The engineering sample of the Haswell chip that the guys over at OCLab.ru had has a base clock speed of just 2.8 GHz, so the i7-3770K was down-clocked to the same level to make for a fair comparison. It is unknown which version the Haswell ES was supposed to represent, but one could assume it was representative of what we would presume to be the i7-4770(K). It could also be a yet unlabelled version though, possibly with or without HyperThreading. Obviously any tests performed may not be representative of the final product, especially considering the CPU was an engineering sample.
The tests from the Haswell engineering sample were run on an unnamed Z87 motherboard.
Further hardware for the test bench included Corsair Dominator 2666 MHz memory running at 1333 MHz, a Corsair AX1200 PSU, and a Corsair Neutron 240 GB SSD. Windows 7 x64 was used as operating system.
| CPU / Test | Super PI 32M (m:s) | Super PI 1M (s) | PiFast (s) | Wprime 32M (s) | Wprime 1024M (m:s) |
|---|---|---|---|---|---|
| Ivy Bridge i7-3770K | 11:49.094 | 13.1 | 25.5 | 13.97 | 7:11.8xx |
| Haswell Engineering Sample | 11:27.505 | 14.1 | 24.01 | 12.8x | 7:11.181 |
Anyone looking at these results is likely to be disappointed, especially if you look at the SuperPi 1M test, where the Ivy Bridge chip ran it a full second faster than the Haswell ES. Of course, these numbers mean nothing until we know more about which version of the engineering sample was used.
Don't lose hope just yet though, raw computing power isn't the only update that we'll see coming with Haswell. The new Haswell chips will carry a new iGPU that is supposed to perform 40 percent better than the current Intel Graphics HD4000 and more. 2.8 GHz might not even be close to the base clock speed that we'll be seeing upon release. Leaks indicate that the Haswell i7-4770K will have a base clock speed of 3.5 GHz and a boost clock of 3.9 GHz. The TDP will also be going up from 77 W to 84 W.
The Haswell chips are expected to launch in Q2 2013.
Update: Francois Piednoel (Principal Engineer / Performance Architect at Intel) has tweeted the following:
“if you see some #intel #Haswell numbers online, I doubt their veracity,since they are nowhere close to what I have in my lab, please ignore”here and “#intel #haswell is very healthy in our labs, we don't see any case where it is slower than IvyB. .ru numbers are funky!”here.