32 GB VRAM for less $1k sounds like a steal these days, and I’m sure it’s not getting cheaper any time soon.
Does anyone here use this GPU? Or any recent Arc Pros? I basically want someone to talk me out of driving to the nearest place that has it in stock and getting $1k poorer.
I’m going to be brutal with you. I spent a few thousand dollars on 176GB of AMD vram because I was happy with getting vram for cheap and I hate Nvidia. It works and its nice to be able to run bigger models at usable performance, but if you need serious concurrency or good support for diffusion, you NEED Nvidia. AMD(and likewise Intel) just doesn’t have the environment support for non-server GPUs. Again, coming from someone who’s using this shit daily.
If you understand this limitation, then yes those B70s are cool as are AMD Pro 9700 which might have slightly better support rn. You may consider nvidia V100s which are old and cheap. I always recommend people start with 3090s (as a general powerhouse) or a pair of 5060tis (for really hood llm support) though. It will make your life easy if you can afford the vram limitation
Thanks for your experience! I’m in a similar boat regarding NVIDIA - plus the budget …
At least in Europe the V100 are only available from China and with a huge markup.
Used 5090 for 3,5k, even a 3060 used is still at 250$ plus.
It’s crazy at the moment - I simply can’t afford self hosting LLMs which is a new thing to say :D
Wouldn’t using the Vulkan backend instead of ROCm help a ton with concurrency and diffusion, at a marginal (1-2%) performance loss?
Vulkan helps with speed. Must benchmarks prove that out. Concurrency is a mixed bag. You can get some with llama.cpp bit vllm is concurrency king.
Just a couple of weeks ago llama.cpp released tensor parallelism which helps, but its still a experimental feature.
Unfortunately, I don’t know of any diffusion runners that work in vulkan. If someone has expertise, let me know!
In my experience Vulkan actually drops performance ever so slightly, but it also improves compatibility (especially on the chips AMD advertises as “AI” then promptly forgets to support via ROCm, like the gfx1101/02/03 family, gfx1150, etc.), which is why I recommend it - as you said, and I’ve just noted, AMD is famous about doing as little in their AI toolkit for the average user as possible, leading to very limited support on consumer hardware.
Thank you! This is really helpful. 32 GB V100 or pair of 5060ti s looks very interesting, and about the same price. Does running multiple GPUs require any special hardware? I mean apart from the motherboard with 2+ PCIe x16 slots?
This is something I learned the hard way.
Consumer hardware is limited by multiple factors when it comes to PCIe connectivity.
Your graphics card might be a 16 lane card (referred to as “x16”), but sometimes, not all of them are used. Aforementioned 5060ti - I believe only uses x8. Some devices like graphics cards can use a physically smaller slot with an adapter for a loss in performance (a few frames in game play performance)
Similarly, your motherboard might have a x16 slot and another x16 at the bottom. That second slot might only function as x8 or even x4. Does this matter? Sort of. Inta-card communication aka peer to peer communication can affect affect performance and that can compound with multiple cards.
Even worse, some motherboards may have all sorts of connectivity but may have limitations like only 2 out of the bottom 4 slots, PCIe and m.2, can work at a time. ASK ME HOW I KNOW.
Your CPU controls PCIe. It has a hard cap in how many PCIe devices it can handle and what speed. AMD tends to be better here.
Enterprise gear suffers from none of this bs. Enterprise CPUs have a ton of PCIe lanes and enterprise motherboards usually match the physical size of their PCIe slots to their capacity and support full bifurcation*
PCIe lanes are used up by and consumable by m.2, MCIO, and occulink to name a few. That means that you can connect a graphics card to either one is those of you can figure out the wires and power**
Bifurcation is a motherboard feature that lets you split PCIe capacity, so a 16x slot can support two x8 devices. My motherboard lets me do this on just the main slot and in a strange x8x4x4 configuration. I have an MCIO adapter (google it) which plugs into the PCIe and gives me 3 PCIe adapters with those corresponding speeds.
it also has 2 m.2 slots which connect to the CPU. One is them, I use for a nvme ssd like a normal person. The other is an m.2 to PCIe adapter which gives me an x4 PCIe slot. For those keeping track, that’s 24 PCIe lanes so far. That’s the maximum my processor Intel 265k can handle
But wait! The motherboard also has a kind of PCIe router and that thing can handle 8 more lanes! So I use the bottom 2 PCIe lanes on my motherboard for 2 cards at x4 each. The thing that kills me is that there are more m.2 ports. But the mobo will not be able to use any more than 2 devices at once. AND even though that bottom PCIe slot is sized at x16, electrically, its x4.
Do your research (level1techs is great) and read the manuals to really understand this stuff before you buy
My mobo for reference ASUS: TUF GAMING Z890-PRO WIFI
How reliable is this setup for local inference? For instance how many tokens/sec?
I’m asking cause I’d guess sharing bandwith like that would have some cost in speed
I find llama.cpp with Vulkan EXTREMELY reliable. I can have it running for days at once without a problem. As far as tokens/sec that’s that’s a complicated question because it depends on model, quant, sepculative, kv quant, context length, and card distribution. Generally:
Models’ typical speeds at deep context for agentic use. Simple chats will be faster
that’s sick, thanks for sharing
Wow, I didn’t think you were running 176GB worth of GPUs on a consumer board! I don’t have an extra board, and my gaming PC that has 9070XT is not a good basis for multi GPU build - it has a cheap mATX motherboard with too few slots and lanes. So it’s going to be a new build. Used EPYC boards look interesting for that.
I wish I bought an epyc board last year instead of my rig. Would have been far fewer headaches and, with the price of RAM, I would have quintupled in value now!
It’s getting better, but yeah, I can’t run a lot of models.