Gamers demand a lot from their computers, starting with the PSU. Therefore, almost every PSU manufacturer sells products optimized for gaming PCs. We introduce ripple and noise testing in this roundup to further improve our power supply evaluations.
In previous power supply reviews, we focused on a specific power output range and tested performance and efficiency.
This time, we asked PSU manufacturers to send us products developed for a very specific and very demanding group: gamers. Are the so-called gaming PSUs really optimized for this segment? Or is that designation just an empty promise created by marketing departments? We looked at 12 different products to find answers.
Following the requests of several manufacturers and our readers, we decided to include ripple and noise testing in our PSU reviews. According to the vendors asking about this test discipline, including these tests will highlight some obvious differences that should make it easy to draw conclusions regarding the electrical quality of a PSU, especially when it comes to high-performance units. As you'll see, this roundup will go on to show that not all manufacturers have done their homework in this area, and in some cases they don’t live up to our expectations--or the product specifications.
Ripple and noise testing is used to determine how accurately the circuits work to smooth out the DC voltage output. The goal is to produce a flat output, like you would get from a battery. Circuits of diodes and capacitors take care of this task as they convert alternating current (AC) to direct current (DC). Depending on the quality of these rectifier modules and components, the result shows more or less ripple and electrical noise. According to the ATX specifications, this value is not to exceed 120 mV for the 12 V rails. For all other ATX PSU voltages, the limit is 50 mV. A power supply that doesn't exceed these tolerances is good to go.
Huge Assortment, Huge Test
Up until now, we've generally compared five PSUs in our roundups. This is a natural limitation of test products, since we're paying for time in a professional testing facility. But it’s probably also because we're asking for products in very specific segments in order to make the comparisons as relevant as possible.
This time, we asked the manufacturers to send us gaming-oriented PSU products, without providing any specific criteria, and we received a veritable avalanche of submissions. Therefore, this review offers a broad spectrum of the market, represented by a total of 12 PSUs. Their power ratings fall between 580 and 850 W, and the efficiency certifications range from 80 PLUS to 80 PLUS Gold. Prices differ quite a bit, with the cheapest starting at around $90, while the most expensive offerings sit around twice that number. In addition to the manufacturers represented in our previous tests, Antec, Chieftec, Cooler Master, Corsair, Cougar, Enermax, and Seasonic, this time we have four newcomers to our test labs: be quiet!, NZXT, OCZ, and Sparkle.
- New Tests: Ripple And Noise
- Antec TP-750
- Antec TP-750: Results
- be quiet! Straight Power E8
- be quiet! Straight Power E8: Results
- Chieftec BPS-750C
- Chieftec BPS-750C: Results
- Cooler Master GX 750
- Cooler Master GX 750: Results
- Cooler Master Silent Pro Gold 700
- Cooler Master Silent Pro Gold 700: Results
- Corsair CMPSU-AX850
- Corsair CMPSU-AX850: Results
- Cougar SX700
- Cougar SX700: Results
- Enermax Modu 87+ 700 W
- Enermax Modu 87+ 700 W: Results
- NZXT HALE90-750M
- NZXT HALE90-750M: Results
- OCZ Fatal1ty 750 W
- OCZ Fatal1ty 750 W: Results
- Seasonic X-750
- Seasonic X-750: Results
- Sparkle SCC-750AF
- Sparkle SCC-750AF: Results
- Test Configuration, Hold-Up Time, Inrush Current, Peak, And Short Circuit Tests
- Efficiency According To 80 PLUS, Temperature, And Standby Overview
- Efficiency Under Different Load Profiles
- Conclusion And Recommendation

The article doesn't appear to measure noise from during switching and how much noise is introduced to the CPU and bus.
Anyway, it will be a cold day in hell before anyone gets me to switch out my Silverstone 1500 Watt PSU.
You are either lying or very unlucky (got a bad PSU)......I have crossfire 5870s and an i7 965 and all stock cannot exceed 650W at the wall (about 550W actually used) no matter what I try.
I call BS. I could run your setup with my PC power & cooling 750w unit.
http://www.techspot.com/review/289-geforce-gtx-480-sli-versus-radeon-5870-crossfire/page9.html
My next build will be in an NZXT full tower, and I am particularly interested in the Antec 850W, which I believe is the same line as the 750 tested. The report mentioned something about the cables being a bit short for full towers, I'd like to know exact cable lengths, and does anyone know if the 850W model cables may be a little longer?
You're kidding right? You mention the NZXT power supply as "being one of the best", but no mention or recommendation for the Seasonic (even tough according to your own tests the Seasonic handily beats the NZXT)?
Seasonic is one of the, if not thee, best power supply houses in the land. Most of the times if you see a real good power supply from another brand, it turns out it was a Seasonic unit. I'm not saying there are no other good manufacturers out there... But, to not even to get a nod or anything in the conclusion??
Review fail.
So, essentially it will pick away at your RAM. No thanks.
"...its single +12 V rail (preferable to all of the units with dinkier +12 V output spread over multiple rails)..."
I call BS. I'd much rather have multiple +12V rails. With OCP set appropriately, there's no issue of power being "trapped" anywhere, yet it will be safer in the event of near short-circuit conditions.
I'll be building either a SB or BD rig for myself in late Spring or early Summer, and I already picked up the 560W model of the Seasonic "X" line for it ($71 for the one HardwareSecrets reviewed).
Remember too, that 80+ tests at the absurdly low ambient temperature of 23C. While I think their tests are still generally valid, and will certainly clear out the liar-labeled units, for a certified unit to fail by a couple percent under real world conditions is not too unusual.