digital-garden-anabasis/pages/projects.feedfarm.md

# FEEEDFARM

## Modes of Operation
rendering X learning X mining

### 1/ [AI](projects.feedfarm.md#AI) _how?_
### 2/ [3D](projects.feedfarm.md#3D)] 
### 3/ [coins](projects.feedfarm.md#coins) _?_


## GeForce GTX 1080 Ti 
### Specs
Maximum Graphics Card Power (W) 250

#### Cores
CUDA Cores 3584  
Graphics Clock (MHz) 1480  
Processor Clock (MHz) 1582   

#### Memory
Standard Memory Config 11 GB GDDR5X
Memory Interface Width 352-bit
Memory Bandwidth (GB/sec) 11 Gbps

#### Supported Technologies
SLI, CUDA, 3D Vision, PhysX, NVIDIA G-SYNC™, GameStream, ShadowWorks, DirectX 12, Virtual Reality, Ansel


## Questions
### Interkonektivity via SLI
sli becomes NVlink 

##### https://en.wikipedia.org/wiki/Scalable_Link_Interface
>SLI allows two, three, or four graphics processing units (GPUs) to share the workload when rendering real-time 3D computer graphics.

>Not all motherboards with multiple PCI-Express x16 slots support SLI. 

##### https://www.gpumag.com/nvidia-sli-and-compatible-cards/

>**High-Bandwidth Bridge or SLI HB Bridge** (650 MHz Pixel Clock and 2GB/s Bandwidth) – This is the fastest bridge and is sold exclusively by Nvidia. It’s recommended for monitors up to 5K and surround. SLI HB Bridges are only available in 2-way configurations.

##### resume?
- [SLI is not needed for 3d renders](https://linustechtips.com/topic/896397-is-sli-really-needed-for-rendering/)

- [SLI and VR](https://www.quora.com/Is-SLI-worth-it-for-VR?share=1)

- [ SlLI in gamemachine](https://www.quora.com/Is-SLI-worth-it-in-a-gaming-machine)
- SLI and nvidia quadro?

### VRam share?
- is it possible?

### Machine Learning
- https://www.pugetsystems.com/labs/hpc/NVIDIA-GTX-1080Ti-Performance-for-Machine-Learning----as-Good-as-TitanX-913/

#### Hardware setup
- https://blog.slavv.com/the-1700-great-deep-learning-box-assembly-setup-and-benchmarks-148c5ebe6415 !!!
- https://medium.com/the-mission/how-to-build-the-perfect-deep-learning-computer-and-save-thousands-of-dollars-9ec3b2eb4ce2
- http://guanghan.info/blog/en/my-works/building-our-personal-deep-learning-rig-gtx-1080-ubuntu-16-04-cuda-8-0rc-cudnn-7-tensorflowmxnetcaffedarknet/
- https://www.servethehome.com/deeplearning10-the-8x-nvidia-gtx-1080-ti-gpu-monster-part-1/
- multi gpus
	- https://hackerfall.com/story/which-gpus-to-get-for-deep-learning
	- https://medium.com/analytics-vidhya/setting-up-a-multi-gpu-machine-and-testing-with-a-tensorflow-deep-learning-model-c35ad76603cf
- cluster networking
	- https://timdettmers.com/2014/09/21/how-to-build-and-use-a-multi-gpu-system-for-deep-learning/ 


## 3D

### Blender
#### sheepit
- cant manage plenty of stuff

## render farms
- https://www.videomaker.com/article/c01/18933-make-a-render-farm-out-of-old-computers-for-a-cost-effective-way-to-speed-up-rendering
- https://www.toolfarm.com/tutorial/in_depth_render_farms-2/

### davinci resolve
- [windows video tutorial](https://www.youtube.com/watch?v=Xernl1v01K4_)
- [linux video tutorial](https://www.youtube.com/watch?v=10cpBZ86vrc)
- [DaVinci Resolve Remote Render Server Complete Guide (Windows/Linux)](https://forum.level1techs.com/t/davinci-resolve-remote-render-server-complete-guide-windows-linux/138807)


