Thursday, October 22nd 2026
GALAX GeForce RTX 5080 HOF Extreme Comes with Dual 12V-2x6 Power Connectors
GALAX has unveiled the specifications for its GeForce RTX 5080 Hall Of Fame Extreme edition graphics card, the only RTX 5080 model featuring two 12V-2x6 power connectors. Interestingly, the card has a TGP of 400 W, and the official GALAX announcement states that the two power connectors are not required for simultaneous use. Instead, they are offered as alternatives to suit different PC builds and cable management preferences, so even if both connections are used, only one will be active. One standard connector is located on the top of the GPU, in the usual spot. The second 12V-2x6 power cable is positioned at the bottom of the card, concealed behind a GPU shroud near the PCIe x16 connector. This serves as an alternative power supply, allowing system builders to hide visible cables for a cleaner look.
For the GPU itself, GALAX has implemented a 20-phase power delivery system for the GPU and a 4-phase power delivery design for the GDDR7 memory, along with a digital PWM controller, eFuse, and HOF-branded inductors. All of this is mounted on a customized white PCB with gold plating on key connections. In OC Mode, which raises the TGP to 400 W, the GPU clock speed reaches 2,805 MHz. In regular Gaming Mode, the TGP is reduced to NVIDIA's reference 360 W, with clock speeds at 2,655 MHz. Since the 12V-2x6 connector is sufficient for these TGP designs, a second power source is unnecessary. GDDR7 memory is running at 30 Gbps on a 256-bit memory bus for a total of 960 GB/s memory bandwidth.
Currently, the GALAX GeForce RTX 5080 Extreme is only available in China, and it's unclear if it will be sold in Western markets. However, the card is likely to attract enthusiasts worldwide, though it is expected to come with a hefty price tag. The current fixed price in China is 19,999 Yuan, which is approximately 2,995 US Dollars at the time of writing.
Source:
via VideoCardz
For the GPU itself, GALAX has implemented a 20-phase power delivery system for the GPU and a 4-phase power delivery design for the GDDR7 memory, along with a digital PWM controller, eFuse, and HOF-branded inductors. All of this is mounted on a customized white PCB with gold plating on key connections. In OC Mode, which raises the TGP to 400 W, the GPU clock speed reaches 2,805 MHz. In regular Gaming Mode, the TGP is reduced to NVIDIA's reference 360 W, with clock speeds at 2,655 MHz. Since the 12V-2x6 connector is sufficient for these TGP designs, a second power source is unnecessary. GDDR7 memory is running at 30 Gbps on a 256-bit memory bus for a total of 960 GB/s memory bandwidth.
25 Comments on GALAX GeForce RTX 5080 HOF Extreme Comes with Dual 12V-2x6 Power Connectors
You get half the Vram and 69% the performance for 50% higher price that a 5090s MSRP!
Get it while you can! FOMO! FOMO! FOMO!About 3 weeks ago the 9070XT were still at the $799 price that most had been for months and months, but just recently they all went up to $899 at my local Micro Center. Guess it depends on where you are. What doesn't make any sense to me is that most of these cards on the shelf were probably there after already being purchased and having a $799 price on them, but now that AMD has increased their prices all the cards that were already paid for before the price increase have magically gone up in price. Same happens for the Nvidia cards too....consumer is the one taking it in the shorts at every possible turn.
But we live in the moronic timeline and it shows.
The GTX 970 had more relative performance to the flagship and that GPU had another 20-40% in the tank for OC.
What is the argument for it not being a midrange GPU outside of price?
The 5080 is unqiuely the weakest xx80 class GPU releative to the flagship Nvidia has ever produced. It doesn't beat the last gen flagship and only beats it's predecessor by 11%.
The 5080 is roughly half a 5090 core and VRAM wise. It's literally the middle. Historically Nvidia's mid-range sits between 50 - 76% of flagship performance. Either way you slice it, it's mid-range.
Mind you, the 5090 has quite a few disabled cores. If you compare the % of disabled cores to the 1080 Ti, it sits between the 5090 and RTX 6000 Pro. The only saving grace is that the die is rather large but even accounting for that it would seem Nvidia simply rebranded xx80 Ti to xx90 and upcharged. The real xx90 class GPU is the RTX 6000 pro but you've got to pay $17,000 - $22,000 for that now.
64 bit chips were usually display adapters
128 bit were the low end
192 bit *see below
256 bit were mid range
384 bit high end
512 bit rare/enthousiast
Nvidia pulled that trick with the 680,980,1080,2080,4080 and 5080 and amd with the 6000 series.Not even 66%, it's close to 61% if you push it with RT/PT.
Name are supposed to be helpful to customers indentifying a product but companies like to build up customer trust in a particular branding and then use that trust to extract more money.
The logic of resources versus the flagship is more solid, but the GB202 is such an anomaly that the argument feels… eh.
Anyway. They were even using hbm in consumer cards. Nvidia almost made it available with volta, amd did with fury and vega. Wide buses are not obsolete, they were phased out as a cost cutting measure. "Shrinkflation" in action. (some cited technical difficulties with gddr6/gddr6x but gddr7 works just fine obviously)
The 5080 is a medium sized die on an extremely mature 5nm node. Nvidia did this a lot, if they reuse a node, the second gen products bigger. kepler>maxwell, pascal>volta>turing but not with ada to blackwell, except for the 5090. GB203 is actually a little smaller, even more "mid range" than AD103.
Finally the latest rumors caused a 200$ spike on the 5080 in China. Even that extreme 3000$ 5080 is now a 3200$ 5080 probably.
It was never a 80 class card to begin with.
Making these kind of comparison should, in my opinion, always try to get as much of the picture as possible.
In regards to GB202 being an anomoly, it is larger than most but the 5090 has much more disabled cores relative to the flagship than most flagship GPUs as well. 1080 Ti for example have much more enabled cores of the full die. When you factor that in, it's not abnormal. It's probably evens out to around an xx80 Ti class GPU. Realistically the RTX 6000 pro is the xx90 class GPU this gen but like other GPU tiers, what you get has been shifted down by one. Nvidia would much rather sell the chips for $17,000.