Drazic,
running dual pumps will give u ~1-2oC (instead of running just one) off that system if you were using something like 120.2 rad, having bigger rad and also two of them you'll need something much better than 1048 pump's (even if they are two).
Depending the cpu block you want to use you have to set your pumps in different oriantation (parallel or series). Personally i use a Storm waterblock which needs high pressure to perform at its best so i have 2xL30's in series to have maximum pressure, sometimes i can hear the water flowing through the block jets...
What block you going to use btw.... you haven't mentioned it.
If i had another 120.2 rad on that loop that headpressure would be less but water temps would be better hence might have seen another ~2oC off the cpu but no more than that as if i wanted maximum waterblock efficiency i would need a scale bigger pump than 2xL30's, orelse waterblock efficiency would drop cause of not enough pumping power. Similar thing happened to smids 3 radiator setup and single not powerfull pump
2x120.3's are overkill for a prescott even at 4ghz, even though the more rads the better but not always. The more rads the more head pressure you'll loose and pump will struggle. Going with two of these massive rads, you'll need big pumps to have decent headpressure on the waterblock to perform (that is if the waterblock that you're using is needs high pressure to perform). I've gone through prescott 478socket & 775 and running a presler now using same radiator (120.2 that is and also have a 6800GT overclocked vga on the same loop) and watercooling setup. Temps never got more than 50oC and all cpu's were overclocked about 4.15Ghz with decent vcores in some cases.
My opinion on this is to use a single 120.3 and two better pumps in series or in parallel depending the waterblock you going to use, i.e. if it is a restrictive one you'll need high pressure, so go with series pump orientation. If the block is less restrictive go with pumps in parallel to improve flow.
Remember, you need a radiator to dissipate heat from the components you are using. Which means the cpu and vga heat energy, which are approx 220watts at full load both. If you check thermochill charts you'll see that a 120.2 radiator is good enough to dissipate that much heat using relevant spec fans and ambient temperature.
If you are stubborn and want to use no matter what two 120.3's make sure to use something like 2xL30's or something bigger, the rest setup on the pic you posted looks ok with each rad positioned as is, might need two T lines to bleed the system quicker as having two rads you going to have more coolant running through that system.
Also smids is right on this, HT cpu's are single cores, so running prime95 actually gives almost maximum heat output. If you notice you'll see that running single prime, cpu temp (on a single core HT cpu) will go lets say 50oC and when you start second prime instance it will move from 50oC to 52oC, whilist in a dual core cpu if one prime95 instance is running cpu temp will go up to 5oC and running a second instance it will move from 50oC to 56-58oC's (numbers are just random on watercooling, on air they should be even more).
For example on my 920D running at 4.15ghz, 1.44vcore on watercooling, cpu reaches 38oC after an hour running prime95. If i start the second instance temp will instantly rise from 39oC to 43oC and then stabilize at 44oC. On my single HT core, temp would go up to 46oC with one prime95 instance and 47oC with two
This means that with a single HT core while you run single prime95 you get about maximum heat output, but to get 100% max heat output you'll need two whilst in dual cores you have about 60% being used when running single prime instance and 100% when running two. Much of the temperature issue and observation here can be also explained that since they are under the same heatspreader, when one core is running the heatspreader spreads the heat over the surface anyway so you can't really tell the difference by cpu temps, so is kinda pointless to compare it by max temps but only by watching how fast the temp differences rise on the beggining of the test, only then you can tell the difference in between single and dual.
sorry for the long post just speaking by personal experience on watercooling and intel cpu's