## AI
#incubation https://www.makeuseof.com/4-unique-ways-to-get-datasets-for-your-machine-learning-project/

### spleeter
!!!! #incubation !!!!
EASYTOGO
### [incubation.openAI](incubation.openAI.md)
#### [jukebox](https://github.com/openai/jukebox)
- [ ] 16gb VRAM?
- https://github.com/openai/jukebox
- https://www.theverge.com/2020/4/30/21243038/openai-jukebox-model-raw-audio-lyrics-ai-generated-copyright
- https://github.com/openai/jukebox/issues/136
#### openAI gym
- https://medium.com/curiouscaloo/a-gpu-ready-docker-container-for-openai-gym-development-with-tensorflow-9be3d61504cb

### [tools.magenta](tools.magenta.md)

### tensorflow
- https://www.tensorflow.org/tutorials/generative/dcgan
- https://joelgrus.com/2016/05/23/fizz-buzz-in-tensorflow/

### EBsynth
- https://ebsynth.com/

### runwayML
- https://runwayml.com/

## coins
- Efektivni vs. efektni?
- bitcoin? [[ethereum]] ?

???????		???????
???????		
???????
???????
???????
???????
???????		???????

## Hardware
### motherbords
#### new
- 2 way sli? ? ? ?
	- https://www.alza.cz/msi-mpg-z390-gaming-pro-carbon-d5471041.htm ? 

#### aukro
- https://aukro.cz/asus-p8z68-v-socket-1155-ivy-bridge-sandy-bridge-ready-zaruka-6980209156
- https://aukro.cz/asus-crosshair-iv-formula-amd-890fx-rozjede-i-fx-procesory-6978083856
- **!!!** https://aukro.cz/asus-p9x79-deluxe-intel-engineering-sample-q8av-i7-2490x-3960x-6980118152 **!!!**
- **!!!!!** https://aukro.cz/zakladni-deska-cpu-ram-pouzita-plne-funkcni-6980112877 **!!!!!**

#### with quad SLI //VR solution 
- [reference](https://pcpartpicker.com/products/motherboard/#h=3,8&sort=price&l=4)
- [reference bez zdurazneni SLI](https://pcpartpicker.com/products/motherboard/#h=3,8&sort=price&L=4&m=8)

##### sorted by SLI capabilites
- 16/16/8/8
	- https://www.gigabyte.com/Motherboard/X399-DESIGNARE-EX-rev-10/sp#sp
		- CZ
			- 12000,-
	- https://www.asrock.com/MB/AMD/Fatal1ty%20X399%20Professional%20Gaming/index.asp#Specification
- 16/8/8/8 - socket LGA 2011
	- 
	- ` 4 PCI-E 3.0 x16  
2 PCI-E 2.0 x1 (in x4 slot),  
,  
\*\*\*4 PCI-E3.0x16 slots are running at 16/16/NA/8 or 16/8/8/8\*\*\*  
\*\*\*PCI-E slot #1 supports (x4) and slot #4 (x16) is disabled when Intel Core i7-5820K , i7-6800K or any processors with only 28 lanes is installed\*\*\*\*\*\*PCI-E slot #1 supports (x4) and slot #4 (x16) is disabled when Intel Core i7-5820K , i7-6800K or any processors with only 28 lanes is installed\*\*\* `
	-  https://www.senetic.cz/product/MBD-C7X99-OCE-O	
		- 8000,-
- 16/16/16/16 - socket 2099
	- https://www.asus.com/Commercial-Servers-Workstations/Pro-WS-X299-SAGE-II/	
		- https://www.bhphotovideo.com/c/product/1558809-REG/asus_pro_ws_x299_sage.html
			- 13000,-
	- https://www.asus.com/us/Motherboards/WS-X299-SAGE/
		-  https://www.zbozi.cz/hledani/?q=ASUS%20WS%20X299%20SAGE&razeni=nejlevnejsi
			- 12000,- 
	??? / !!!
	https://www.asrockrack.com/general/productdetail.asp?Model=X470D4U
	
### Summary
Speed of PCI connections is important. Speed of 8x seems like minimum for rendering and machine learning. PCI lanes need to offered by motherboard and also CPU. 

The pci slots may not operate full bandwith even in case they offer the right hardware slot. Also motheboard dependend. Asus seems to be greatly specific with their motherboards.