The Dual Rail v. Single Rail Challenge

[jdogg707]

The Contenders

Enermax Noisetaker Series ATX12V Ver2.01 600W PSU: $169.99

PC Power & Cooling Turbo-Cool 510 SLI: $239.00


Test System

AMD Athlon 64 3200+ (90nm) @ 2.608GHZ (1.568V)
Asus A8N-SLI Deluxe Motherboard (BIOS Rev. 1003)
PDP Patriot 1GB (512MB x 2) PC3200 w/XBL Technology (TCCD) (2-3-3-10) (2.75V)
2 x BFG 6800GT OC PCIe Video Cards (Core 1: 43C/58C) (Core 2: 41C/57C) (370/1000)
Western Digital 74GB Raptor SATA Hard Drive
Hitachi 250GB SATA 7200RPM Hard Drive
Sound Blaster Audigy 2 ZS Sound Card
NEC 3500A-BK DVD+/-RW
Cooler Master Centurion 5 Mid Tower Case (80mm Intake, 120mm Exhaust)
Thermalright XP-120 w/Thermaltake Thunderblade 120mm Blue LED Fan
Logitech Z-5500 5.1 Speakers
Logitech MX1000 Laser Mouse (USB)
Microsoft Multimedia Keyboard (USB)
HP PSC 2210 AIO Printer (USB)
Logitech Communicate Pro Webcam (USB)


The Premise

Since the emergence of SLI enabled motherboards, there has been a lot of talk about how to power the system. The main focus of the argument has been between Dual-Rail PSU's and Single Rail PSU's, and which provides better system stability/Rails. I decided to take one of the most powerful Dual Rail PSU's and compare it to the gold standard in single rail PSU's, as I, like many of you, want my system to benefit from the best power supply.


The Method

Tests were performed over a three day period. Readings were taken with a digital multimeter. I will list these findings by each benchmark I tested with. For all tests all fans are set to full speed, including the one on the Enermax PSU.

Idle Measurements

Enermax: 12V - 11.84, 5V - 5.05, 3.3V - 3.232

PPC&C: 12V - 11.84, 5V - 4.999, 3.3V - 3.264


The Benchmarks

3D Mark 2003 Pro - 5 Consecutive Runs

Enermax: 12V - 11.776, 5V - 5.05, 3.3V - 3.232

PPC&C: 12V - 11.776, 5V - 4.999, 3.3V - 3.28


3D Mark 2005 Pro - 5 Consecutive Runs

Enermax: 12V - 11.776, 5V - 4.96, 3.3V - 3.216

PPC&C: 12V - 11.776, 5V - 4.972, 3.3V - 3.248


Prime 95 - 12 Hours, Maximum Heat and Power Setting

Enermax: 12V - 11.67, 5V - 5.05V, 3.3V - 3.232

PPC&C: 12V - 11.776, 5V - 4.96, 3.3V - 3.248


PC Mark 2004 Pro - 5 Consecutive Runs

Enermax: 12V - 11.78, 5V - 5.05V, 3.3V - 3.216

PPC&C: 12V - 11.776, 5V - 4.999, 3.3V - 3.248

Half-Life 2 - 3 hours

Enermax: 12V - 11.78, 5V - 5.05V, 3.3V - 3.232

PPC&C: 12V - 11.776, 5V - 4.96, 3.3V - 3.248


Conclusion

I was a little confused by the results of these tests, because even though there weren't a ton of tests performed, I felt like I had found the results I had set out to find. I really went into these tests expecting the PPC&C unit to completely dominate the Enermax unit, with stronger, more stable rails. Surprisingly though, the Enermax held it's own and proved that dual rail PSU's can handle the load of a SLI system. I did video encoding (MPEG2, Divx, and Xvid), light Photo Editing, and daily office tasks on each PSU. So which one am I keeping? I don't know yet. I like the PPC&C because of the sleeving job done and the excellent 5 year warranty. The only reason I would send it back is because of Enermax's on-par performance, and equally excellent three year warranty (Not to mention the fact that it is almost $70 cheaper). Both units are of equal build quality, and I must say that if weight is the measure of a good PSU, both of these are heavy SOB's! Therefore it is my conclusion that if you have a quality dual-rail PSU, it will work just fine with a SLI system (at least a comparable system to mine), and that it isn't necessarily better to go with a single rail PSU.

