What Makes a PSU SLI Compatible? Got Answers Here!

[ImJacksAmygdala]

I'm still looking into what makes a PSU "SLI compatible". There are a few PSUs that advertise this and most of them have dual 12 volt rails. The only difference I have found is that the "SLI compatible" power supplies have at least 18A on both 12 volt rails and have the PCI-E 6 pin connectors so you don't need an adapter. They also all support the ATX12V ver 2.0 standard.

There is also a power requirement chart at Asustek for the A8N-SLI that doesn't really help. Click on the Power link there.

http://usa.asus.com/products/m...a8nsli-d/overview.htm#

Newegg has two supplies that advertise "SLI compatible"

http://www.newegg.com/app/View...tion=17-103-461&depa=0

http://www.newegg.com/app/View...tion=17-103-460&depa=0

I wish there was more reviews on those power supplies. I think people at the Anandtech forums might have disliked the dual 12volt rail power supplies because most of them have uneven current ratings between the 12volt rails and they were below 18A in most cases which caused problems for some people.

I agree it is extremely confusing and frustrating. Your average dual 12volt rail or single 12volt rail PSU might work great with some GF6600's but not dual high end cards. If you look at these threads you can see that not all power supplies are created equal in SLI systems. Some people even went as far as recommending people avoid dual 12volt rail power supplies and just get a high current single 12volt rail PSU with at least 24A.

http://forums.anandtech.com/ca...px?catid=29&entercat=y

http://forums.anandtech.com/me...wer+AND+supply+AND+sli

http://forums.anandtech.com/me...wer+AND+supply+AND+sli

http://forums.anandtech.com/me...wer+AND+supply+AND+sli

http://forums.anandtech.com/me...wer+AND+supply+AND+sli

I'm still researching this aspect of SLI.

I was looking at Monarch computers site and they advertise both the Enermax Noisetaker EG701AX-VE-SFMA 600W $165 and the Enermax Noisetaker EG701AX-VE-SFMA 535W $99 as SLI compatible. Both Power supplies are ATX 12V V2.0 compatible, but even the expensive one only has 17A on the second 12volt rail. Maybe SLI just require the ATX 12V V2.0 standard or a high current single 12volt rail. I have not been able to find much user feedback to back this up though. The other thing I noticed is that Monarch adds the current from both rails. Maybe the ATX12V ver 2.0 standard works this way although many on the Anandtech forums said it did not. Ryan "ViRGE" Smith agrees with the folks at Anandtech forums when he says this.

To be completely spec compliant, the second PCIe 16x slot also needs its 75 watts of power, but considering the relative massiveness of that kind of power, there?s no easy way to do it. One way is to simply draw more power from the EPS12v connector, but if we back up for a second, there is that dual-rail design to talk about. Because power and power draw can?t be perfectly balanced among the rails on a whim, the total power a system can use will never be rail1 + rail2, which leads to a problem when a device needs more power than just what a single rail can provide. What we?re seeing then is that with some dual-rail power supplies and even with the high-end ones, is that SLI boards their video cards are favoring the EPS12v connector too much, and that dual-rail supplies can?t provide enough power to the motherboard as a result. This causes a very odd situation where single-rail power supplies with a high amperage rating on their 12v line works better on some of these SLI boards, which is causing a lot of customer confusion, and some resentment.

Any recommendations on a good power supply for a SLI system that would have a +3500, 2x512mb, dual 6800GT, sound card,and a DVD-RW? Should I only consider ATX 12V V2.0 Power Supplies? I was hoping to get a PSU that already has a 24pin power connector and at least two 6pin PCI-E connectors so I didn't use up four 4pin plugs using adapters.

Hope this is clear as mud for you too...

So then I stumble on to this little jewel of an article at HomeLAN Fed.

http://www.homelanfed.com/index.php?id=29023

This finally puts things into perspective... Thanks Ryan "ViRGE" Smith!

Power, Power Everywhere, But Not a Volt To Drink?

