TechAntics

Пікірлер

@krystofzdansky3102 Күн бұрын

can you show it with pc pls

@PedjaMigrantArhiva-mb9jb 5 күн бұрын

is there any way to change commands?

@rizkysenja4964 5 күн бұрын

What about 3070 ti 8gb?

@dxgraal749 6 күн бұрын

In my experience using this exact model, i feel that both sticks tend to stick to the x or y axis when playing around them. This could be unpleasant in precision games where you would be required to use small fine adjustments such as shooters or Rocket League (in my case). However, i still think it is a nice cheap choice if you intend to use it on games that do not require really fine adjustments. Great video! :)

@logunsrun5 6 күн бұрын

How do you run it?

@garythepigeon8600 8 күн бұрын

how do i do it with pc

@official_RANDOM_dev 9 күн бұрын

Acer Swift Go 14 (13th Gen i5-13500H | Intel Iris Xe) is better

@senpaiv6936 6 күн бұрын

Only as a workbench, not an all-rounder, a good gpu is a must these days and I am sure intel iris is not it

@notrninja 9 күн бұрын

So should i buy it over the hp victus 15 with rtx3050 4gb

@locvan6183 9 күн бұрын

Yes ,lenovo loq performance better because it has 6gbVram

@GoofBall-z5v 9 күн бұрын

Is it safe?

@doesthingswithcomputers 10 күн бұрын

You can set the layers to offload from gpu to other devices…

@calvinbeukes43 10 күн бұрын

I updated mine and then it inverted the screen and deleted all my games? Any fix for this

@alikzildag8274 5 күн бұрын

İ have the same problem.İ try all panel but doesnt work

@mondoburger9535 10 күн бұрын

will this work on arm macs?

@mytechantics 10 күн бұрын

Quick timestamps if you're in a hurry: 1:38 - Unboxing & build quality 3:06 - Connectivity options 3:37 - Keyboard & touchpad 5:15 - Lenovo Vantage 6:41 - The display & gaming experience 8:30 - Speakers & fans loudness test 9:26 - Cinebench 2024 benchmark tests 10:24 - AI image generation test with Stable Diffusion XL 11:01 - The battery life 11:26 - Final remarks Enjoy!

@RecapHistory-740 11 күн бұрын

For me not show AutoAWQ in the list.

@indiraharron 11 күн бұрын

very easy to set up? brother this shit is complicated as hell

@subarashiikylox 8 күн бұрын

@flyingbusa3407 2 күн бұрын

But it's free

@indiraharron 2 күн бұрын

@@flyingbusa3407 it isn’t that good, i got it

@MuhammadDanyalKhan 11 күн бұрын

Can we add custom emotion?

@yamiverse9442 11 күн бұрын

hello nice video ! i have elitebook 8470 hp does it work on it or not someone help a brother here !

@Batnano 13 күн бұрын

LOL this guy thinks 6gb VRAM is a normal thing to have

@Luis-sq9gu 14 күн бұрын

finally I can get into the Darkest Fear trilogy, Resident Evil Uprising, and the Silent Hill point and click games :D

@Tudor715 16 күн бұрын

bro is so much lag

@JarppaGuru 16 күн бұрын

what eva you run next year new model is better to take your money bcoz second is so long time

@zhenia8 17 күн бұрын

can it be stretched to full screen?

@drinerqc7434 17 күн бұрын

Rtx 6000 ada is rly good to

@ALEXHANS1383 17 күн бұрын

Another video which could have been a blog post.

@mytechantics 16 күн бұрын

It actually is, and the post features some more recent updated info, here: techtactician.com/best-gpu-for-local-llm-ai-this-year/

@patrickflanigan3903 18 күн бұрын

Currently using the AMD 7900 XTX for the meta-llama/Meta-Llama-3.1-8B-Instruct.

@CaimAstraea 18 күн бұрын

what about the rtx titan

@Graybeard_ 19 күн бұрын

4:54 you said the RTX 4070Ti was 12gb vram, but it is 16gb. What about AI Accelerator cards? Asus makes one that goes in a pcie slot. Will these pair/work together with the GPU cards to increase productivity/workflow?

@Olds79Starfire 19 күн бұрын

No it's not. The Ti super has 16gb

@Graybeard_ 19 күн бұрын

@@Olds79Starfire He was mousing over the TI super when he said 12gb. You cab see 16gb in the listing he is mousing over which is why I put a timestamp in my comment. Try it.

@notapro_ballsdex 20 күн бұрын

Hey, what's the file or .exe to use to start the program ? I already install it and i can't find it on my pc and using the start_http file again doesn't seem to work

@MeteoricPerson 21 күн бұрын

it booted up the first time but other times it doesen't. any fix?

