• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

So, who's getting a 5XXX series card?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
I know. but when it worked! i was getting up to 180% FPS with 2 cards in crossfired 290x's in most games , when i could find a crossfire profile that worked or matched up to the game engine.
I think there all a bunch of quiters :p
 
And what you described is, sadly, a best case scenario. It was rare (think Dogman/Sasquatch/Nessie rare, lol) to see ~100%/~2x scaling.

Nothing like getting an average of 50%-60% scaling by paying 2x and using 2x the power, LOL! I mean, sometimes you just had to, I suppose, but yeah, just terrible scaling in most cases.

There has to be a better way! :p
 
And what you described is, sadly, a best case scenario. It was rare (think Dogman/Sasquatch/Nessie rare, lol) to see ~100%/~2x scaling.

Nothing like getting an average of 50%-60% scaling by paying 2x and using 2x the power, LOL! I mean, sometimes you just had to, I suppose, but yeah, just terrible scaling in most cases.

There has to be a better way! :p


LOL NOOO NOO NO that's just because they didn't put in enough effort to make it work properly :).
I actually feel that's partially true btw because lots of games that weren't working well id force it to use another game with a similar engines profile and it would be great.
Yea Bring it back and fix it so it works better :) what else are we gonna do with all these pcie5 slots and 1200+W power supplies?
 
Power PCIe RAM when it becomes the norm with your excess PSU power?

Multi-gpu is definitely a thing outside of video outputting. You can't remove the bottleneck of multi-gpu video outputting without changing the system architecture I think or you limit the FPS to 30/60 or something like that, but I doubt that would always work too. The bottleneck being time to display the image. The image can certainly be ready and rendered even with memory being diverged across multiple nodes (although maybe not anymore with current GPU pipelines), but when it comes to making that image ready for the video output buffer, well that requires near instant transmission from the remote node to the display node.

This generations GPU pipelines for gaming is super optimized to display as good of an image as possible in a certain timespan (ala Fake Frames). All three GPU companies have changed up the pipeline to rely heavily on cache and local ML branch decision making to optimize each frame presented to the display. I think this alone prevents multi-gpu setups for gaming. Intel's dual-gpu battlemage would be an interesting whitepaper if they did try something new.
 
"Galax Introduces RTX 5080 and 5070 Ti HOF GPUs Featuring Power Failure Status LEDs"

"Galax has introduced the RTX 5080 and RTX 5070 Ti Hall of Fame (HOF) Edition graphics cards, now available on their global website. These GPUs come with a built-in LED system designed to help users monitor the power connection status visually. The LED array, located in the center of the card, changes colors based on the power supply's condition. If the 16-pin power connector isn't fully or properly plugged in, the LEDs light up yellow. If there are power delivery problems despite a correct connection, the LEDs turn red. When the connection is good and stable, the lights either stay off or display a decorative color. This system aims to help prevent issues like melted connectors or hardware failures caused by bad power connections, a concern that has grown since the adoption of the 12VHPWR power standard."

 

"ASUS BTF 2.5 Connector on GeForce RTX 5090 Stays Cool Even at 1,900 W Load"

"In the first trial, the card drew approximately 670 W, roughly matching the RTX 5090's typical maximum, and the GC-HPWR connector stabilized between 30°C and 35°C over ten minutes. ASUS engineers rated the metal connector for up to 1,000 W continuous operation, and the results confirmed a comfortable safety margin for maximum load. Tony Yu even increased the draw to 1,300 W, yielding a peak connector temperature of 38°C. In a final extreme test, the load was set to 150 A, driving total consumption above 1,900 W. Even under these conditions, the metal connector held at about 41°C, while the power cables reached roughly 70°C, demonstrating the adapter's superior thermal performance. Yu's experiments also showed that the GC-HPWR and 16-pin connectors can share the power load. At a 200 A setting, each connector supplied around 1,200 W to 1,400 W from separate power lines. Although ASUS plans to ship retail RTX 5090 cards with a single connector, this test suggests a possible path for future ultra-high-power GPUs designed for extreme overclockers."
 
The only problem with this ASUS is that they have already changed the connector, so we can expect that they will change it again in 1-2 generations. In short, you buy a BTF series graphics card, and you don't know if you can use it when you change the motherboard in a few years. You are also limited to a few overpriced ASUS motherboards, which are unavailable in many countries. It's like you buy a PC and can't really change anything. When anything is slow, you have to replace most of the PC.

It once again shows that Nvidia connectors are a total failure, but we have known that for a long time already. Why can't they solve it and release a connector that will be available without changes for the next 10+ years? Standard 6-8 pins have been in use for so many years, and no one was complaining. 3-4 changes to one connector type and a flood of complaints is enough to make any respectable brand design something better.
 
Thay grey... can barely see the numbers. Mine js all.. rainbowy, lol. Didn't know you could change it.

Also, has it been updated? My cpu %, the right one/overclocking. Doesn't break 100% but shows it's hitting peak boost.
 
I hope that desktop market goes with slotted power for high-powered cards rather than cables. You can get so much more using PCB even though it costs similar.
 
Back