

[Homeles]

So if ShintaiDK that is an Intel-biased links from Anandtech you will not consider AT a credible source ???

Sorry, your English is a bit poor here; come again?

And you keep missing the point as well.

No, I don't. As mentioned earlier, Steamroller has 8% higher IPC in this benchmark. I do believe that's less than Haswell.

However, as I've already mentioned, it's pointless to draw conclusions off of this leaked bit of information. That means that Shintai is making a mountain out of a molehill. But your reading comprehension skills weren't up to snuff to catch that, it seems.

 

[PPB]

Yes, I agree. This image is so enlightening. Quality source.

No problem, fixed already with a link that redirects to the CB15 thread on this very subforum.

Considering how often hardware.fr is linked by the AMD-biased of this forum, their credibility is rather suspect...

Yes, I do. Keep failing to miss the point.

Keep the falacies for yourself, and bring in the proof of other renders using ICC!

PS: Dont dismiss Hardware FR so easily, only because they have the bad habit of showing real world software when they compare CPUs. Silly them.

 

[AtenRa]

Sorry, your English is a bit poor here; come again?

you said,

Considering how often hardware.fr is linked by the AMD-biased of this forum, their credibility is rather suspect...

And i asked, if Intel-Biased forum members linked often from Anandtech, would that make AT less credible ???

However, as I've already mentioned, it's pointless to draw conclusions off of this leaked bit of information. That means that Shintai is making a mountain out of a molehill. But your reading comprehension skills weren't up to snuff to catch that, it seems.

But it was you who spoke about regression, wasn't not ??

 

[PPB]

No, I think it was ShintaiK.

I know, gets hard to distinguish the FUDers sometimes. Even harder when they have similar avatars!

 

[Homeles]

No problem, fixed already with a link that redirects to the CB15 thread on this very subforum.

Your link is still broken.

Keep the falacies for yourself, and bring in the proof of other renders using ICC!

PS: Dont dismiss Hardware FR so easily, only because they have the bad habit of showing real world software when they compare CPUs. Silly them.

Funny that you tell me to keep fallacies to myself, and then go ahead and use one.

It'd be nice if hypocrisy was against the ToS here.

And i asked, if Intel-Biased forum members linked often from Anandtech, would that make AT less credible ???

Okay, thanks for clarifying.

We're on the AnandTech site. It doesn't take a genius to figure out that your analogy isn't applicable.

But it was you who spoke about regression, wasn't not ??

A 7800 performs less than a 6800. That's the very definition of regression.

 

[AtenRa]

A 7800 performs less than a 6800. That's the very definition of regression.

In that sense then yes, 7850K has lower ST performance in R15 than 6800K. But it is the MT performance that we care in R15 and all those applications. We only care about ST performance in relation to IPC, single thread performance between products is meaningless in R15.

 

[Homeles]

In that sense then yes, 7850K has lower ST performance in R15 than 6800K. But it is the MT performance that we care in R15 and all those applications. We only care about ST performance in relation to IPC, single thread performance between products is meaningless in R15.

Ah, I didn't catch that it was a single thread score in the chart.

 

[AtenRa]

Ah, I didn't catch that it was a single thread score in the chart.

Well we were talking about IPC and Single Thread performance in R15, at least i was and i believe ShintaiDK too.

 

[inf64]

That spiraled out of control pretty fast. From Shintai claiming it's a "fail" based on one benchmark to others bickering about compiler bias etc.
Patience is a virtue that is really lacking in people these days

 

[Homeles]

Well we were talking about IPC and Single Thread performance in R15, at least i was and i believe ShintaiDK too.

Regardless, IPC is absolutely meaningless on its own, as well as clock speed. The two have to be paired together. In this case, Steamroller has a 8% higher IPC (quoting inf64), but a >8% regression in clock speed.

This has been Shintai's point all along. You keep ignoring the clock speed side of the equation.

Anyway, as I've said multiple times now, there's no point in trying to digest these scores.

That spiraled out of control pretty fast. From Shintai claiming it's a "fail" based on one benchmark to others bickering about compiler bias etc.
Patience is a virtue that is really lacking in people these days

It's just one person bickering, and he's still going, it seems. Even if he comes back and somehow "proves" that Cinebench isn't well-representative of real world rendering software... no one else gives a flying flock except for him. It's a dumb point to try and prove. He's delusional if he thinks that we and everyone else are going to stop using and looking at Cinebench scores just because he wants us to. It's well ingrained in the hardware community, whether he likes it or not.

 

[AtenRa]

Regardless, IPC is absolutely meaningless on its own, as well as clock speed. The two have to be paired together. In this case, Steamroller has a 8% higher IPC (quoting inf64), but a >8% regression in clock speed.

This has been Shintai's point all along. You keep ignoring the clock speed side of the equation.

Anyway, as I've said multiple times now, there's no point in trying to digest these scores.

It's just one person bickering, and he's still going, it seems.

Well actually i was the one that said we measure IPC clock to clock, anyway more benchmarks will soon be available to see about the Kaveri performance.

 

[ShintaiDK]

Well actually i was the one that said we measure IPC clock to clock, anyway more benchmarks will soon be available to see about the Kaveri performance.

Even if you want to keep bickering about IPC, its not 30%, its not 20% and its not 10%. So my statement was correct.

You simply went for the person instead of the ball right away. I assume due to the "bad news".

 

[ShintaiDK]

In that sense then yes, 7850K has lower ST performance in R15 than 6800K. But it is the MT performance that we care in R15 and all those applications. We only care about ST performance in relation to IPC, single thread performance between products is meaningless in R15.

MT performance is lower as well sofar.

6800K 326
7850K 311

 

[Vesku]

I predict AMD buries the CPU results under a small mountain of GPU aware results.

 

[PPB]

It's just one person bickering, and he's still going, it seems. Even if he comes back and somehow "proves" that Cinebench isn't well-representative of real world rendering software... no one else gives a flying flock except for him. It's a dumb point to try and prove. He's delusional if he thinks that we and everyone else are going to stop using and looking at Cinebench scores just because he wants us to. It's well ingrained in the hardware community, whether he likes it or not.

Just tell me you couldnt prove anything from the start and will still look at CB because it fits your agenda and it's all good. The rest of the people that actually USE renderers, will stick to benches from those real renderers :biggrin:

 

[Homeles]

Just tell me you couldnt prove anything from the start and will still look at CB because it fits your agenda and it's all good. The rest of the people that actually USE renderers, will stick to benches from those real renderers :biggrin:

It doesn't fit my agenda. I don't care about Cinebench. And in case you haven't noticed, I've been dismissing these results.

Sorry bud, you've come up short again.