Mac Pro 6 core slower than i7 980x?

[aquablue]

Why is the W3680 Xeon processor in the new Mac Pro 6 core 3.33 so much slower on this benchmark (http://www.cpubenchmark.net/high_end_cpus.html) than the consumer i7 980x? In the real world, is this gap as pronounced as this? Is it because Xeon's are just slower overall due to their need to correct errors?

 

[TheBritishBloke]

Xeons are server parts, not consumer parts.. And many CPU benchmarks only tesr so many cores.

 

[aquablue]

but the 980x has the same core number, no? Why is it so much faster than the Xeon?

 

[skiffx]

it also shows that Intel Xeon X5680 @ 3.33GHz is slower than Intel Xeon X5670 @ 2.93GHz which doesnt make sense.

 

[sboerup]

it also shows that Intel Xeon X5680 @ 3.33GHz is slower than Intel Xeon X5670 @ 2.93GHz which doesnt make sense.

agreed, they have some splainin' to do.

 

[Umbongo]

Don't rely on those numbers. They are on different hardware, there is no indication of what else was running during tests and the number of samples for Xeons is tiny compared to those of consumer parts.

 

[thagomizer]

Although 980x and W3680 have same CPU performance, ECC RAM used by W3680 is 2 wait states slower than regular RAM used by the 980x, all other things being equal. And if they used registered ECC RAM it might be slower still.

 

[skiffx]

Although 980x and W3680 have same CPU performance, ECC RAM used by W3680 is 2 wait states slower than regular RAM used by the 980x, all other things being equal. And if they used registered ECC RAM it might be slower still.

So why does apple bother using server tech which is slower instead of opting out for desktop chips/memory?

 

[Inconsequential]

So why does apple bother using server tech which is slower instead of opting out for desktop chips/memory?

Because stability is more important than out and out speed.

It's a workstation, NOT a overclockers dream.

Besides the performance difference of the CPUs are at max 1-2% from benchmarks i've seen, ECC vs non-ECC etc so that graph has ALOT of explaining to do.

 

[thagomizer]

agreed, they have some splainin' to do.

True. True.

 

[eponym]

LOL @ getting worked up over 1,500 Pts. in a synthetic test where one has hundreds of samples and the other a dozen. Not to mention a benchmark that makes not indication about hardware and overclocking.

 

[derbothaus]

There are other sources claiming other results. It depends on what you are doing which processor wins out. Here is that same site in the overclocked area with the w3680 on top. More headroom.

http://www.cpubenchmark.net/overclocked_cpus.html

This guy was able to get his w3680 to 5.5GHz. Stable.

http://forums.guru3d.com/showthread.php?t=314538&page=2

 

[Chilz0r]

ITT: e-pen comparison debate.