GTX 1080s: Testing a Legend

I have 3 3090s, all in various gaming computers. I’m not sure what to do with them- the VRAM is only 24GB so doesn’t seem worth the power costs to run an open weight model. Any suggestions on how I can run them as some type of cluster? Is it worth exploring NVLink?

I'm still using a 1080ti in my main PC, though I don't play as many games as I used to. Control is the last title I picked up where I really felt like I was pushing the graphics capability of my rig.

I remember when I built my PC I was surprised when I'd come across a post of someone running ancient (at the time) chips like the i7-870, and now I'm realizing I am one of those people.

My 2700X and 1080ti do plenty of what I need these days. I meant to upgrade both a couple years ago but between job changes and then tariffs and whatnot it just never quite made sense to commit to the jump. I have a hard time imagining what an ideal upgrade would even be now with how expensive everything's gotten. I'm mostly hoping my machine continues to hold up for another couple years.

Funny enough, I have the 1080 and I tested qwen 3.5 9b on it today. I was positively surprised with the experience. It's no Opus but it did OK on the couple of queries and tasks I gave it. It's pretty impressive for a 10 year old gpu!

It's amazing how great the 10 series was. The 1060, a budget card, had the same amount of VRAM as the previous generations top tier card (6GB) with the 1080Ti having 11GB! the 5060 for example has only 8GB, the same as the GTX 1070 and 1080.

My old 1080 Ti now sits inside a 4U rackmount server. It's an EVGA SC2 HYBRID version and I am still amazed that the AIO cooler hasn't failed yet or leaked everywhere. One of the best things I've purchased in retrospect since it was my main GPU up until the middle of 2025, even though at the time in 2017 I felt the price of 1.5k NZD was extreme.

My 1080ti is my local inference machine now. With the release of the Bonsai 27B models I can run a genuine dense model with usable context. GOAT card.

Haha that’s amazing. I thought these things were practically worthless. Gave mine away for free on Reddit.

Those were great and 1070 was a great deal. Upgraded to a 3080 from 1070 last year, they are quite reasonably priced now.

Mine is still going strong, it can run Gemma 4 26b. It can still play all the games I care about at near max settings at 1440p at 60 FPS (like Elden Ring).#

My PC did coincidentally die today (not GPU related) so I think it's time for an upgrade, what do you recommend?

They still have great value in the used market. Swapped mine for a 3090, also used, for ~20% rebate.

I just threw together my old 1060 in a steamOS - adjacent box running cachyOS. It’s great for couch co-op platformers. The GPUs were overkill for most games then and can still run new indie 2d games now.

Crazy for me to still own two

What's legendary about the 1080? It's a smaller 980ti on a much smaller node.

I had a 1080 in my previous build, it was an OC model that went from 180 up to 250W and could perform about halfway between a stock 1080 and a 1080ti. That was a nice card but by late Covid it was really showing its age.

Haha that’s amazing. I thought these things were practically worthless. Gave mine away for free on Reddit.

My 1080ti is my local inference machine now. With the release of the Bonsai 27B models I can run a genuine dense model with usable context. GOAT card.

What exact runtime are you running Bonsai 27B with right now and what variants of the weights are you using?

I have a 1080ti currently not plugged in. More info please.

I'm still using a 1080ti in my main PC, though I don't play as many games as I used to. Control is the last title I picked up where I really felt like I was pushing the graphics capability of my rig.

I remember when I built my PC I was surprised when I'd come across a post of someone running ancient (at the time) chips like the i7-870, and now I'm realizing I am one of those people.

Mine is still going strong, it can run Gemma 4 26b. It can still play all the games I care about at near max settings at 1440p at 60 FPS (like Elden Ring).#

My PC did coincidentally die today (not GPU related) so I think it's time for an upgrade, what do you recommend?

Those were great and 1070 was a great deal. Upgraded to a 3080 from 1070 last year, they are quite reasonably priced now.

They still have great value in the used market. Swapped mine for a 3090, also used, for ~20% rebate.

Long may it continue. I think you know when you know. Diminishing returns are essentially immediate at todays mid hardware. So what was elite in 2021 is now superseded by todays mid in 2026. I just shifted to being behind the market lead. I experience the same perceived performance upgrade at maximum value. I've started doing this with phones as well. For what I do, it makes no difference.

