I've read quite a bit about the so-called 35W limit for the Optiplex SFF's PCIe x16 slot. But unless Dell used inferior quality motherboards with substandard copper traces, I have a feeling the limitation is more likely based on the small 240W PSU with its 17A (+12v) rail, rather than on any electrical limitations of the board itself. The poor air flow in the small case is probably another major reason for the 35W limitation.
Or does someone have evidence this is not the case? Recently I have come across several Optiplex SFF users that have been using 60W cards with no ill effects... other than risking overload of their PSU. I'd be interested in hearing from other Optiplex 790 SFF owners and their experience.
Its not 35w its 25w and its on the slot as such. X1 slots the limit is 10w.
USFF Has no slots.
VIDEO BELOW 5450 INSTALLED IN GX280 SFF
Actually, that's not my board. This is mine:
Mine has no wattage declaration printed on it. And the PCIe x16 slot is blue, the PCIe x16 (x4 electrically) slot is black. The owners manual has no disclaimer anywhere for the slot wattage limitation.
But my point still stands no matter 25W or 35W. I feel it is the PSU that is the "limitation", not the physical slot or the board's copper traces. The PCI Express interface itself is on the CPU, so that wouldn't be an issue either.
The limitation is the power supply. That's why an external unsafe fire hazard hack job works.
Exactly. Thanks for verifying that, Speedstep. In fact, shortly after I posted this (back in Sep), I installed a GTX 750 Ti low profile Gigabyte gfx card in my Optiplex 790. Maxed out at 100% GPU usage, the draw on the +12V rail is an estimated 14 to 14.5 amps max. Since then, I have been running this setup now every day for many hours per day including gaming.
Nice that you have got it work and lucky me to finally find someone with a similar setup I am trying to figure out!
GTX 750 Ti does not have external power connectors, right? So the situation is different than what Speedstep presented earlier. I was already considering switching to GPU with external power connectors to overcome the 35W PCIe x16 -limitation, like is done in the Speedstep's video.
I don't understand this part "+12V rail is an estimated 14 to 14.5 amps max". Is not that like 150W+? Or did you mean 14 to 14.5 watts? Have you been able to figure out what is the max wattage that can be pushed thru the slot?
Did some digging. According to this it would like to have 20A. I understand that these requirements are little over what is required in reality, but how do you know the GPU is getting all the power it needs exactly? Would not it take all it could get @ 100% and the figure you gave would either be a) what it needs or b) what the slot is able to deliver (it could require more for full function)? How do you know which is the case?
The estimated max draw for the card you have is 20A. Sure in reality it is less. But how do you know that the 14A delivered to the GPU working @ 100% is all that it needs and not just all that could be delivered?
Thanks for the reply SammyVV. I wasn't concerned about the so-called wattage limitation of the PCIex16 slot. There is no such disclaimer on my motherboard or any of the manuals. Therefore I am only concerned with the current demand on the +12V rail and the heat produced. (The PSU's +12V rail is rated for 17A)
I estimated the amps (yes, I meant amps) by taking an amp reading with my Kill-A-Watt meter set for current. It read 1.4 amps at its max on the primaru side of the PSU. Since the PSU's transformer is a 10:1 step-down ratio (120V primary/12V secondary), I estimated the total secondary amperage at ~14 amps not counting power factor loss*. Since the +12V rail is responsible for nearly all of that amp draw, I assumed the most it could be drawing was 14A or so.
I ran the Firestrike benchmark and got a score of 3847 (I5-2400 and GTX 750 Ti). Max current draw was an estimated 12.4A
* Since I'm taking my reading on the primary side of the PSU, power factor would be in my favor. I.e. the secondary side isn't using all that power. Some is loss to heat, and capacitve and inductive reactance.