Power Consumption
A number of readers have repeatedly requested that we add power-consumption data to the SBMs, and this month we are now able to provide total system power (of just the machine itself) measured from the wall outlet.
Low-level overclocking with Riva Tuners was quick and easy, but it dramatically raised our idle-power consumption as the high clock speeds were running both in 2D and 3D. When overclocking with CCC, the 2D clocks cycle down to a very low level, so the extra 40 MHz gain that Riva offered compared to CCC did provide some extra FPS in games, but it came at the expense of full-time lower energy efficiency.
Under load, we see the added power demands that our increased voltages and clock speeds have on system power consumption.
For SBM performance testing, power-saving features such as CPU Enhanced Halt and CPU EIST Function were both disabled. Also, Gigabyte’s Dynamic Energy Saver Advanced was not installed until all testing was complete, as we did not want any interference when reaching our maximum overclocked speeds. After installing and enabling these utilities, power consumption in idle mode dropped to 89 W while load consumption dropped to a more-impressive 180 W. However, no performance testing was conducted with these settings and we can’t assume there was no loss in performance when enabled.
Latest Buyers Guides News
Latest Buyers Guides reviews
- 19/01 – Five Overclocked GeForce GTX 560 Cards, Rounded-Up
- 23/12 – System Builder Marathon, Dec. 2011: System Value Compared
- 21/12 – System Builder Marathon, Dec. 2011: $600 Gaming PC
- 20/12 – System Builder Marathon, Dec. 2011: $1200 Enthusiast PC
- 19/12 – System Builder Marathon, Dec. 2011: $2400 Performance PC
First(s) I'd like to say thanks, for the review... Ah yes, once again OC'ing proves the nay-sayers the fools we've always known they are: "Well why don't you take that money spent on fans and get a better CPU?" LMFAO be quiet and stick with Dells and the other mass-produced JUNK...
I'm a little curious about the way you determine your ram selection (aside from price)? Doesn't really seem like a "top end," aka: Corsair, Geil, OCZ, Kingston, Mushkin etc?
What thermal paste are you using in this test also? That is a CRUCIAL determining factor in your OC'ing results. 5-10C can make or break a top-end OC.
What about replacing the stock TIM on the GPU-sink? Everyone knows that stock "paste" is pure garbage, along with the extremely inefficient TIM-type "cloth" they use on the ram-chips b/c of poor engineering tolerances aka, huge GAPS between the chip and sink. I took apart my old 7800GS+ and was shocked at the ~2mm gap that was "filled" (smashed) with this white-ish-paste-cloth type TIM gunk?! Id rather have thermal pads in place of that mess... anyhow using IC7 Diamond/AS5—I've seen min 15C-20C drops in GPU Load temps! That's certainly an eye-opening temp drop. And definitely going to yield a better OC, even for the say of stability and hardware longevity...Anyhow thanks 4 the review!
Overall a good build, though a bit CPU-heavy and PSU-light for this level. Surprised at the OCd GPUs stability given the low-rated PSU - is that one of the newer Delta-built 430W Antecs? The older Seasonic 430W PSUs are generally considered one of the least effective aftermarket Antec PSUs and more often get bundled with (now obsolete/EoL) desktop/high-end-HTPC (Fusion) cases.
I do see a pattern though... despite the hype, Intel's mid-binned 45nm parts (E7000/Q8000) seem to suffer serious stability issues at moderate and high OC. Given that Intel's safety spec says that 45nm CPUs will start to die beyond 1.3625v and that many enthusiasts will think twice about running such parts beyond the 1.4225v mark I do think that its pretty cheeky that Toms ran the benchmarks above with a CPU barely hanging in there at a whopping 1.4875v, which is also beyond the limit Toms was supposed to use (1.45v).
And it wasn't even that much use, other than proving that the extra L2 cache probably isn't worth the money to the more cash-strapped gamers. A lot of the improvement in gaming benchmarks can instead be put down to the much improved drivers added to the fact that Toms was much more adventurous with OCing this time around (its the same Sapphire Value HD4850 as the last SBM!).
v12v12 - Patriot are a good high-end RAM supplier... in the US. Less well known and more expensive over here. OCZ modules would be more par the course for Europe. And those Sapphire Value cards are cheap because of an outrageously non-reference architecture. That's why ATT won't work on them. They also leave out heatspreaders on anything except the voltage regulators - a good decision, as you said they're more of a hinderance on anything other than hot-running power circuitry (like the regs) so Sapphire leaves the bare RAM chips just under the fan. Not just for cost - several tests suggested that those modules ran cooler and clocked higher than identical ones on other competing HD4830/50s that used discrete heatspreaders or the integrated heatspeader on the reference design slot-cooler.
again another good build but i have bg problem with em, they are on the U.K sit, i would liek to see a U.K one done, i know this will probly not be taken note of but it is hard to ind the samecompants at the equiv price here
but still great as usual =D
As above, we could really do with these done on UK sites/shops. Scan for example or Overclockers UK.....