@nortonwedge 21 күн бұрын

I snagged an open box 3090 from amazon for $1300 CAD (usually2500-3k), paired it with a 7900x3D for $500 (usually $700), with 64GB 6400 ram. Been running various image and video models on comfyui, works pretty well running large batches.

@oj397 21 күн бұрын

What was the first game?

@mytechantics 20 күн бұрын

The Sims 3 mobile

@olican101 21 күн бұрын

What if I don't have a max sequence length?

@EpicNukeCod 22 күн бұрын

im your 1000th subscriber

@bougiri 23 күн бұрын

4060 ti 16 gb vram?

@Nnamdxxx 23 күн бұрын

I have a 3060 12 gig and have had fun with ComfyUI but when you get into Videos you gonna need more VRAM.

@miroslavmajer5155 23 күн бұрын

Or buy Macbook - it's actually cheaper for LLM comparing to build PC for LLM. Mac have sharable ram between CPU and GPU. So, in case od MacbookPro with 128GB RAM, you can actually have GPU with 128GB RAM

@MrDoBo95 24 күн бұрын

Idk.. I'm running big llama 3.1 8b q8, qwen2.5, mistral nemo 12b q4km, even Gemma 27 fully as a try, which didn't go in usable speed but worked fine with my 3200 ddr4 ram in addition to 12gb vram, Llama vision llms and homey for my Hass for detection, flux schnell with text encoding in comfyui And imagine: it all runs on my 6700xt, which to this day did not make a single error or fail. The only little thing to do with the 6000 gen is to add the HSA override 10.3.0 as it's not recognised automatically. It's not that hard to add --HSA_OVERRIDE_GFX_VERSION=10.3.0 in a command line. I upgraded from a 6600 as 8gb just wasn't quite it, especially for the image creation but the 6700xt is a dream. I'm either getting a second one or go with a 7900gre additionally with 20gbs of ram. Stop creating dumb shit with bullshit claims about amd. The mi300x flagship is a lot better than the NVIDIA h100, they just became the standard too early.

@thebrainfan 27 күн бұрын

Great video, unfortunately I've already bought a 3080 with only 10GB VRAM.

@glenswada 26 күн бұрын

Doesnt really matter. I have slower 8GB card (aka 4060) and runs 8B great (responses in 1-2 seconds). The larger 70B will not fit into vram so will spill into local ram anyways. I had to upgrade from 32 to 64Gb ram to get 70B to work. But response times for 70B is something like 2 to 8 minutes. Not even a 4090 will run these larger models at a speed that will make people happy. So best to use 8B mostly. Only use 70B when you want a more detailed answer. Be nice to run 405b but it will not perform on cunsumer based gpu's.

@liora2k 28 күн бұрын

Good video , thanks for putting this together. I have some comments around AMD GPU, everything can run on them , you need to know how to …

@unicornhorn1003 Ай бұрын

hi can you do a video on how to download v2? IM HAVING SO MUCH TROUBLE WITH THE GIT CLONING IDEK HOW TO DO IT PLEASE

@LemmingwayArk Ай бұрын

While downloading Dolphin, there is one file that is still hanging on 0% for a while now. This is so annoying. How can I have WebUI download all the dolphin files at once, without one hanging on 0%?

@sircrashalot1815 Ай бұрын

Does this work on iOS? I am trying to usb tether as I don’t have a wifi adapter.

@imqqmi Ай бұрын

Writing new characters is so much fun! Llama 3.1 14b Q3 gguf is a pretty good LLM for 12GB 4070ti card. It really plays along well with the written character. Enabling flash attention and tensor cores speeds it up a lot too. Haven't had any luck training a lora though, giving errors after applying it to the model.

@BogdanTestsSoftware Ай бұрын

All the content on YT presents only Nvidia this and Nvidia that. I would like first a _feature availability / capability_ comparison between AMD Radeon vs Intel Arc vs Nvidia. I only include Nvidia to be able to relate to all other comparisons I´ve seen. I wouldn´t buy even a used Nvidia other than it really comes out as a steal (I mean a cheap deal, not actually stolen goods!!). I find Nvidia licensing & software practices as being toxic *for myself and for the market*. Now once capabilities are clear and set, I would like to know how performance stacks up for each. I.E.: is Radeon 2x (twice) as slow or 10x?? If performance is around 2-3x slower for Radeon or Intel, I would find the tradeoff quite acceptable. 5-10x would be a bit harder tradeoff, but I would still avoid Nvidia, the best I can. Nvidia is at "kill it with fire, before it spreads" toxicity and infectiousness levels, just my $0.02.