Throughout the growth of the personal computer in the 80?s, 90?s, and now in the 3rd millennium, it has been a design that not unlike humans has had to endure some growing pains every now and then. Though format wars such as those between DVD + and ? are often the most recorded of such growing pains, the PC has seen a much more constant pain in the form of power consumption and heat generation. With the greater computing power that computers have gained, the need for more electricity (and by extension more heat dumped out) rears its head out every few years, and those few years have once again rolled around.

Since the creation of the ATeXtended(ATX) specification back in the 90?s, it has required some minor revisions every now and then to keep up with power consumption, and such changes are pretty quiet. Besides the recent creation of the EPS12v standard(more on that later), the last major revision was the addition of the ATX 12v connector, the small 4pin square connector originally needed to provide extra juice to the then-new Pentium 4, and since then adopted by Intel and AMD alike. Since 2002 however, the ATX standard has run in to some issues related to power that it was never originally intended to cover in depth: video card power. We first saw this problem back in 2001 when 3dfx was showcasing its power hungry Voodoo 5 line, which with the high-end V5 6000, needed its own external ?Voodoo Volts? power supply. But with 3dfx?s death, the issue was buried once again until 2002, when ATI released the Radeon 9700.

With the 9700 out and ATI not going the way of 3dfx, it became a real problem that had to be dealt with, more so once the GeForceFX 5800 hit the streets also needing a power connector. The unfortunate side effect of this is that the arms race between ATI and Nvidia has been spilling over in to motherboard and power supply design, and that things only became worse once the AGP Pro slot failed to catch on and Nvidia bumped up their power requirements to 2 independent connectors on the GeForce 6800 cards. This of course lead to the popularity of ?dual-rail? power supplies that could power the motherboard/CPU and video card on separate rails, and what evolved is a sort of power anarchy among motherboards, video cards, and power supplies.

To the credit of the ATX group, their has been some effort on their part to reign in on this with the introduction of new standards for PCI-Express parts. While the ATX 12v connector was kept for locality reasons(it can go right next to the CPU, making life far easier for motherboard designers), the 20pin ATX plug was replaced with the 24pin EPS12v plug, which adds another 4 pins to help deliver more power to the PCIe slots(which can deliver more power than their AGP counterparts, up to 75W now) without taking power away from the CPU and other components. Video cards received some special attention here too, with a 6pin PCIe connector created that allows for power supply designers to more appropriately route power instead of having to worry about overbuilding the typically low-power 4pin molex connector to handle the demands of a video card.

However, in making this new standard the ATX group didn?t manage to lay down any plans for SLI or certain adaptors, which is where our anarchy problems today start. The first issue that some people are finding is that with SLI boards comes some new and interesting power distribution requirements. To be completely spec compliant, the second PCIe 16x slot also needs its 75 watts of power, but considering the relative massiveness of that kind of power, there?s no easy way to do it. One way is to simply draw more power from the EPS12v connector, but if we back up for a second, there is that dual-rail design to talk about. Because power and power draw can?t be perfectly balanced among the rails on a whim, the total power a system can use will never be rail1 + rail2, which leads to a problem when a device needs more power than just what a single rail can provide. What we?re seeing then is that with some dual-rail power supplies and even with the high-end ones, is that SLI boards their video cards are favoring the EPS12v connector too much, and that dual-rail supplies can?t provide enough power to the motherboard as a result. This causes a very odd situation where single-rail power supplies with a high amperage rating on their 12v line works better on some of these SLI boards, which is causing a lot of customer confusion, and some resentment. Thankfully for vendors, what?s a new power supply when you?re already shelling out upwards of $1000 just for a board and a pair of video cards, but it?s still a thorny issue.

To the credit of the boys over at Asus, they are taking a slightly different approach to the situation as you can see here, as their SLI board includes a molex plug on the board to help power the second PCIe slot. Unfortunately this hasn?t completely alleviated the problem among dual-rail power supplies, and it?s also encouraging another problem that?s best described as ?connector hell.? Compared to just 5 years ago, we now have motherboards that don?t just need an ATX/EPS power plug, but need the second 12v plug, the molex plug, and their video cards(which for the sake of argument we?ll include as part of the motherboard as the larger whole) need the PCIe power connector. For user of a single-rail power supply, this requires just coming up with enough plugs on a strong enough power supply for everything(which thankfully isn?t as hard as it used to be), and it generates a precarious balancing act for dual-rail users. In short, it?s getting to be too hard to just plug in all the power connectors for a computer.

