"7 Watt" Ivy Bridge my arse!

Page 11 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,787
136
That's because Intel's SpecCPU scores get a ~20% boost from being compiled with ICC, remember?

So you are basically saying that their results are completely discredited while other guys like ARM is showing proper results? In the case of AMD, where they were rumored to have been purposely crippling(and ICC was the only viable compiler) that makes sense, but when comparing two totally different architectures, comparing top to top makes sense. Also, just because they look similar in architectures at the high level, do not mean they are in actual implementation.

I do believe 2x+ advantage in single threaded IPC difference holds in real world as well.

It helps that the processors actually do properly throttle.

Not just that. Core 2 clock gates far better than Pentium 4, so you wouldn't have unnecessary units executing. TDP would be different for processors that can't turn off FPU when its only ALU being used versus one that can do it properly for example. Or you'd have units throttling when elsewhere is bottlenecked, with minimal to no performance degradation.

Plain videos or more obnoxious flash stuff? That was really just one example.

Modern processors with sophisticated power management don't go anywhere near TDP with flash, you should know that simply by looking at reviews. And like I said, while it might reach "TDP" for a second or so, from a cooling perspective it might be a lot lower.

If that's the case why even bother with a relatively expensive ULV Core series derived processor in them?

I don't think this is really relevant.
 

Exophase

Diamond Member
Apr 19, 2012
4,439
9
81
So you are basically saying that their results are completely discredited while other guys like ARM is showing proper results? In the case of AMD, where they were rumored to have been purposely crippling(and ICC was the only viable compiler) that makes sense, but when comparing two totally different architectures, comparing top to top makes sense. Also, just because they look similar in architectures at the high level, do not mean they are in actual implementation.

I do believe 2x+ advantage in single threaded IPC difference holds in real world as well.

No I'm not saying their results are discredited, I'm saying that you get better results using a better compiler and Intel has access to a better compiler for x86 than ARM does for ARM. That doesn't make the scores invalid, but it does explain why they would have a bigger advantage in something ICC compiled than they do in something that isn't ICC compiled like JITed Javascript code.

Whether or not this compiler benefit matters to the individual is going to depend on how much they use programs that are compiled with ICC.

In the real world, the differences depend so much on what you're doing that I'm reluctant to even try to give an average number for them. This is based on my experiences running my own software on both my Chromebook and i5-2500K desktop.

Not just that. Core 2 clock gates far better than Pentium 4, so you wouldn't have unnecessary units executing. TDP would be different for processors that can't turn off FPU when its only ALU being used versus one that can do it properly for example. Or you'd have units throttling when elsewhere is bottlenecked, with minimal to no performance degradation.

Yes, that all helps too.. power management has improved in a lot of different ways..

Modern processors with sophisticated power management don't go anywhere near TDP with flash, you should know that simply by looking at reviews. And like I said, while it might reach "TDP" for a second or so, from a cooling perspective it might be a lot lower.

This is another over-generalization. Flash can literally include any sort of user code and therefore of course a flash program can peg a CPU - and if we're talking a CPU with turbo enabled then the entire point of turbo is to let it clock as high as it can and therefore as close to TDP as it can.

Whether or not anything does actually do this is another question.

I don't think this is really relevant.

Not relevant to what? What were we talking about again? >_>