12v rail

[x2 3600 rules sazakky]

Which parts of the computer use the 12v rail. I have a Athlon 3500 am2, 1gb ddr2 533, 2 hard drives and a dvd drive. My generic 550w PSU has 18amps on the 12v rail.
Will a x1900xtx run in my comp.

 

[Maximilian]

My guess would be no. Its 550 watts but only has one 12v rail? My 520 watt has somthing like 30+ amps across three 12v rails. The x1900xt is a power hog anyways, there was a recommendation on ati's website for an ample 12v rail but i cant find it anymore.

 

[Captante]

No, its not sufficient ... you need approx 26 amps on the +12v rail to run an X1900XTX without risking problems.

 

[Puffnstuff]

A decent ps would have well in excess of 30a on the 12v rail. Hell we've got an old enermax 485 that has 18a x 2 on the 12v rails. Take a look at the newer ps models and they've got upward of 50a on the 12v rails. My enermax galaxy 850 has 120a across 5 12v rails. Amps are the key here and the 12v rail is the most taxed on a modern system.

 

[TheDrD]

Whats the point of having 5 12v rails?

 

[Puffnstuff]

The galaxy is set up for a triple sli so that each would tax a different rail among other things. Not that I would use that but I'm about futureproofing with my most recent setup which is why I got this case and that ps. I don't skimp on my ps's. To each his own though. A smaller lower capacity ps that is being taxed harder is also heating up more. Heat kills electronics and I want my stuff to last a while besides this is my build philosophy. To each his own.

 

[TheDrD]

How does a Single Rail PSU work?

24 Pin Connector- 12v (I think)
8 Pin Connector- 12v(I think)
6 Pin Graphics Card- 12v

3 Rails?

 

[Zepper]

OP,
. The PSUs I'm selling in my FS/T listing (link is in my sig. below) could handle your system because the power is properly allocated among the rails for a modern system.

The DrD,
"How does a Single Rail PSU work? "

All +12V wires (the yellow or yellow with stripe ones) are connected to one point within the PSU (just like the +5 and +3.3 wires) - IMO, it is the correct way to build a PSU. The split rails concept is yet another bad idea from your friends at Intel.... Many so-called "split rail" PSUs actually aren't.

Check out the new Silverstone OP-620 review at http://www.jonnyguru.com - 50+ amps on the single +12 rail! More companies will be returning to the single rail to reduce complexity.

Almost nothing on the mobo or add-on cards actually uses +12V directly. The few things I know of that are found in every system that use +12V directly are drive motors and fans. All other parts that are powered off the +12 rails use lower voltages (+5V or under) with local regulation off the +12 rail (CPU(s), PCI Express, some video cards). Some parts use the +5 or +3.3 directly, but even those are often locally regulated and filtered downwards too. The local voltage regulators are more efficient and stable with a higher difference between the input and output voltages - thus the use of the +12V rail rather than the +5V rail for supplying critical high-power devices. Also the current draw on the +12 rail(s) will be MUCH lower (less than half) for delivering the same amount of power, thus preventing the connector melts that were a fairly common occurrence when the +5V rail was heavily relied upon.

.bh.

 

[tcsenter]

Originally posted by: Soviet
My guess would be no. Its 550 watts but only has one 12v rail? My 520 watt has somthing like 30+ amps across three 12v rails. The x1900xt is a power hog anyways, there was a recommendation on ati's website for an ample 12v rail but i cant find it anymore.

That recommendation is a worst case scenario for the power requirements of the entire system, to account for the widest range of power supply rating methods (from the conservative to inflated). The X1900XTX requires not more than 10A MAX of +12V current according to reasonably accurate measurements by Xbit Labs (9.75A).

Case in point, ATI and partners recommend 38A for CrossFire setups and 30A for single card. If one card requires 30A, how could two cards require only 8A more?

All +12V wires (the yellow or yellow with stripe ones) are connected to one point within the PSU (just like the +5 and +3.3 wires) - IMO, it is the correct way to build a PSU. The split rails concept is yet another bad idea from your friends at Intel.... Many so-called "split rail" PSUs actually aren't.

Nothing inherently wrong with split rail concept. There is something wrong with not knowing which connector is powered from which rail, thus not being able to calculate how much load will be placed on any given rail. That is where ATX12V 2.x went terribly wrong.