Even the vendors themselves are getting confused and are running in to issues related to all of this. Because not all power supplies come with a PCIe power connector, the design allows for an adaptor that will take molex plugs and let them be adapted to a PCIe connector to allow these older supplies drive new PCIe cards, but the PCIe connector design was built with Nvidia?s power-hungry 6800 Ultra in mind, so it was designed to carry 2 molex plug?s worth of power.

But remember at the same time that the new PCIe standard allows for more power on the PCIe card slot than AGP did, so there?s not as much of a need for that second molex plug?s worth of power; so to save end users some grief, when ATI built the adaptors for their PCIe X800 cards, they designed such an adaptor that takes a single molex, and built their cards in turn to only draw on one rail. Their plan backfired a bit though, as their adaptor wasn?t designed correctly and is missing a ground connector. On the plus side, it?s not the important ground connector that makes the difference between frying the card or not, but on the down side, it is the ground that the card uses to complete a circuit to make sure that the PCIe connector is plugged in. Because of this the card will fail to work thinking that it doesn?t have a PCIe power connection, but the fix isn?t much harder than replacing the cable with a proper one for $10 or so. But for ATI this is a lesson in it not always being wise to try to cut corners no matter the intentions, and yet another exhibit of the anarchy and connector hell problems that are continuing to evolve.

We wouldn?t be talking about this issue if we didn?t have a solution though, so a few people are probably going to cringe right about now: we need another new power standard. Call it ATX+, ATXX(X), ATY, I don?t really care, but high-end home/gaming hardware has worked itself to the point where just like high-end workstation hardware it needs its own standard. At the very least, the EPS12v connector needs to be built to provide more power, either by increasing tolerance or adding more pins, and the ATX 12v connector needs to be built in to the EPS12v connector too. And while we?re at it, increase the amount of power a PCIe 16x slot can provide so that it?s well overbuilt if it has to be, as long as the external power connector is no longer necessary(though I?ll be benevolent and say it can stay for compatibility with mid-range boards that don?t have the extra juice).

Now I?ll be the first person to admit that such a suggestion is about as ugly as they come(my sincerest condolences to the motherboard designs out there), but at this point, do we have much of a choice? High-end computing isn?t known for its falling power needs, and with the movement to dual-core CPUs later this year we?re on the verge of another jump in power usage, so now is as good as any time to try to make a more permanent fix. Otherwise, if we don?t put an end to this power anarchy and reign in on connector hell, how many more people are going to be left having to keep buying unnecessary new parts for their computers just to satisfy their computer?s power needs? The answer is greater than zero, and that?s too many people in the year 2005.

Yep clear as mud! Atleast this explains what folks at the Anantech forums found about high current single 12volt rail power supplies vrs dual 12volt rail power supplies.

 

[Algere]

I was looking at Monarch computers site and they advertise both the Enermax Noisetaker EG701AX-VE-SFMA 600W $165 and the Enermax Noisetaker EG701AX-VE-SFMA 535W $99 as SLI compatible

Typo

Both Power supplies are ATX 12V V2.0 compatible, but even the expensive one only has 17A on the second 12volt rail.

The second 12V rail can have 18A if the first 12V rail is only using 17A (35A total per Enermax specs).

the second PCIe 16x slot also needs its 75 watts of power

Something I don't get here. Since SLI (on desktop) is two 8X when split, does power consumption follow as well? From my understanding 16X is 3A*3.3V + 5.5A*12V (75W) while 4X/8X is 3A*3.3V + 2.1A*12V (35W). So would SLI follow the power consumptions of 2*16X PCIe or 2*8X PCIe?

the total power a system can use will never be rail1 + rail2

Well... rail 1+2 does power an entire system - drives, MB, & all, but if he's (Virge) just talking about the 75W 16X PCIe slot power spec then yes 75W alone comes from one 12V rail. OTOH PCIe gfx cards that utilize a 6-pin PCIe power plug can* feed off of both 12V rails unless 12V consumption from the motherboard is disabled which I haven't heard being done as of yet.