I definitely recommend either of these excellent PSU's for heavy power users!

 

[mofrack]

Excellent post!

I've used Enermax PSUs for years, and they are great units. The PCP&C looked to have rails that were a bit more stable.

I personally prefer single-rail PSUs, but this is good news to anyone doubting their high-end dual rail units.

 

[Whitewolf]

Great post m8!
I am building a system with an Enermax 495 (dual rail 485W) and although I am going to use an X800XL (no SLI for me) I was scared by all of the people here saying bad things about dual rail psus.

 

[FastEddie]

I'd like to see this completed with a pair of 6800 Ultra's. And just one test--Prime95 for a straight 24 hours.

 

[TheInvincibleMustard]

Dual-Rail PSUs are perfectly capable of supporting systems such as this one, but the important thing to remember is that it is a dual-rail PSU ... if you max out the 18A on one of the rails, the other rail doesn't just automagically pick up the slack. Getting things wired correctly, etc. is certainly possible, but I would imagine it would also depend upon the individual system and the will of the end-user to wire everything correctly.

Nice work, though!

 

[GRAFiZ]

Nice work bro, the Enermax looks solid, however, most will argue the PCPower&Cooling will increase overall stability. Who knows for sure?

Hey, question... not exactly directly related, but how do you keep your BFG 6800GT's so cool?

I have the same set and they run around 65c to 75c (load).

Thanks!

 

[jdogg707]

Originally posted by: FastEddie
I'd like to see this completed with a pair of 6800 Ultra's. And just one test--Prime95 for a straight 24 hours.

I did Prime for 12 hours, but this weekend I'll see what happens with 24 hours. As far as the Ultras go, I have no doubt they will make an impact, but it would be interesting to see just how big of one.

 

[jdogg707]

Originally posted by: TheInvincibleMustard
Dual-Rail PSUs are perfectly capable of supporting systems such as this one, but the important thing to remember is that it is a dual-rail PSU ... if you max out the 18A on one of the rails, the other rail doesn't just automagically pick up the slack. Getting things wired correctly, etc. is certainly possible, but I would imagine it would also depend upon the individual system and the will of the end-user to wire everything correctly.

Nice work, though!

That's the thing, it is possible though to run a setup like this with those limitations. I don't think there is really a way to see how close I am to maxing the rails out, but if there is a way, I would like to test it. I really would like to put this issue to rest, becuase I think it is confusing a lot of people. It's hard when Nvidia is handing out certifications to both, and then people come here and see that it might just not be that easy.

 

[jdogg707]

Originally posted by: GRAFiZ
Nice work bro, the Enermax looks solid, however, most will argue the PCPower&Cooling will increase overall stability. Who knows for sure?

Hey, question... not exactly directly related, but how do you keep your BFG 6800GT's so cool?

I have the same set and they run around 65c to 75c (load).

Thanks!

To keep the BFG's so cool, I have a high power exhaust fan that runs around 2000RPM with max airflow of 90CFM. My intake fan also is running at full speed with max airflow of about 34CFM. Other than that, I have no idea!

 

[FastEddie]

This is very good work in putting this together. I note that you are very close to the max tolerance with the 6800GT OC's, as is indicative of your 12v reading in Prime95. 6800 Ultras would shred that rail. The other consideration to note is that two peices of hardware (psu's in particular) are never 100% the same, and usually perform +/- 5% to the manufactured line.

I know you have been following this argument closely, hence the three days of your time spent providing these results. You know then that I have stated that there were a few successes with dual rail psu's, but successes in powering the system, not in running at or near a constant PEAK. Your Prime95 test shows this with the 6800GT OC's in the drop of available current on the main 12v rail. Add considerations such as a warmer ambient temperature and you would see further deterioration of that 12v rail. Add further the power drain from the 6800 Ultra's, and you are hard on that PEAK wall.

