An picture of an unreleased graphics card PCB has appeared on-line claiming to be that of an Nvidia GTX 1180. The PCB options connections for eight reminiscence modules, a average GPU die, 10 part VRM, twin fan headers, and 6+Eight pin energy connectors. It additionally comes with an NVLink connector. Odd.
The picture reveals either side of a probably finalised reference PCB, though whether or not or not that is the fabled GTX 1180 remains to be very a lot up for debate. The reminiscence config will enable for both 8gb or 16gb chips in every slot, for a complete of both 8GBs or 16GBs of, what we will solely assume to be, GDDR6 reminiscence.
Already constructed the very best gaming PC? Here are the best gaming keyboards to match.
The present GTX 1080 options 8GB of GDDR5X reminiscence. While we’d like to see the flagship card are available with 16GB of GDDR6, that may be ludicrously costly. 8GB of speedier GDDR6 reminiscence would nonetheless provide higher bandwidth at lesser voltages, and it appears just a little unusual for Nvidia to bump the reminiscence so considerably with out all that a lot impetus from shoppers, devs, the competitors, or anybody aside from fanatics.
As for the ability supply, the leaked PCB picture from Baidu (through VideoCardz) depicts some significantly chunky VRM parts for a reference Nvidia GeForce design. This leaked GTX 1180 PCB has sufficient room for 10 entire phases price of energy componentry, probably 8+2 part, powered by 8+6 pin connectors. That’s greater than even the GTX 1080 Ti, which was constructed on the identical PCB because the Titan Xp.
There’s additionally the NVLink connector — or what appears virtually equivalent to 1 in any case. These connectors are used on Nvidia’s actually costly GPUs to extend bandwidth past what’s obtainable to system over PCIe. It’s additionally the premise for the NVSwitch join that powers the big 16-GPU DGX-2 AI system Nvidia rolled out earlier this 12 months.
Nvidia’s present multi-GPU interface, SLI, which is sort of a bit much less bandwidth-heavy than NVLink, hasn’t confirmed all that standard with the Pascal technology. Developers simply aren’t all that fussed on the concept it appears. As such, it appears doubly odd that Nvidia would put an much more costly interface on a GeForce card for the few stalwart multi-GPU aficionados nonetheless left standing.
So possibly this is not the subsequent GeForce GTX 1180. The NVLink connector, the dual-fan headers (Nvidia reference designs normally solely have one radial fan), the chunky energy supply on a reference PCB… it simply appears just a little too costly for the likes of us players, trying fairly a bit like what you would possibly anticipate from a high-end workstation card. Not to say us players are low cost, however most of us will not be spending hundreds of {dollars} on single graphics playing cards both.
Source