*Depends on the PSU's 12V configuration.

 

[Elfear]

This whole PSU/SLI issue has my head spinning. From what I can gather, the dual-rail PSUs aren't necessarily the problem but to be safe it's better to get a strong single-rail unit if you're going to run SLI with two high-end cards.

Not trying to hijack your thread or anything but my question goes right along with your subject. I've narrowed down a suitable PSU to a couple of models (I'm sure there are others out there but these seem to get talked about a lot or are inexpensive). Which would you guys go for out of these:

OCZ 520W Powerstream $140 http://www.atacom.com/program/...number=&KEYWORDS=O5_20

Sparkle 550W $88
http://www.ewiz.com/detail.php?p=PS-SP550I&c=pw

Zippy 500W $125
http://www.newegg.com/app/View...tion=17-103-708&depa=0

Enermax 550W $129
http://www.newegg.com/app/view...tion=17-103-437&depa=0

The OCZ is obviously the nicest but also the most expensive. I need it to power 2x6800GT, oced winchester, and normal peripherals.

 

[Algere]

Zippy per rep.

 

[FastEddie]

Great research ImJacksAmygdala. :thumbsup:

This is an excellent explanation of power draw in dual rail -vs- single rail power supplies. The point made is that in dual rail psu's, the power draw is never rail 1 plus rail 2 because these units aren't designed to accumulate power like that. What happens in reality is one of the two rails gets maxed, to perform at a constant 100% of it's output.

There are a few psu's that seems to combine the ATX1.3 standard and the ATX2.0 12vEPS standard in using a hybrid connector that has a 20-pin+4-pin=24-pin connector, in addition to the 12v AUX 4-pin connector used to supply power to the cpu socket. These psu's also have the -5v pin which is absent from the ATX 2.0 12vEPS standard (a seperate issue in relation to the new MSI NEO 4 Platinum SLI's Creative onboard sound chip requiring a -5v pin to enable the onboard sound). They fall outside of the ATX2.0 standard, but that may be why they are working so well.

 

[Insomniak]

Awesome post. This is why I'm thinking of buying a CoolerMaster Stacker any day now - two PSU mounts.

 

[ImJacksAmygdala]

Thanks guys... I thought this was a topic that had a lot of people searching for information.

Althought that article helps puts things into perspective it still doesn't really answer the question of what a consumer would want to look for in a dual 12volt rail PSU for SLI that has all the nice new connectors.

I think I'm going to have to agree that for 2X 6800GT, an OCed +3500, plus the rest it would be better to get a high current single rail PSU like the one from OCZ. It recieved the Gold award in The Tech Report 8 PSU round up, but at the time of that article SLI wasn't out yet and it requires adaptors for PCI-E or a total of 4 4pin molex plugs just for video cards which is not ideal...

I think its high time a review site like Anandtech addresses this issue in a PSU review.

 

[McArra]

TT purepower 680W is also supposed to be for SLI setups and has 3 12V rails, 2 of them with 15A and another one with 8A, for a total of 38A. This is the PSU I will receive next week, is supposed to be really really powerful. I'm not using SLI though, I'm an AGP owner.

 

[Insomniak]

Originally posted by: McArra
TT purepower 680W is also supposed to be for SLI setups and has 3 12V rails, 2 of them with 15A and another one with 8A, for a total of 38A. This is the PSU I will receive next week, is supposed to be really really powerful. I'm not using SLI though, I'm an AGP owner.

Amps still aren't additive - if any of your devices need more than 15A, like an A8N-SLI Deluxe, then you're boned.

 

[ImJacksAmygdala]

This is the perfect example of why I made this topic...

 

[Insomniak]

Originally posted by: ImJacksAmygdala
This is the perfect example of why I made this topic...

I feel your pain.

 

[tsmori]

I went with the PC Power and Cooling Turbo-Cool 510 SLI. I've used them for years. I had bought one of those Ultra PSUs, but the PCP&C unit feels more substantial. The modular cables were nice, but not very flexible and also not long enough for a full-tower case.

