Read more.New add-on aims to improve system stability by adding power directly to the PCIe bus.
Read more.New add-on aims to improve system stability by adding power directly to the PCIe bus.
Erm, the power has to come from somewhere, and I don't know many PSUs that don't use the same rail for the motherboard line as for the molex lines...
Not only that, but doing it through this route is cutting out the entire 6/8/12/whatever phase circuitry that the motherboard has.
I can't - PCI express has very clear power requirements per slot. Intel and AMD also have very strict guidelines with their chipsets when it comes to power use.We can imagine a scenario where two or three overclocked GPUs could start to draw more power than the PCIe bus could normally provide, potentially impacting system stability
If mobo makers are selling motherboards that are unable to do their job, then they need to be returned as not fit for purpose.
In fact, another thought - We have PCI Express power that's coming through a minimum of 4 phase circuitry (I've not seen a modern board with less) to ensure a clean power supply to the board, and then we have something from the PSU being directly linked into the motherboard, without any evidence of a filter (and given it's size, it's very unlikely).
Looks like snake oil to me.
Well, their one gpu partner keeps making cards that disregards PCI-E spec, so they had to come up with something. Remember, the 580 is only sub 300 watts due to the driver. Even gpuz turns it into a 350+ watt card. Snake oil may be needed to give a a false sense of security to Nvidia buyers, also these cards aren't selling in any kind of major volume so a new revenue must be generated. Even is it's just gimmicks.
This product is geared to the high end overclockers, shoving several cards with massive overclocks while under LN2. That breaks any spec sheet - PCI slot by default delivers 150W, 6-pin 75W, 8-pin 150W. If each card is pulling >400W, an extra bit of juice is more than welcome. Not for $20 though.
It all depends on context.Looks like snake oil to me.
I've grouped these together as they are kind of aiming in the same direction.
@B_G: I know exactly what you mean with the 'wattage unlocking', but AFAIK the extra wattage is pulled from the power connectors and not the PCI Express slot. More than happy to be proved wrong on this though (and then I could see its application a little bit more).
@borandi: I've been there and done the whole LN2 / Vapo thing, although it's been many years. The part I'm struggling to get my head around is that simply applying more voltage / current to a lane like this, doesn't make it useful, doubly so when you have different phase passes between the two supplies and then claiming it's use is for extreme overclocking - it's the exact kinda thing you'd want to avoid. Any modern PWM on a motherboard could supply 150W over the PCI Express slot with ease (even multiple slots), so adding power in this way seams really, really crazy.
It also comes back to the power draw on the slot - does any card actually pull the additional power over the slot and not the connectors? In all honest, I'd be surprised if there wasn't a safety cut-off, it was certainly not uncommon on AGP systems.
Keep in mind that the data bus signal is entirely separate to the +12V supply over PCIe as shown here. The signalling bus is just as (if not more) important as the main supply, and the part I'd argue is going to suffer first when doing extreme overclocking of the likes of LN2 with a good power unit.
Thoughts?
I can't comment on the likely efficacy of this because I don't understand the finer details of power regulation - but I can say that Biffa over on OcUK has had issues with mobo ATX connectors melting/dying in some form or another in the context of three/four high-end, overclocked GPUs Folding 24/7 on motherboards that did not allow supplementary power from a motherboard-mounted molex socket.
@Agent,
Seems like it could be shaping up to be an exclusive EVGA feature to me, I'm no mobo designer but I have to assume that the 1oz copper pcb(2ozs in the case of gigabyte) is rated for only so much power. You start randomly injecting into the bus and your engineers have to take that into account. Could it possible promote instabality rather than stability ?
There are currently 1 users browsing this thread. (0 members and 1 guests)