mouacyk

If a Russian can resolder 16GB of GDDR6 onto an RTX 3070, surely someone on Extreme Hardware can glue together a broken standoff .

Would have liked to see all the VRAM go through a memory check. Also, does anything know how to address above 8GB on the modded GPU? Didn't see any use case that demonstrated this.

What are your use cases? For twitch fps like CS, my last experience with wireless several years ago wasn't so good. I believe it was the MX Master. I'm still using a G502 to this day, but would like to unplug.

Guys, $330 for 10900KF at MC is unbeatable. Too bad I don't live next to one and already run a 9900K direct-die.

I hope it's also bots putting up these sales posts. Otherwise, it's also a waste of human potential and time...

hello

It's for real and some of the folks on here may be old enough to remember this classic. https://blizzard.gamespress.com/Diablo-II-Resurrected

Slow day.

Welcome.

It's an old case now from Silverstone. Unfortunately, it does not support ATX but I wished it did. It does support up to mATX, so there's expansion opportunities. I have been looking at the Thermaltake Tower 900 for its accommodation of bigger mobo's, so you never have to compromise on VRMs. VRMs on mATX Z390's are abysmal, except the insanely expensive Gene. Also installation should be much easier on the Tower 900. However, it is $300 and doesn't seem to have a bottom intake fan to aid in the vertical air flow.

As much as it is a PITA to work with, can't seem to give up the vertical air flow and external cooling. The recent mods will allow for side 120mm radiator and better looking passthrough fittings to my GTX360 rad. Definitely tempted to get a MO-RA.

Only makes sense since Tesla already invested 1.5B directly.

No doubt they were trying to get money while they could. They knew eventually that dedicated hardware would outdo their tech. Here we are. A first look at Unreal Engine 5 - Unreal Engine (June 2020) Using Turing Mesh Shaders: NVIDIA Asteroids Demo | NVIDIA Developer Blog (December 2018) Mesh Shader Possibilities – Nathan Reed’s coding blog (reedbeta.com) (September 2018)

@UltraMegaIt certainly is, and not just interesting but almost mandatory to help linearize the exponential difficulty of resolution increases. Anyone remember "Unlimited Detail" graphics from several years ago? (213) Euclideon Unlimited Detail - A fair comparison - YouTube Makes me wonder if something transpired between that company and what's happening with the advent of geometry shaders, direct or not. While that was an entirely software solution, its aim was no different - hence its name. If one starts to see the whole picture, it makes sense why RTX IO and SAM are being introduced now. They are required to stream meshlets efficiently to geometry shading hardware.

Apparently, mesh shading requires hardware support a la "geometry shaders". It dawned on me yesterday why I had not been interested in NVidia's Asteroid demo when I still ran a 1080TI. It needed hardware that was only available in Turing onward. So while it would be great to get remasters of classics to run better on modern GPUs, it's impossible to run this technology on older GPUs.

More excited for the near 10x performance uplift on Ampere. Combined with DLSS, this will truly make 8K gaming at 60fps a reachable goal. Any reason why your source link is not a hyperlink?

Just an fyi to anyone installing water blocks -- check that standoffs are fully tightened. Somehow, I got a Bykski block for my 3080 that had a delta of 36C on GPU temp, similar to everyone else on the internet clamoring about how great water blocks are at keeping GPUs in mid 50C's. That didn't feel right, having seen a fully unlocked 1080 TI with XOC BIOS pull ~400W+ in low 40C's in my same loop. Turns out, several standoffs around the die mount surface had loose standoffs and prevented optimal contact. Afterwards, delta dropped to a typical 15-17C delta. a = stock mount; b = thicker washer; c = tightened standoffs with thicker washer

It was probably problematic for NVidia in yesteryears that two low or mid-range cards could be combined at a lower cost and get more or equal performance to a higher-tier card. There was just no way for marketing to sell that kind of a solution, without reflecting harm on themselves. And as we have seen, they also intended to increase the pricing gap between the performance ranges, so allowing SLI made even less marketing sense. All good free things must come to an end, as the masses will exploit it to the detriment of enthusiasts and investors alike.

Where is the market, I can't find it.

Can't imagine any userbase booting up Windows 10, open up the Start menu and stare at it all day. The only other userbase that may have to stare at windows decorations all day are sys admins. And even they might prefer Windows 98-esque styles for less distraction. The rest of everyone else should be looking at the content area in a spreadsheet, an app, a browser, or an IDE.

Ashes of the Singularity did explicit multi-GPU pretty well, even across vendors. I'm guessing it had to have been a bridge-less configuration, and do wonder if a bridged configuration (were that it was possible for any cards) would offer more performance. One day, explicit mGPU will be ubiqituous and all cards will have a universal bridge connector

Looks like a decent jump from your stock run. Throwing my 1080TI and 3080 1080p results out there. 1080TI / 2100MHz 12,000MHz @1080p: 3080 / 2070 MHz 21,000MHz @1080p: