Clarkdale 10% faster than C2D clock-for-clock

[geoffry]

So says Nordic

http://www.nordichardware.com/news,9099.html

 

[VirtualLarry]

Cliffs: Clarkdale 10% faster in SuperPI. No other benchmarks listed.

 

[geoffry]

True.

I'm not a benchmark expert but why is Superpi usually the only thing quoted on these early leaks? From what I've seen it happens a lot.

 

[ilkhan]

its free and its fast. But its only single threaded.

 

[Gikaseixas]

If it apply to to many others (benchmarks) it's a good improvement.

 

[jones377]

Originally posted by: geoffry
True.

I'm not a benchmark expert but why is Superpi usually the only thing quoted on these early leaks? From what I've seen it happens a lot.

Because all these "leaks" are probably pre-approved by Intel intended to generate buzz on the Internet.

 

[Idontcare]

Originally posted by: jones377

Originally posted by: geoffry
True.

I'm not a benchmark expert but why is Superpi usually the only thing quoted on these early leaks? From what I've seen it happens a lot.

Because all these "leaks" are probably pre-approved by Intel intended to generate buzz on the Internet.

This is absolutely true.

Originally posted by: Gikaseixas
If it apply to to many others (benchmarks) it's a good improvement.

It never does, unfortunately we've seen this many times in the past. It's the same with vantage, and sandra memory bandwidth.

Originally posted by: ilkhan
its free and its fast. But its only single threaded.

It's also notoriously memory latency sensitive, exactly the kind of thing that a well-tuned IMC would benefit over a discreet NB memory controller being queried thru a FSB...there's a reason it is being compared to C2D and not PhII or i7.

Personally I will find it a tad disappointing if Clarkdale manages a measly 10% IPC bump in Superpi over a legacy C2D product as Superpi tends to be a "best case result" when it comes to bumps in single-threaded IPC.

 

[geoffry]

Originally posted by: Idontcare

Originally posted by: jones377

Originally posted by: geoffry
True.

I'm not a benchmark expert but why is Superpi usually the only thing quoted on these early leaks? From what I've seen it happens a lot.

Because all these "leaks" are probably pre-approved by Intel intended to generate buzz on the Internet.

This is absolutely true.

Originally posted by: Gikaseixas
If it apply to to many others (benchmarks) it's a good improvement.

It never does, unfortunately we've seen this many times in the past. It's the same with vantage, and sandra memory bandwidth.

Originally posted by: ilkhan
its free and its fast. But its only single threaded.

It's also notoriously memory latency sensitive, exactly the kind of thing that a well-tuned IMC would benefit over a discreet NB memory controller being queried thru a FSB...there's a reason it is being compared to C2D and not PhII or i7.

Personally I will find it a tad disappointing if Clarkdale manages a measly 10% IPC bump in Superpi over a legacy C2D product as Superpi tends to be a "best case result" when it comes to bumps in single-threaded IPC.

Thats why I enjoy bringing stuff here, someone is bound to know about the topic a tremendous amount and clarify things for me.

 

[aigomorla]

wprime is multi threaded.

 

[iCyborg]

They did the same trick with i7s too, the first "leaks" I saw were, conveniently for Intel, Super-Pi numbers.