[Article] Probably IvyBridge-E would be a flop as well

Yesterday Tom's Hardware released some preliminary benchmark results of the upcoming IvyBridge-E CPUs. They only compared others with the highest end SKU, namely, the 4960X. The results were very lackluster once again. (Previously, both the SB to IB and IB to Haswell were lackluster) I felt the benefits of IB-E over SB-E were even less than the benefits of IB over SB.

 

Now, there was one important improvement with IB-E over SB-E. Power consumption. At full load, the power consumption delta was more than 80W. This is at the same clock speeds. With the power consumption being lower than the predecessor and the IHS being soldered to the die, we might see some sweet overclocking numbers though. Too bad Tom's Hardware didn't do  any overclocking tests. That could be the deciding factor.

 

But if we assume that IB-E would clock to the same values as the SB-E, this doesn't look like a big upgrade either. You will definitely see a noticeable improvement with multi-threaded applications by upgrading to a 6-core CPU, but at what cost? The sweet spot, the 4930K, will still cost like $600. That's about 70% more than a 4770K, while giving less than 50% better results even at best case. On average, the 4930K would be like 20% faster. Maybe even less. Like I said, it is the overclocking potential that is going to make a difference.

 

Another noteworthy thing though. Why the heck would anyone buy a 4820K over a 4770K? It will perform about the same (because there isn't a noticeable difference between dual channel and quad channel memory and most people will settle for a maximum of 2 graphics cards; the enthusiasts will go for 6-core CPUs anyway), but 4770K has newer motherboards with 6x native SATA-III and USB3.0 support and a lot of fancy stuff. We all know that there is not going to be a x89 chipset, so the IB-E platform will be 2 years behind current technology. Sucks! Intel should have released IB-E last year.

 

I was planning to get the IB-E and now I'm skeptical. I will wait for the overclocking results - retail overclocking results to be precise. (We all know how Haswell retail CPUs clocks worse than the engineering samples.

 

Why, why, why doesn't Intel want my money? :(