Bench run with HT disabled. i7 4770k @4.5ghz. Gtx 670 sli at stock. 64 player cq server. Ultra, x4 msaa at 1920x1200.
Frames, Time (ms), Min, Max, Avg
10988, 180000, 6, 105, 61.044
Only a short run sure, but the fps drop is quite considerable compared to with HT enabled, probably lost about 20fps. Game was virtually unplayable, very jerky. Cpu use maxed at 99%, (65% normally), gpu's at 70%. Sli was enabled in that run, which is known to be hampered by an i5 in this. Ill try a run sometime with just one card if i get time.
Setter there is no way an i7 is 35% more powerful than an i5 in games (99% cpu usage verse 65% usage)....It may be a heat issue affecting performance or sli not working actually effecting your gameplay without HT on who knows.
PARADOX................Heres why
Battlefield 4 Beta..........
Resolution at all times 1920x1200 4xmsaa, Win 7
Single Card 7950
3570k clocked at 4.6ghz in ultra uses 75% cpu utilisation max, single gpu 1200Mhz 6000Mhz memory vsyn on 98%, Mins 45 fps, Max 60, Avg 60.
Crossfire 2x7950........Heres why
3570k 4.6ghz, in ultra 85% cpu, gpu's 1150mhz 6000Mhz mem, Vsyn on, min 60, max 60 avg 60.
So when you use vsyn with 2x7950's you use 85% cpu at 4.6Ghz with a 3570k. GPU1 uses 100%, GPU2 is 40%. Vsyn caps both gpu and cpu usage as is expected!!!
Therefore 3rd or 4th gen i5 is not a bottleneck using vsyn at 60fps which is what most people run.
I appreciate that with vsyn off and 120Hz you might show a bigger difference with an i7 but I only have a 60hz 28" monitor that runs 1920x1200 so cant test at higher frequencies.
So my opinion is that the 3rd and 4th gen i5 or i7 is perfect for bF4 just as recommended by DICE.
Do 670 have a very narrow memory bus speed and bandwidth? (Are they 256 bit cards)