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

Intel Arc Pro B70 Review

pugetsystems.com

Post reply on HN
Print view View original

91–100 of 131 posts

Re: Intel Arc Pro B70 Review

#91

Post by ycui7 »

muyuu wrote: post #78
ycui7 wrote: post #76

Intel Arc B70 when released, can only produce 1/3 of the token of RTX PRO 4500. Well, it also cost 1/3 of RTX PRO 4500. It lacked software support the for the primary target application, running LLM. The officially supported vllm fork is 6 version behind mainline. It did not run the latest hot new open models on huggingface. Parallel two of B70 reduce token rate, not improve it. So, the software behind B70 is basical…

There are nonlinearities to exploit in that calculus. Given enough VRAM to host a larger model that you're targeting, just the size can push you past the usability threshold at a much better price.

Problem is the more B70 you have, the slower the inference it gets(due to terrible software atm). A single B70 is almost barely faster than CPU inference. If you have 4 B70, you might as well run interference on CPU and be faster with cheaper DDR5 instead of GDDR6.

Re: Intel Arc Pro B70 Review

#92

Post by ycui7 »

magicalhippo wrote: post #54
canpan wrote: post #31

Earlier quoted context omitted.

Just ran llama-bench at home with the similar priced AMD AI PRO R9700 32G. The phoronix numbers look extremely low? Probably I misunderstand their test bench. Anyway, here are some numbers. Maybe someone with access to a B70 can post a comparison. Tried to use the same model as the article: llama-bench -m gpt-oss-20b-Q8_0.gguf -ngl 999 -p 2048 -n 128 AMD R9700 pp2048=3867 tg128=175 And a bigger model, because testing…

For reference in case it's interesting to someone, a 5090 on Windows 11 with CUDA 13.1 | model | size | params | backend | ngl | test | t/s | | --------------------- | ---------: |--------: | -------- | --: |------: |----------------: | | gpt-oss 20B MXFP4 MoE | 11.27 GiB | 20.91 B | CUDA | 999 | pp2048 | 10179.12 ± 52.86 | | gpt-oss 20B MXFP4 MoE | 11.27 GiB | 20.91 B | CUDA | 999 | tg128 | 326.82 ± 7.82 | | qwen35…

You can get 120TPS (144 peak) with Qwen3.6-27B on RTX PRO 6000 with autoround when MTP enabled. It runs faster than sonnet api calls.

5090 gets maybe 100TPS with MTP

Re: Intel Arc Pro B70 Review

#93

Post by ycui7 »

Mindless2112 wrote: post #45
canpan wrote: post #31

Earlier quoted context omitted.

Just ran llama-bench at home with the similar priced AMD AI PRO R9700 32G. The phoronix numbers look extremely low? Probably I misunderstand their test bench. Anyway, here are some numbers. Maybe someone with access to a B70 can post a comparison. Tried to use the same model as the article: llama-bench -m gpt-oss-20b-Q8_0.gguf -ngl 999 -p 2048 -n 128 AMD R9700 pp2048=3867 tg128=175 And a bigger model, because testing…

As of b8966, it is still not great. | model | size | params | backend | ngl | test | t/s | | --------------------- | --------: | ------: | ------- | --: | -----: | -------------: | | gpt-oss 20B MXFP4 MoE | 11.27 GiB | 20.91 B | SYCL | 999 | pp2048 | 851.81 ± 6.50 | | gpt-oss 20B MXFP4 MoE | 11.27 GiB | 20.91 B | SYCL | 999 | tg128 | 42.05 ± 1.99 | | gpt-oss 20B MXFP4 MoE | 11.27 GiB | 20.91 B | Vulkan | 999 | pp2048…

At this speed, people end up paying more on electricity than api calls. (California electricity)

Re: Intel Arc Pro B70 Review

#94

Post by KronisLV »

lambda wrote: post #35
MostlyStable wrote: post #6

Is Intel still making GPUs? I have heard so many conflicting things about will they/won't they stay in the market.

What do you mean, are they still making GPUs? This is a discrete GPU that has just recently been released, and it's one of the most popular GPUs in its class at the moment, due to 32 GiB of RAM for under $1000, which makes it great for LLM inference.

> What do you mean, are they still making GPUs?