I'm not sure if the PCP&C unit has dual 12v rails, but it does have the 6-pin connectors for PCI-e cards. It is however, not an economical piece of equipment. I know a lot of people balk at paying $100 for a PSU, but this baby ran over $200.

 

[Skinjob]

I've been looking at this one:

Link Depot 14CM LED Fan 550W Power Supply

Native 24-pin
Modular, wrapped cables including SATA and video
140mm fan
+12V @ 30A

Looks to be a killer according to the specs/features/price, but seems to be a new comer to the PSU market. Anyone had experience with or know of any reviews of Link Depot PSUs?

 

[Insomniak]

Originally posted by: Skinjob
I've been looking at this one:

Link Depot 14CM LED Fan 550W Power Supply

Native 24-pin
Modular, wrapped cables including SATA and video
140mm fan
+12V @ 30A

Looks to be a killer according to the specs/features/price, but seems to be a new comer to the PSU market. Anyone had experience with or know of any reviews of Link Depot PSUs?

I heard these are rebadged SuperFlower/TTGI PSUs. If that's the case, they're a rather good deal for the money.

 

[Skinjob]

Originally posted by: Insomniak

I heard these are rebadged SuperFlower/TTGI PSUs. If that's the case, they're a rather good deal for the money.

If you go to http://link-depot.com/powersupply.html they actually list Super Flower like it's a Link Depot product line. Not sure who actually makes them, but there's certainly a strong family resemblence.

So you think it's worth a shot for $95? I didn't really want to spend $150-200 on a PSU.

 

[ImJacksAmygdala]

Well I found an interesting review on the Enermax EG701AX-VE(W)SFMA V2.0 600 Watt PSU that advertises SLI support. Sadly although the review is fairly informative, it does not use a SLI system for testing. So far from what I can tell the power supplies that advertise SLI support are atleast 500W, are ATX 12V V2.0 compatible, and have dual 12volt rails of atleast 18A each, and offer 2 PCI-E 6pin plugs. It is also interesting that this paticular power supply is an active PFC or Power Factor Correction supply which might balance the load of two video cards more efficiently across the dual 12volt rails. If you look at the specs the current from both rails is added together to 35A. If you read between the lines PFC could effect the results described in the HomeLanFed article about combining current from 12volt rail1 and 12volt rail2.

http://www.bigbruin.com/reviews05/enermaxeg701ax/

http://www.bigbruin.com/review...iew.php?file=spec1.jpg

http://www.monarchcomputer.com...=M&Product_Code=100365

http://www.newegg.com/app/View...tion=17-103-461&depa=0

So I think I have finally justified for me atleast that the Enermax EG701AX-VE(W)SFMA V2.0 600 Watt PSU isn't just hype when it advertises "Support nVIDIA SLi Technology". It does seem to be one of the few power supplies out there that has 2 PCI-E 6pin plugs, is native 24pin, has dual 12volt rails at 18A each, and has power factor correction.

To bad we don't have any PSU reviews to prove this though.

In the end I think that a high current single 12volt rail PSU like the OCZ PowerStream line may actually scale better to an individuals power need and budget. They are great 24pin native supplies, but keep in my that you will have to use two PCI-E 6pin adapters and it takes a total of four 4pin molex plugs to power two video cards this way.

Worth reading 8 PSU review
http://techreport.com/reviews/2004q4/psus/index.x?pg=1

ENERMAX Noisetaker 600W Support nVIDIA SLi Technology PCI Express Ready Dual 12volt rails +12V1@18A, +12V2@18A - $179
http://www.newegg.com/app/View...tion=17-103-461&depa=0

OCZ PowerStream 600W PCI Express Ready, dual 12volt rails +12V1@20A, +12V2@18A - $209
http://www.newegg.com/app/View...tion=17-104-153&depa=0

ENERMAX Whisper II 2.0 SLI 535W Support nVIDIA SLi Technology, PCI Express Ready, Dual 12volt rails +12V1@20A, +12V2@18A - $109
http://www.newegg.com/app/View...tion=17-103-460&depa=0

OCZ ModStream 520W PCI Express Ready, Single 12volt rail +12V@26A - $120
http://www.newegg.com/app/View...tion=17-104-155&depa=0

OCZ ModStream 450W PCI Express Ready, Single 12volt rail +12V@26A - $90
http://www.newegg.com/app/View...tion=17-104-154&depa=0

Although I haven't seen a proper review, I think these power supplies make a good example of the features/power/current/price scale consumers should be looking for to run SLI systems.

