Updated: May 13, 2019
Seven months. Disclaimer: The following post contains Sarcasm.
It was brutal. Well ehm, actually it wasn't that bad. Either way this is the Fermi-based GF108 graphics processor surrounded by its quartet of GDDR3 chips. Each of those hooks into the silicon via 32-bit wiring so that's the 128-bit bus populated. Oh and they're running at 1.8Gbps. So you get an absolutely earth-shattering 28.8GB/s of Pure Unrestrained Bandwidth to that tiny little core. And they're 256MB in size so that's 1GB VRAM. Oh, I didn't even say what this is, it's a GeForce GT 540M. Ripped out of my old Toshiba P750 laptop.
GF108 is around 116mm² in size (I didn't measure it, but Wikipedia is usually correct, right?) and contains about 585 million transistors. That's 0.58 billion transistors. That's not actually a lot of transistors. It's built on TSMC's now ancient 40nm process.
So you're looking at a pair of the Fermi Streaming Multi-processors (they're basically GPU cores) and each contains 48 CUDA cores and 8 texture units. (the smaller Fermi chips have a different SM layout, GF110 and GF100 only have 32 CC per SM)... But you get a grand total of 96 shader processors and 16 texture units on this massively powerful beast of a GPU. Fermi has a decoupled core/shader clock so the GPU core (ROP, Raster, Geometry, etc) is running at 672 MHz and the shader processors are 'doubled-pumped' to 1344 MHz. Did I mention this GPU has 4 ROPs? Four.
Anyway performance today is abysmal and it's weaker than basically every Integrated GPU available, even the lowly UHD 620 is likely comfortably ahead. This GPU is now quite literally, obsolete. Still I didn't have a picture of it so here it is. Did hold me over while I had to live at my Aunt's though.
Oh, and Pusheen says hi. :3