Well, it seems that the Nvidia Website for Aus has the RTX2080TI in stock for 1899 AUD.

I'm contemplating buying 2 for myself.

Yeah, I already have the Titan XPs in SLI and in all honesty the only real game I'm waiting for is the new Metro and 3 Kingdoms Total War, which may or may not be worthy of dropping 4 grand AUD for a set of cards, especially if CPU limited. My retirement plan this December is to sit at home and play vidyagaems.

The TDPs are indeed something very high and looking at the board design, it does seem to be a lot beefier with enlarged heavy copper ground planes for better thermal performance. So I expect thermals to be a limiting factor for any clockspeed bumps.

HDR seems nice but I'll be honest and say I've been spoiled by Dolby Vision HDR on OLED. That shit is awesome. Come to think of it, for 4k, I'm better off buying a new .45 pistol hahaha

lol the good ole vacuum cleaner GTX480.....long time never hear of that monster.

Now all the industry needs to do is make an OLED Monitor with a 30 inch size with Dolby Vision and I'll definitely pour a mint into my PC.

Maybe it's also worth waiting for the benchmarks and IQ tests to come out, but me knowing me, I'm a fool waiting to be parted with his money.

With the inclusion of many AI based features and the combination of ray tracing and raster graphics which each have their own processing quirks, it's hard to say. It really depends on the game and the load, the API and how well the devs have done their homework.

Traditionally longer pipelines mean less CPU "calls" but microbursts of activity favouring raw clock speed, shorter GPU pipelines favour more parallelism and a steady feed. In theory lah....

But there are a few things in play....

- The Turing's onboard scheduler is newly built and a lot more intelligent when it comes to dealing with complex loads. It's a lot more efficient this time around.

- Dual datapaths. Remember when hyperthreading came out and it made intelligent use of a scheduler and branch prediction to maximize the efficiency of processors by best using both ALUs (Integers) and FPUs (Floating Points). Nvidia deals with it by offering 2 seperate datapaths for each execution unit. But it's up to the dev to ensure that instruction ratios are optimal to ensure that pipeline is fed efficiently.

- 64kB L0 instruction cache for each EU with further breakdown of memory units. Makes for better data fetching delays, but incase there's a mistake (a cache miss), I predict that it'll be more painful vs the Pascal. But that also means potentially a lot less CPU or RAM calls for a miss due to the fact there's another cache layer.

So?

Generally this kind of architecture means 2 things, parallelism is increased, so pipelines can be cut short, with an emphasis on even better shader performance.

So what does this mean? This graphics card probablywon't be limited by CPU clock performance most of the time, but by shader and programming optimizations on the part of the developer and will probably be less affected by clock speed.

Note/Disclaimer : I may be 100 percent wrong lol. I spend most of my days working with C3MR LNG plants and doing RFIs for chem processing units rather than computer engineering these days

lol protip gais, if you're gonna flash your Lenovo Y740 to a less power conservative VBIOS for more performance, make sure you use the MSI BIOS instead of the Razer one.

Recovery was a bitch.

Probably not as I don't have enough time for gaming these days and I'm far away from my main gaming rig in Brisbane. Covid travel restrictions

Building a house here in Perth, if it goes according to plan and gets finished on time, I might jump on the next gen. See how lah.