Recent content by EdZ

The big limitation of segmented chips is that it puts a massive constraint on bandwidth and latency between the 'chiplets' (even with EMIB, the problem is far worse with through-substrate links). Thinking of a multi-die package like a multi-socket system crammed into a small space gets you into...

A few nit-picks: The adhesive-attached front plate only needs to be removed to disassemble the cooler into its individual components. To remove the cooler entirely form the card (e.g. for watercooling) you only need to remove the screws in the backplate and the whole thing comes off, same as...

At it's core, the Type C connector has 4 fixed (two lanes duplicated for passive reversibility) pins for USB 2.0, 4 ground pins, 4 power pins, 4 control pins (duplicated for passive reversibility, and 8 pins for high bandwidth lanes. Those can be USB 3.1 Gen 2 lanes, or DisplayPort, or...

You can adapt a Type-C output with DP alternate Mode to a DP connector, but you cannot go from DP output to a Type-C connector with DP Alternate Mode. Such an adapter would need to incorporate the USB connection as part of the Type-C standard, and that USB connection would need to be closely...

Can we get an 'En-bearded' forum banner?

Did some very basic benchmarking of the 2080Ti vs. 1080Ti for VR using FCAT*. For things already comfortable within the 11ms limit there is little change, but start pushing things and the difference is dramatic. e.g. Raw Data where 2xSS would be completely in ASW on the 1080ti is damn near rock...

This does rather remind me of the situation with the HG Osmi. In both cases (pardon the pun), a company created a custom case for their boutique PC business, with the case being designed only to be part of a premium integrated setup, with manufacturing cost secondary to achieving the desired...

There's HDR, and then there's 'HDR'. There are a whole pile of monitors that are basically SDR panels, but that accept an HDR input and attempt to squish that entire range down into what they can actually display (similar to the old 'HD' TVs that accepted HD inputs and displayed them on an SD...

Not quite the same as the reference cooler. The vapour chamber/finstack block is similar but not the same, and the midplate (provides heat distribution for the VRM and memory with some thermal isolation from the GPU) appears to be absent.

"It's too expensive and the new features don't work yet" is somewhere we've been before, with the introduction of Hardware Texture & Lighting, and with the introduction of fully programmable shaders (CUDA cores). The 2080 and 2080Ti sit right on the price/perf curve with the 10xx series in...

It seems like once again the FE is the way to go if you don't want a GPU that is covered in plastic embaressment.

Got the SFF bike lift assembled, this weekend I should get the insulation installed in the shed workshop then I can wheel the VFR in and start pulling it to bits. I'd love one, but the upfront cost here is huge and charging it is a pain: a Zero S without the fast charger tank (still takes 2...

Great for HPC use, but probably not much impact for the consumer side of things.

Correct, though I don't think actual NVLink bandwidth has been announced (GV100's PCIe varaint needed two card-edge NVlink connectors, and still didn't have a full lane breakout available). Bandwidth and cache-cohernacy are unrelated to SLI's use of AFR (or SFR as in the initial guise). There...

NVLink is a really neat interconnect, but it doesn't actually do anything to tackle the issues that make multi-GPU difficult in the first place: needing to split draw calls between discrete devices (even with a high bandwidth NVLink bridge both GPUs still have to be treated as different NUMA...