There was recent talk of them pulling back from the consumer segment, though obviously the leaks have also predicted Battlemage not being a thing so go figure: https://youtu.be/NYd2meJumyE?t=638 (timestamped)

That said, them not releasing a B770 in the consumer segment also sucks, since there are games and use cases that the B580 comes in a bit short for.

Re: Intel Arc Pro B70 Review

#95

Post by ycui7 »

muyuu wrote: post #78
ycui7 wrote: post #76

Intel Arc B70 when released, can only produce 1/3 of the token of RTX PRO 4500. Well, it also cost 1/3 of RTX PRO 4500. It lacked software support the for the primary target application, running LLM. The officially supported vllm fork is 6 version behind mainline. It did not run the latest hot new open models on huggingface. Parallel two of B70 reduce token rate, not improve it. So, the software behind B70 is basical…

There are nonlinearities to exploit in that calculus. Given enough VRAM to host a larger model that you're targeting, just the size can push you past the usability threshold at a much better price.

When you get 4 of these, the idle power alone is 120W. That is a lot of electricity if left on 24/7.

At that power consumption, you also end up being more expensive than API calls and many times slower. It starts to feel very stupid to run local interference.

If the client is very keen on privacy, then they can pay for the NVIDIA.

I end up returning my B70s, and bought RTX PRO 6000.

Re: Intel Arc Pro B70 Review

#96

Post by als0 »

dwoldrich wrote: post #65

Hi Intel, I'm itching to buy an Xe3P! Or, Nova Lake? Crescent Island? Celestial? Jaguar Shores? Whatever the hell you name it doesn't matter to me, I just want a workstation with one of them bad boys attached to 160GB of RAM for legit inference power! I've been saving my money not paying for Claude Code so I can run my own agentic coding setup at home on yours. Please don't charge too much for the workstation class c…

> Please don't charge too much for it

Intel wouldn’t decide to do this even to save their own life

Re: Intel Arc Pro B70 Review

#97

Post by icelancer »

Readerium wrote: post #82
2001zhaozhao wrote: post #43

There's a tradeoff between dense models and MoEs on memory usage vs. compute for the same quality. For example, Qwen3.5 27B and Qwen3.5 122B A10B have similar average performance across benchmarks. The 122B is much faster to run than the 27B (generates more tokens at the same compute). The 27B, on the other hand, uses ~4x less VRAM at low context lengths (less difference at high context lengths). Right now, different…

LLMs are memory bandwidth bound not compute bound.

This is only true for some parts of the time cost function.

Re: Intel Arc Pro B70 Review

#98

Post by ycui7 »

bigyabai wrote: post #11
speedgoose wrote: post #7

Time to first token is a very important performance metric, as I figured out using a Mac Studio M3 Ultra (that is quite slow on this aspect). But 32GB for a TDP of 230W is perhaps not super interesting. Especially because you probably want to have more than one card. It's a lot of heat. You could use the cards for heating up a building, but heatpumps exist.

A lot of the TDP is reserved for running the shader units at full-power. My RTX 3070 Ti only pulls ~110w of it's 320w running CUDA inference on Gemma 26b and E4B.

B70 idles at 30W, while RTX PRO 4500 idles at 9W (measured to be 5W at wall).

B70 runs at 1/3 token output rate of RTX PRO 4500 and consume 3X idle power when do nothing.

Re: Intel Arc Pro B70 Review

#99

Post by ycui7 »

unethical_ban wrote: post #26

It looks like, if one can afford it, the R9700 is worth the extra money. I read that Intel is getting out of the dGPU space, but then again, their iGPUs are really getting good. I can't understand why they'd give up the space when the AI market is so insane.

Exiting dGPU for gaming, but staying in the LLM world.

Re: Intel Arc Pro B70 Review

#100

Post by ZiiS »

XCSme wrote: post #17
ZiiS wrote: post #9

Earlier quoted context omitted.

The drivers often need per game optimisations these will be missing but I doubt Intel would nerf them, just rely on you not paying a lot for RAM the game won't use.

I actually meant it in a different way. I would get it for local AI stuff, but being able to game on it would be a huge plus, otherwise I would need two different machines.

Much as I want diversity; a 3090 would be a billion times better for games and can probably hold its own for a broader AI workload. Anything other then running highly quantised models that don't fit in 24GB with realativly small contexts.
Post reply on HN