. @notRandomDude

@ashxhart @Alibaba_Qwen @Apple @digitalix So you are using 2 models on the dgxs and the m5u simultaneously. How do they interact with each other. They work on the same problem and talk to each other via the shared kv cache? That's so cool. Is this a better approach than to run one huge model on let's say bonded m5us?

0 0 0 78 0

View Details

. @notRandomDude

5 days ago

@MehdiHacks @JamesMcPherson Thank you Mehdi. Currently trying to build a setup for me and my roommate so I really sit on that fence. I do agree though that the m5u is probably the better long term invest and has more headroom to grow in terms of capabilities. Tough one

0 0 0 22 0

View Details

. @notRandomDude

5 days ago

@JamesMcPherson With the 64gb sparks now getting released I also think there is no going down in price now anymore for the 128gb model. So if I want the spark I really should get it now. This is difficult, but thank you again so much for your input!

0 0 0 10 0

View Details

. @notRandomDude

5 days ago

@JamesMcPherson Ah thanks for the detailed response James! Yeah it's a tough one. Ash Hart was leaning a bit towards the m5u because of the memory bandwidth and the untapped potential. Both of you them together could do some real damage I reckon.

1 0 0 21 0

View Details

. @notRandomDude

5 days ago

@ashxhart @Alibaba_Qwen @Apple @digitalix Ah yes I see you using the tensorfold glm 5.3 flash build. That's so cool. What's your speed on that? Kind of makes me want to get both. This is hard. But thank you again for the input.

1 0 0 150 0

View Details

. @notRandomDude

5 days ago

@ashxhart @Alibaba_Qwen @Apple @digitalix Thanks for your take! I also saw your your testing with MCDMA on a M5U connected to 2 Sparks. But as I can gather the bonded M5U seem to be a lot more promising. Any plans to still try to test more with the Sparks in the MCDMA context?

1 0 0 131 0

View Details

. @notRandomDude

2 weeks ago

@NVIDIAGeForce #RTXON

0 0 0 1 0

View Details

NVIDIA GeForce @NVIDIAGeForce

2 weeks ago

We're giving away a CONTROL Resonant custom wrapped NVIDIA GeForce RTX 5080 to celebrate its official release with #RTXON. To enter: 🟢 Share this post 🟢 Comment #RTXON T&Cs: nvda.ws/4z24Gcq

38K 31K 34K 3.0M 3K

View Details

. @notRandomDude

6 months ago

@Ren_aramb I agree with the pick and wanted to buy a small position today. But to be honest... This kind of quick rise scares me. If we are entering an environment that is less about research and more about hyping stocks that have some loose connections to the AI. That can go wrong quick

0 0 0 99 0

View Details

. @notRandomDude

6 months ago

@boring_invest Well said

0 0 0 69 0

View Details

. @notRandomDude

6 months ago

@Ren_aramb x.com/i/status/20466… Came across this post and the detail of research was incredible. Really strengthens my believe that this is the pick for glass core substrates