I still recommend, with the state of the current dual rail psu's, to use a strong single rail psu for SLI.

 

[DAPUNISHER]

Always nice to see members that actually deserve the 1337 title :beer: Dual rail PSUs also give some dual Opteron setups nothin' but grief so I'll be sticking with single rail PSUs myself for the forseeable future.

 

[jdogg707]

Originally posted by: FastEddie
This is very good work in putting this together. I note that you are very close to the max tolerance with the 6800GT OC's, as is indicative of your 12v reading in Prime95. 6800 Ultras would shred that rail. The other consideration to note is that two peices of hardware (psu's in particular) are never 100% the same, and usually perform +/- 5% to the manufactured line.

I know you have been following this argument closely, hence the three days of your time spent providing these results. You know then that I have stated that there were a few successes with dual rail psu's, but successes in powering the system, not in running at or near a constant PEAK. Your Prime95 test shows this with the 6800GT OC's in the drop of available current on the main 12v rail. Add considerations such as a warmer ambient temperature and you would see further deterioration of that 12v rail. Add further the power drain from the 6800 Ultra's, and you are hard on that PEAK wall.

I still recommend, with the state of the current dual rail psu's, to use a strong single rail psu for SLI.

I completely agree and have been following what you, and many other forum members have been saying for some time. This weekend I'll start a loop of Prime95 and run it for 36 hours, testing rails at different times, with both systems. I'll repeat the test with the Enermax. I really want to see if I can get one of these PSU's below the accepted variance threshold. Any idea what programs I could run, other than maybe Prime that would do that?

 

[DAPUNISHER]

Looping 3Dm'05 in cmbo P95TT *play with the priority* should put some stress on the PSU. Unless you keep it warm in the comp room or crank up the thermistat though, you won't be able to see the worst temp senario till summer.

 

[darkeyed]

Nice job on putting this test together.

The other part to this issue is most people do not buy the top of the line dual rail systems. The other thing that people buying one have to be sure of is how it is set up and does the motherboard/video card pull to many amps from one of the rails. I am sure that dual rail PSU can work in many systems but you need to do your homework. I went with a single rail system for the machine I am building now mostly because insuring I dont run on the edge of one of the rails is not easy to figure out. I would rather not have a PSU be my limiting factor in my system.

 

[DAPUNISHER]

I think you make a very good point darkeyed, buying a PSU with a single strong +12 makes things more of a no brainer. I think there is a whole debate in if dual rail units will continue to be manufactured long term or not.

 

[jdogg707]

Originally posted by: DAPUNISHER
I think you make a very good point darkeyed, buying a PSU with a single strong +12 makes things more of a no brainer. I think there is a whole debate in if dual rail units will continue to be manufactured long term or not.

I honestly hope they don't. In practice, they offer no real benefit to users, and they make buying a PSU much more complicated than it should be.

 

[GRAFiZ]

Hey... try this program:

http://testmem.nm.ru/snm.zip

It's called S&M.

People over at xtreme use it to test stability.

I've played with it a few times and it is a POWER SUCK!

Trust me, I've never heard my power supply bog down as much as it did while running the CPU (FPU) test.

Most people see their CPU temp climb around 5c higher then it does during Prime95 tests.

Someone also did a test on the power consumption while it was running at got these results:

CPU Burn-in v1.017......................: 144W
Prime95 In-place large FFTs v23.7.1: 153W
OCCT v0.91:...............................: 153W
S&M FPU (v1.3):..........................: 160W

I've run it and was impressed, it's quick to find instability in an overclock, faster then any other test I've run. I get a feeling you'll see some very different results while running it.

Good luck bro.

 

[jdogg707]

Originally posted by: GRAFiZ
Hey... try this program:

http://testmem.nm.ru/snm.zip

It's called S&M.

People over at xtreme use it to test stability.

I've played with it a few times and it is a POWER SUCK!

Trust me, I've never heard my power supply bog down as much as it did while running the CPU (FPU) test.

Most people see their CPU temp climb around 5c higher then it does during Prime95 tests.