Does anyone have a different view or see something I'm missing? Besides a proper SLI PSU review?

 

[Algere]

Originally posted by: Insomniak

Originally posted by: McArra
TT purepower 680W is also supposed to be for SLI setups and has 3 12V rails, 2 of them with 15A and another one with 8A, for a total of 38A. This is the PSU I will receive next week, is supposed to be really really powerful. I'm not using SLI though, I'm an AGP owner.

Amps still aren't additive - if any of your devices need more than 15A, like an A8N-SLI Deluxe, then you're boned.

At the very least wattage can* be additive & wattage is the product of amp & volts for a PCIe video card w/ 6-pin connector . If the A8N-SLI does indeed need 15A, One could guess that 11A (130W) max. is consumed from the PCIe 16X slots alone leaving about 4A (give or take) left for the motherboard's other functions assuming PCIe video card slots in SLI working in 8X mode ignore specs of power consumption for 8X PCIe cards. On the other hand if video cards in 8X mode under SLI do follow power consumption specs of 8X PCIe, that would only be 4.2A (50W) max. from the PCIe video card slots leaving about 10 or so amps left for the motherboard & if that's the case. I don't see how a A8N-SLI can consume that much - 15A.

*Dual rail PSU with PCIe slot/MB on one rail & 6-pin on the other.

 

[Algere]

If you read between the lines PFC could effect the results described in the HomeLanFed article about combining current from 12volt rail1 and 12volt rail2.

Didn't read the article but from what I remember Active PFC requires a bit more power for operation than an equivalent model with passive or no PFC.

So I think I have finally justified for me atleast that the Enermax EG701AX-VE(W)SFMA V2.0 600 Watt PSU isn't just hype when it advertises "Support nVIDIA SLi Technology". It does seem to be one of the few power supplies out there that has 2 PCI-E 6pin plugs, is native 24pin, has dual 12volt rails at 18A each, and has power factor correction.

To bad we don't have any PSU reviews to prove this though.

I feel like a broken record for saying this but check Falcon Northwest & VoodooPC. They both use that exact PSU (not exact; active PFC, other regular) for SLI systems. Assuming you believe they test those PSUs before making it an available option for such configurations.

 

[FastEddie]

Originally posted by: Algere

I feel like a broken record for saying this but check Falcon Northwest & VoodooPC. They both use that exact PSU (not exact; active PFC, other regular) for SLI systems. Assuming you believe they test those PSUs before making it an available option for such configurations.

Because someone else is using these in their builds doesn't justify the argument that dual rail psu's are perfectly fine to use in SLI builds. That's a cyclic response. That's like saying, Because Ford uses these tires, these tires are good. It's meaningless in any functional argument.

 

[ImJacksAmygdala]

Well the whole purpose of this thread is to help explain what makes a PSU "SLI compatible" whether it has a single 12volt rail or dual 12volt rails. I think if you actually read all of the information provided in this thread its safe to conclude that it doesn't matter if the PSU is a single 12volt rail or a dual 12volt rail. What really matters is that there is enough current provided on the 12volt rails with a high enough overall power rating for the system it is providing power. This is where people are running into trouble, and its understandable when some dual rail power supplies only offer 15A on a 12volt rail. I now understand why power supplies advertising "SLI support" are all dual 12volt rails, and I also understand why some single 12volt rail power supplies are still fully capable in SLI systems. I'm sure there is more to the story when you start considering ATX PSU's that require 24pin and 6pin adapters though.

