Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

phoronix.com

Post reply on HN
Print view View original

61–70 of 131 posts

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#61

Post by adastra22 »

shwaj wrote: post #50
adastra22 wrote: post #37

Earlier quoted context omitted.

It was many generations before vector operations were moved onto graphics chips.

I think they’re using “vector” in the linear algebra sense, e.g. multiplying a matrix and a vector produces a different vector. Not, as I assume you mean, vector graphics like SVG, and renderers like Skia.

Nope, I mean it in the first sense. That happened with the GeForce 256 in 1999, and shader registers (the first programmable vector math) were introduced with the GeForce 3 in 2001. Before that 3D graphics accelerators -- the term GPU had not yet been invented -- simply handled rasterization of triangles, and texture look-ups. Transformation & lighting was handled on the CPU.

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#62

Post by freeqaz »

bigwheels wrote: post #27
api wrote: post #17

A not-absurdly-priced card that can run big models (even quantized) would sell like crazy. Lots and lots of fast RAM is key.

How does LPDDR5 (This Xe3P) compare with GDDR7 (Nvidia's flagships) when it comes to inference performance? Local inference is an interesting proposition because today in real life, the NV H300 and AMD MI-300 clusters are operated by OpenAI and Anthropic in batching mode, which slows users down as they're forced to wait for enough similar sized queries to arrive. For local inference, no waiting is required - so you c…

I think the better comparison, for consumers, is how fast is LPDDR5 compared to the normal DDR5 attached to your CPU?

Or, to be more specific, what is the speed when your GPU is out of RAM and it's reading from main memory over the PCI-E bus?

PCI-E 5.0: 64GB/s @ 16x or 32GB/s @ 8x 2x 48GB (96GB) of DDR5 in an AM5 rig: ~50GB/s

Versus the ~300GB/s+ possible with a card like this, it's a lot faster for large 'dense' models. Yes, even an NVIDIA 3090 is ~900GB/s of bandwidth, but it's only 24GB, so even a card like this Xe3P is likely to 'win' because of the higher memory available.

Even if it's 1/3rd of the speed of an old NVIDIA card, it's still 6x+ the speed of what you can get in a desktop today.

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#63

Post by musicale »

makapuf wrote: post #29

Funny they still call them graphics cards when they're really... I dont know, matmul cards ? Tensor cards ? TPU ? Well that sums it up maybe, what those are are really CUDA cards.

> what those are are really CUDA cards

That don't run CUDA?

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#64

Post by musicale »

mft_ wrote: post #16

I have no idea of the likely price, but (IMO) this is the sort of disruption that Intel needs to aim at if it's going to make some sort of dent in this market. If they could release this for around the price of a 5090, it would be very interesting.

Intel made a dent in the consumer gaming market with Battlemage.

They made a dent in the HPC market / Top500 with intel MAX.

It will be interesting to see if they can make a dent in the AI inference market (presumably datacenter/enterprise).

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#65

Post by shwaj »

adastra22 wrote: post #61
shwaj wrote: post #50

Earlier quoted context omitted.

I think they’re using “vector” in the linear algebra sense, e.g. multiplying a matrix and a vector produces a different vector. Not, as I assume you mean, vector graphics like SVG, and renderers like Skia.

Nope, I mean it in the first sense. That happened with the GeForce 256 in 1999, and shader registers (the first programmable vector math) were introduced with the GeForce 3 in 2001. Before that 3D graphics accelerators -- the term GPU had not yet been invented -- simply handled rasterization of triangles, and texture look-ups. Transformation & lighting was handled on the CPU.

Oops, sorry I misread. That makes more sense in context.

Yeah, I remember all the hype about the first Nvidia chip that offloaded “T&L” from the CPU.

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#66

Post by pointyfence »

knowitnone3 wrote: post #3

Any business people here that can explain why companies announce products a year before their release? I can understand getting consumers excited but it also tells competitors what you are doing giving them time to make changes of their own. What's the advantage here?

It's more than a year. They're sampling this to customers in the second half of 2026. It's a 2027 launch at best.

Intel has practically nothing to show for an AI capex boom for the ages. I suspect that Intel is talking about it early for a shred of AI relevance.

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#67

Post by thenaturalist »

Mars008 wrote: post #60
thenaturalist wrote: post #24

Earlier quoted context omitted.

The technology, maybe - and if on local. The public co valuations of quickly depreciating chip hoarders selling expensive fever dreams to enterprises are gonna pop though. Spend 3-7 USD for 20 cents in return and 95% project failures rates for quarters on end aren't gonna go unnoticed on Wall St.

So far there is no 'plateau' in the nearest future. 'AI' as a science and its applications should develop further for the next several years. Models will get more efficient, but still the bigger the better. This is obvious. Even if models don't scale up well, they can be used collectively in parallel 'brainstorming'. This will still create demand for hardware. Stagnation is still possible in case of recession. In thi…

Your entire argument is whoefully ignoring the CapEx economics of all of this.

But that's the foundation.

And there is a plateau in real money spent on AI chips.

You're ignoring a whole group of economic and finance professionals as well as - if you're inclined to listen to their voices more - Sama calling it a bubble.

If not for AI spending, the US already would be in a recession.

So your argument might sound nice and practical from a purely scientific perspective or the narrow use case of AI coding support, but it's entirely detached from reality.

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#68

Post by schmorptron »

Neywiny wrote: post #20
schmorptron wrote: post #18

Earlier quoted context omitted.

Maybe not that low, but given it's using LPDDR5 instead of GDDR7, at least the ram should be a lot cheaper.

Certainly an interesting choice. Dramatically worse performance but dramatically larger only time will tell how it actually goes

Rumor has it (according to MLID, so no one knows whether it's accurate) that AMD is also looking to use regular LPDDR memory for some of it's lower end next gen GPUs to not have to contend with nvidia over limited and cartelled GDDR7 supply. Maybe they're going to increase parallel bandwidth to compensate it? Or have wholly different tricks up their sleeve.

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#69

Post by baq »

wmf wrote: post #45
baq wrote: post #23

Earlier quoted context omitted.

With this much ram don’t expect anything remotely affordable by civilians.

160 GB LPDDR5 is ~$1,200 retail so the card could be sold for $2,000. The price will depend on how desperate Intel is. Intel probably can't copy Nvidia's pricing.

It’d be a disaster for Intel if it sold for less than 3k, personally I think they’re aiming for break even at 5k a pop at least, and I wouldn’t be surprised to advertise 2x memory at half nvidia price, which would put it at ~15-20k? and a healthy margin which they need like oxygen now. Of course it’s all for naught if it doesn’t perform compute-wise.

Re: Intel Announces Inference-Optimized Xe3P Graphics Card with 160GB VRAM

#70

Post by bildung »

baq wrote: post #69
wmf wrote: post #45

Earlier quoted context omitted.

160 GB LPDDR5 is ~$1,200 retail so the card could be sold for $2,000. The price will depend on how desperate Intel is. Intel probably can't copy Nvidia's pricing.

It’d be a disaster for Intel if it sold for less than 3k, personally I think they’re aiming for break even at 5k a pop at least, and I wouldn’t be surprised to advertise 2x memory at half nvidia price, which would put it at ~15-20k? and a healthy margin which they need like oxygen now. Of course it’s all for naught if it doesn’t perform compute-wise.

I also think they have to be substantially cheaper than nvidia to have any chance, but the pro 6000 with 96G is already available at 7-8k - so half the price would have to be significantly below 4k.
Post reply on HN