Someone also did a test on the power consumption while it was running at got these results:

CPU Burn-in v1.017......................: 144W
Prime95 In-place large FFTs v23.7.1: 153W
OCCT v0.91:...............................: 153W
S&M FPU (v1.3):..........................: 160W

I've run it and was impressed, it's quick to find instability in an overclock, faster then any other test I've run. I get a feeling you'll see some very different results while running it.

Good luck bro.

Awesome, I'll try that when I get home.

Thanks!

 

[GRAFiZ]

Originally posted by: jdogg707

Awesome, I'll try that when I get home.

Thanks!

No problem... I must add the server this guy uses is pretty poor. Sometimes the download works, sometimes it doesnt. For instance a half an hour ago I was able to download it, however, right now the server is down. If you can't connect, just try an hour later or so. It'll work eventually.

It's a fine little program, I was very impressed by it.

 

[GuitarDaddy]

Originally posted by: DAPUNISHER
Looping 3Dm'05 in cmbo P95TT *play with the priority* should put some stress on the PSU. Unless you keep it warm in the comp room or crank up the thermistat though, you won't be able to see the worst temp senario till summer.

I agree, adding 3Dm'05 would be a much more representative test of power draw. Prime95 is a great test for CPU stability, but it creates only CPU power consumption, which should be very small compared to the power draw of the 2 6800GT's.

 

[boatillo]

Now test the Fortron Blue Storm 500w and see how it does for $75

 

[GRAFiZ]

Originally posted by: boatillo
Now test the Fortron Blue Storm 500w and see how it does for $75

I'm sure he'd be happy to if you sent him the $75 bucks.

 

[jdogg707]

Originally posted by: GRAFiZ

Originally posted by: boatillo
Now test the Fortron Blue Storm 500w and see how it does for $75

I'm sure he'd be happy to if you sent him the $75 bucks.

haha, sure would! The meaning of this test was not to compare different brand PSU's, but to compare top of the line dual rail v. top of the line single rail PSU's. But if you want to send me $75.00, I'll buy one and test it!

 

[jdogg707]

Originally posted by: GuitarDaddy

Originally posted by: DAPUNISHER
Looping 3Dm'05 in cmbo P95TT *play with the priority* should put some stress on the PSU. Unless you keep it warm in the comp room or crank up the thermistat though, you won't be able to see the worst temp senario till summer.

I agree, adding 3Dm'05 would be a much more representative test of power draw. Prime95 is a great test for CPU stability, but it creates only CPU power consumption, which should be very small compared to the power draw of the 2 6800GT's.

Over the weekend I will do the following test:

P95 running at Highest Priority - 36 Hours

3DMark 2005 - 10 runs once every twelve hours

This should provide me with an accurate pircture for each PSU. I'll test outputs during 3DMark and at regualr intervals. I'll most likely just create an excel file with the data and anyone who wants it can download it from here once it is done.

I'll also try the program linked above and see what it does. The chipset fan died on my A8N today and I need to get thar replaced before I start doing any hardcore benchmarking.

 

[ImJacksAmygdala]

Thanks a TON jdogg707! Awsome thread! Anandtech please take notice and publish a review that compares different brands. It would be great if we could see a review that compared the OCZ PowerStream 600W, OCZ PowerStream 520W, OCZ ModStream 520W, ENERMAX Noisetaker 600W, ENERMAX Whisper II 2.0 SLI 535W, and PPC&C 510 SLI based on stability, features, warranties, and noise.

The test system should be something like:

Processor: Athlon 64 4000+
Cooling: Thermaltake Silent Boost K8 Heatsink/Fan
Memory: 1024mb OCZ PC3200 EL Platinum Rev. 2
Motherboard: DFI LANParty nF4 SLI-DR
Video: 2x BFG 6800ultra OC
Hardrives: 2X WD Raptor 74Gb
DVDRW
Sound Card

I'm stuck deciding between the OCZ PowerStream 600W, OCZ PowerStream 520W, and ENERMAX Noisetaker 600W. I've heard bad things about the noise issues and availability of the PPC&C 510 SLI and I think I can get a comparable PSU cheaper.

BTW- The chipset fan on the A8N-SLI is appaling! Makes me want the DFI-SLI even more!