 

[FastEddie]

I understand what you're saying exactly ImJacksAmygdala. I'm just tired of hearing arguments based on marketing strategies. There are two psu's on the market that merely got SLI stickers slapped on their labels because they were dual rail. That leads the buying puplic to believe these units where actually designed for the NF4 SLI chipset. Guess what---they wern't. The psu's in question were around long before NF4 SLI was even on the design table.

 

[Algere]

Originally posted by: FastEddie

Originally posted by: Algere

I feel like a broken record for saying this but check Falcon Northwest & VoodooPC. They both use that exact PSU (not exact; active PFC, other regular) for SLI systems. Assuming you believe they test those PSUs before making it an available option for such configurations.

Because someone else is using these in their builds doesn't justify the argument that dual rail psu's are perfectly fine to use in SLI builds. That's a cyclic response. That's like saying, Because Ford uses these tires, these tires are good. It's meaningless in any functional argument.

I like your response :thumbsup:

I don't think I've said it worked perfectly fine with SLI configurations but one could imagine so. It would be a costly mistake for both OEMs if that PSU didn't work at all with their SLI systems, so yes I assumed that particular PSU works with SLI since the use of that PSU by both OEMs was good enough for me to believe it would work with SLI. Seeing that there are no reviews that I know of (yet) with that PSU hooked up to a SLI configuration. Hence the response with probable proof that it does work with SLI.

BTW there's a slight difference between your analogy & my response. My response was that it would work in contrast to not work while your analogy from what I can tell is based on quality or PSU that works (Ford tires) versus PSU that works better. Clearly factory tires that come with Ford vehicles already work.

Perhaps a better analogy/comparison would be flat tires versus air filled tires, no?

 

[Algere]

Originally posted by: FastEddie
I understand what you're saying exactly ImJacksAmygdala. I'm just tired of hearing arguments based on marketing strategies. There are two psu's on the market that merely got SLI stickers slapped on their labels because they were dual rail. That leads the buying puplic to believe these units where actually designed for the NF4 SLI chipset. Guess what---they wern't. The psu's in question were around long before NF4 SLI was even on the design table.

... But the improvements to the existing dual rail design weren't. e.g. higher amperage on rail, shared amp capability, & better distribution of +12V power to devices to even the load better on both rails in contrast to older dual rail design(s) where too many devices were dependant for power on one rail of a two rail design.

 

[FastEddie]

My analogy for Ford was Firestone tires. We all know what a disaster that turned out to be.

You are correct in that there aren't any current reviews of how well dual rail psu's work with SLI based systems. But you are incorrect in believing that there is any shared amp capability between the two rails. That's what makes them not fitted for these types of systems. There is no combining of the two rails. It's not like one rail gets maxed and the second will just pick up the slack. The two rails aren't connected to each other in any way, shape, or form. All you can hope for is that the demand on each individual rail is never pressed to full capacity or beyond, because there is no "beyond" once that rail is maxed.

How these psu's split rails is defined. The square 4-pin 12v AUX connector is one rail. The ATX power connector is another rail. The specs on one particular unit combine all other device connections from the 12v AUX rail. The specs on yet another unit combine all other device connections from the ATX main power rail. It's a balancing act, but one in which either of those rails will have to work at or near 100% of it's output to keep the system running. And as I've said before, that is not efficient use of a psu. A microsecond of demand beyond PEAK and you are in trouble, because the second rail can't assist the maxed rail, or pick up the slack from the maxed rail.

Take a look at the rating stickers on these psu's. The list two +12v rails, and list the max amps for each of those rails.

 

[Algere]

But you are incorrect in believing that there is any shared amp capability between the two rails. That's what makes them not fitted for these types of systems. There is no combining of the two rails. It's not like one rail gets maxed and the second will just pick up the slack. The two rails aren't connected to each other in any way, shape, or form.

When I said shared amps, that's not what I mean by shared amps. Remember this thread? (13th post down)

How these psu's split rails is defined. The square 4-pin 12v AUX connector is one rail. The ATX power connector is another rail. The specs on one particular unit combine all other device connections from the 12v AUX rail.

See thread link above.



P.S. Also note my previous post in this thread (second post down; second reply).