I still run a 2014 Mac mini and a 2012 MacBook air

Ryzen 5000 cpus are extremely cheap right now. I bought a 5900xt (16 cores) to upgrade a server a few weeks ago for ~$250. If you already have DDR4 you could put together a system for pretty cheap if you need it soon.

The downside is that AM4 is basically EOL, so no upgrade path.

> so I think it's time for an upgrade, what do you recommend?

Without knowing what you'll use it for, probably a 9970x + RTX Pro 6000 would do the trick, YMMV.

The RAM crunch has been a quiet victory for those older cards. I have a 10 year old video card in my rig that had at the time a decent but not chart topping 8GB of RAM. Developers are still targeting 8GB because so many of the cards released today still have that. Since I'm still using 1080p the performance is usually acceptable too. Surprisingly the biggest problem is becoming the CPU, with the i5-3570k lacking a few of the later SSE instructions and being completely unable to run some titles.

To be fair, the 5060 with 16GB is not that much more expensive than the 5060 8GB.

As others said, you can run more than one at a time, but even singly, 24 GB can be very useful in a hybrid approach. For instance, I recently had to process hundreds of high-res blueprints. Used Fable for the main orchestration, but all of the OCR and other work I did with a small local model. Didn't waste expensive Fable tokens on simple grunt work.

You can layer models across multiple GPUs for inference. If you pooled it into one system, that would be 72gb of some pretty damn fast memory. Would be great for dense models in the 30-50b range.

Club 3090 will make installing llm easier, for 1 or more cards.

I have a similar plan for my 4090 once it's gaming days are over. I'll retire it into my homelab (I specifically massively overspecced the power supply in anticipation of this) and run whatever local AI models exist in 2028-2029 which can fit on it.

Crazy for me to still own two

I own four in my office. Daily driven, mostly with photo editing tasks but some Blender, too. Work well for us.

What's legendary about the 1080? It's a smaller 980ti on a much smaller node.

Less energy usage, more vram, faster in games (see https://www.pc-kombo.com/us/benchmark/games/gpu/compare?ids%...).

But it is even more about timing. Nvidia shifted to RTX afterwards, opting to go for raytracing and upscaling instead of making strictly faster cards. Which made the last GTX cards very competitive for regular usage, raising its status.

What exact runtime are you running Bonsai 27B with right now and what variants of the weights are you using?

I have a 1080ti currently not plugged in. More info please.

Plug it in, get the cuda drivers, run ollama or equivalent, and go from there. I use an 4GB Quadro card from the same generation and depending on the model can get faster than I can read output. It’s not as good as commercial models, but smaller Qwens and Gemmas work well. The Quadro came with a Dell refurb that cost about as much as the card would have by itself at the time.

(I’m doing this with passthrough on a FreeBSD host with bhyve to a Linux guest. Works great.)

I’d like a beefier card, but until prices come down, even older cards like this can do some interesting things if you give them enough time.

You can layer models across multiple GPUs for inference. If you pooled it into one system, that would be 72gb of some pretty damn fast memory. Would be great for dense models in the 30-50b range.

The downside is that AM4 is basically EOL, so no upgrade path.

> so I think it's time for an upgrade, what do you recommend?

Without knowing what you'll use it for, probably a 9970x + RTX Pro 6000 would do the trick, YMMV.

To be fair, the 5060 with 16GB is not that much more expensive than the 5060 8GB.

I still run a 2014 Mac mini and a 2012 MacBook air

Club 3090 will make installing llm easier, for 1 or more cards.

I own four in my office. Daily driven, mostly with photo editing tasks but some Blender, too. Work well for us.

(I’m doing this with passthrough on a FreeBSD host with bhyve to a Linux guest. Works great.)

I’d like a beefier card, but until prices come down, even older cards like this can do some interesting things if you give them enough time.

Less energy usage, more vram, faster in games (see https://www.pc-kombo.com/us/benchmark/games/gpu/compare?ids%...).

Something I can't fault nvidia for is long term vision. If you ignore all the non-gaming uses, I can't see significant gains to be had by keeping evolving rasterization based rendering, which presumably would hit diminishing returns, upgrades would be less exciting, and be easier for AMD and others to compete against them. The trouble is a variety of reasons has meant that long term is likely going to be over a decade from when it was introduced before it's really commonplace or part of the furniture for the majority of games.

Hacker Times

Hacker Times

GTX 1080s: Testing a Legend

Discussion

Discussion