EPS12V (SSI) solved this by requiring (not optional) that particular connectors/components be powered strictly from particular rails, so calculating the load that will be placed on each rail is possible. No more guesstimation work. If you know the connector, you know the rail that powers it and the maximum supported load characteristics of that rail (assuming the PSU truly complies with EPS12V spec).

 

[stevty2889]

Originally posted by: x2 3600 rules sazakky
Which parts of the computer use the 12v rail. I have a Athlon 3500 am2, 1gb ddr2 533, 2 hard drives and a dvd drive. My generic 550w PSU has 18amps on the 12v rail.
Will a x1900xtx run in my comp.

My 450w Antec couldn't handle a 1900xtx, so I can almost gurantee that it would kill your PSU. The cpu, hard drives, cd/dvd drives, fans, and video card all run off of the 12v rail.

 

[Zepper]

Yes, there is something wrong with split rail as it is often implemented. You may end up not being able to use all the +12 power that your PSU is capable of producing. Say you have a PSU that can produce 36A@12V and two rails with a limiter on 12-1 of 16A and on 12-2 of 20A - say that 12-1 never exceeds 10A draw and that somehow you have accumulated a load of 23A on the other - your PSU could do it as a single rail, but not as a split. The highest limiter on split rails should be 20A per rail per the ATX spec. so even the way a lot of them do it now with 20A limiters on each rail (regardless of spec) couldn't do it. Very few are true dual rail with separate secondary windings for each 12V rail. All rails are usually off of one 12V secondary winding with current limiters of 20A max to "split" the rails.

.bh.

 

[Zepper]

DrD - Why five 12V rails?

Well, if your PSU can crank over 80 Amps total on the +12 (as some of those new super power PSUs can), then you'll have to do more than 4 rails if you want to adhere to the ATX spec of limiting to a max of 20 Amps per rail (who knows, some idiot might grab their 12Vrails while taking a bath and kill himself - we gotta keep some ways to clean the gene pool, don't we?) :roll: . There is no other reason, which is another reason why the split rail idea is a failure in my book.

.bh.

 

[myocardia]

Originally posted by: Captante
No, its not sufficient ... you need approx 26 amps on the +12v rail to run an X1900XTX without risking problems.

More than that: my 256 MB X1900XT's current draw. Note that that is the video card only, not the whole system.

 

[Zepper]

Where is that number coming from - no way a danged video card can be pulling 27A off the 12V rail by itself, it must be off the local regulators at some lower voltage(s).

.bh.

 

[Captante]

Originally posted by: myocardia

Originally posted by: Captante
No, its not sufficient ... you need approx 26 amps on the +12v rail to run an X1900XTX without risking problems.

More than that: my 256 MB X1900XT's current draw. Note that that is the video card only, not the whole system.

I'm pretty sure the 26 amp number came from ATI's website & that it was a minimum ... more would certainly be better, but that number was for a modest system with an X1900XT, not just the video card itself.

I have serious doubts about the accuracy of that indicated power draw from ATI tool.

 

[soccerballtux]

Originally posted by: TheDrD
Whats the point of having 5 12v rails?

That's 5x12v and you're halfway to the 120 volts you get from the socket in the wall. Isn't technology great?

 

[tcsenter]

Originally posted by: Zepper
Where is that number coming from - no way a danged video card can be pulling 27A off the 12V rail by itself, it must be off the local regulators at some lower voltage(s).

Yep, probably the GPU VCC. His card is overclocked quite a bit. Even assuming this added another 30W over default X1900XTX under max load as measured by Xbit Labs (120W), 27A of +12V current is more than twice that (324W).

In fact, the voltage @ 27.3A works out to be 5.5V for 150W and 4.4V for 120W, neither of which make any sense. The GPU VCC should be 2.5V or less. Any way you slice it, ATI Tool is not measuring +12V current.

 

[myocardia]

Originally posted by: tcsenter

Originally posted by: Zepper
Where is that number coming from - no way a danged video card can be pulling 27A off the 12V rail by itself, it must be off the local regulators at some lower voltage(s).

Yep, probably the GPU VCC. His card is overclocked quite a bit. Even assuming this added another 30W over default X1900XTX under max load as measured by Xbit Labs (120W), 27A of +12V current is more than twice that (324W).

In fact, the voltage @ 27.3A works out to be 5.5V for 150W and 4.4V for 120W, neither of which make any sense. The GPU VCC should be 2.5V or less. Any way you slice it, ATI Tool is not measuring +12V current.

Where are you coming up with these crazy numbers, bub? "His" (my) X1900XT is not overclocked any at all, to begin with. And yes, that's the 12v draw, although it's likely the 12v draw for the entire system, since 27A @ 12v is 327 watts.

That screenshot was taken with an X2 4400 at stock speed and vcore, along with the X1900XT at stock speed and voltages, so 27A would be about right, since the processor itself would be using 8-9A of 12v power, along with around 2A total for the HD and fans, that would leave ~16A for the video card, which seems just about right, from what I've heard.

 

[Zepper]

I know of no way to read currents by software in a PC in general - the only way that current could be read is from the vid card itself which runs at much lower than 12V to most components on the card so it must be reading the local voltage regulator(s) right on the vid card.

.bh.

 

[tcsenter]

Originally posted by: myocardia
Where are you coming up with these crazy numbers, bub? "His" (my) X1900XT is not overclocked any at all, to begin with.

According to the ATI Tool screen capture, the default clocks are 500 core and 600 mem. The current operating core and mem clocks are reported as 621 and 720, respectively. The 'find max core and mem' buttons are greyed out (i.e. they have been activated), leaving only the abort and default reset buttons available. Based on my knowledge of ATI Tool, it appears that you are using the 'find max core and mem' overclocking feature of ATI Tool and the actual clocks have been increased to 621 and 720. Correct me if I'm wrong.

That screenshot was taken with an X2 4400 at stock speed and vcore, along with the X1900XT at stock speed and voltages, so 27A would be about right, since the processor itself would be using 8-9A of 12v power, along with around 2A total for the HD and fans, that would leave ~16A for the video card, which seems just about right, from what I've heard.

10A MAX of +12V current for X1900XTX, as I've already established (9.75A measured).

Can't get even 300W from this setup factoring all components at 100% MAX load, which is practically impossible. At any rate, ATI Tool can't accurately calculate or extrapolate total system power or +12V current without being able to measure the voltage drop (delta) to some reasonable degree of accuracy (or using induction clamps). I don't know what ATI Tool is measuring, but it ain't +12V current. It might be guesstimating based on some set of assumptions...

 

[myocardia]

Originally posted by: tcsenter

Originally posted by: myocardia
Where are you coming up with these crazy numbers, bub? "His" (my) X1900XT is not overclocked any at all, to begin with.

According to the ATI Tool screen capture, the default clocks are 500 core and 600 mem. The current operating core and mem clocks are reported as 621 and 720, respectively. The 'find max core and mem' buttons are greyed out (i.e. they have been activated), leaving only the abort and default reset buttons available. Based on my knowledge of ATI Tool, it appears that you are using the 'find max core and mem' overclocking feature of ATI Tool and the actual clocks have been increased to 621 and 720. Correct me if I'm wrong.

That screenshot was taken with an X2 4400 at stock speed and vcore, along with the X1900XT at stock speed and voltages, so 27A would be about right, since the processor itself would be using 8-9A of 12v power, along with around 2A total for the HD and fans, that would leave ~16A for the video card, which seems just about right, from what I've heard.

10A MAX of +12V current for X1900XTX, as I've already established (9.75A measured).

Can't get even 300W from this setup factoring all components at 100% MAX load, which is practically impossible. At any rate, ATI Tool can't accurately calculate or extrapolate total system power or +12V current without being able to measure the voltage drop (delta) to some reasonable degree of accuracy (or using induction clamps). I don't know what ATI Tool is measuring, but it ain't +12V current. It might be guesstimating based on some set of assumptions...

"Correct me if I'm wrong."-- I like you already, but you're wrong. The card is running at the default 3D clocks of 621 core, and 720 memory. The numbers at the top are the default 2D clocks of 500 core, and 594 memory. I have no idea now why the "Find Max" buttons are greyed out, unless maybe I had attempted to see how high it would overclock? Anyway, here's a brand new screenshot, just for you: link, with those buttons not greyed out. As a matter of fact, as you can see from the Asus page pertaining to my card, it's actually slightly underclocked, for whatever reason.

As far as how accurate ATI Tool is, or where it gets it's amperage info from, I have no idea. A google search doesn't turn up anything that's very informative, as far as I was concerned. Possibly the cards themselves have the ability to report what ATI Tool needs to discern that info, like they now do with the three different temps, and the fan speed? I'd personally guess that it's just a rough guesstimate, since it already knows the voltage for both the core, and the memory, and also knows what speed both are running, which makes sense, now that I think of it.:laugh: