Difference between revisions of "24-25 Finances"

From the Linux and Unix Users Group at Virginia Teck Wiki
Jump to: navigation, search
m
(Proposed Purchases)
 
(3 intermediate revisions by 3 users not shown)
Line 1: Line 1:
 
This page contains information, plans, and results of club budget requests and purchases for the 2024-2025 school year.
 
This page contains information, plans, and results of club budget requests and purchases for the 2024-2025 school year.
 
  
 
== Funding ==
 
== Funding ==
Line 37: Line 36:
 
** https://blacksburg.craigslist.org/sys/d/christiansburg-lot-of-dell-poweredge/7795184704.html
 
** https://blacksburg.craigslist.org/sys/d/christiansburg-lot-of-dell-poweredge/7795184704.html
  
 +
* $26.00 server protector (ppitzer)
 +
** We need to protect our servers from any errant passers-by to the server room. Ms. Miku has proven herself particularly adept at this task for VTARA, and I have confidence she will do the same for VTLUUG.
 +
** https://www.amazon.com/SEGA-Luminasta-Hatsune-Anniversary-Version/dp/B0C5PFT4K2
 +
 +
<b>Possible Setup 1 (this is just to test the waters) Ill have better ones</b>
 +
There will be many possible builds depending on what other hardware people want to use or have (technically I am sleeping rn but this build was on my mind)
 +
* $275 Intel® Core™ i9-10850K (daddykeystroke)
 +
** 10th gen is the min required to run intel a770. Never had an i9 before.
 +
** https://www.amazon.com/Intel-i9-10850K-Desktop-Processor-Unlocked/dp/B08DHRG2X9/133-1309834-0230441?pd_rd_w=PjByE&content-id=amzn1.sym.189ba0e8-243a-408f-947d-77c5ac846d2e&pf_rd_p=189ba0e8-243a-408f-947d-77c5ac846d2e&pf_rd_r=58Y2GVQZPJRME7DVFMGB&pd_rd_wg=cLdUJ&pd_rd_r=26b6fc3c-5074-4f3f-b75d-dad029f0d4c5&pd_rd_i=B08DHRG2X9&psc=1
 +
 +
* $179 ASUS Prime Z490-P LGA 1200 (Intel® 10th Gen) ATX Motherboard (daddykeystroke)
 +
** I am ASUS enjoyer (People say they changed with the logo I have not noticed quality drop myself). Cheap, effective, works with 10th gen intel. (ONLY 2 PCIE 16 slots) this is never enough and the p100 faces the wrong way if I remember right.
 +
** https://www.amazon.com/ASUS-Prime-Z490-P/dp/B07ZT3F95V
 +
 +
* $199 EVGA Supernova 1200 P3, 80 Plus Platinum 1200W
 +
** should easily be able to run both the p100 (250w) and the a770 (225w) and I'll likely upgrade my 3060 and 3090 and loan those to the cause also. (need better motherboard for this)
 +
** https://www.amazon.com/EVGA-Supernova-Platinum-Modular-220-P3-1200-X1/dp/B09GQM6CW3/133-1309834-0230441?pd_rd_w=LHZH3&content-id=amzn1.sym.189ba0e8-243a-408f-947d-77c5ac846d2e&pf_rd_p=189ba0e8-243a-408f-947d-77c5ac846d2e&pf_rd_r=PRXCZXN959H21VZ1MZ1C&pd_rd_wg=EIYpB&pd_rd_r=6324a15d-b9c3-41ad-ac8e-6be8d22d7869&pd_rd_i=B09GQM6CW3&th=1
 +
 +
 +
<b> API SERVICES </b>
 +
If we set up a rig we can get google one and openai accounts so we can set up our own agents to do activites for the benefit of society. This might be against everything we stand for.
 +
* $20/month rate limit Openai api
 +
** The comparable local model would be at minimum llama 3.2 11B (about as good as chatgpt 4o on everything but logically stepping through problems) (logic can be improved with prompting techniques consisting of long prompts since context length doesn't matter anymore). We need at least a 24gb of VRAM (we'd be using about 22gb of it a long prompt will push it over)
 +
** https://openai.com/api/pricing/
 +
 +
* $20/month Google One with Gemini subscription
 +
** We get 2TB of storage and access to the web app for Gemini Pro. Currently the api is free at 10req/min (not enough for things you'd use an api for)
 +
** https://one.google.com/about/?g1_landing_page=0
 
[[Category:Infrastructure]] [[Category:VTLUUG:Projects]]
 
[[Category:Infrastructure]] [[Category:VTLUUG:Projects]]

Latest revision as of 15:48, 31 October 2024

This page contains information, plans, and results of club budget requests and purchases for the 2024-2025 school year.

Funding

VTLUUG is currently eligible for:

  • $1,500 from the SBB for the year

We could in the future be eligible for:

  • $500 a semester, more with justification
    • if we become a CS Affiliated Org. This is theoretically possibly doable *right now*, during the Fall '24 semester. Theoretically.
  • $??? from the SEC
    • if we jump through all the required hoops. these seemingly are:
      • attendance to SEC ~3 meetings a semester
      • >=3 hours of community service via the SEC in a *previous* semester
      • signing up for some stuff

Proposed Purchases

For each of the following, use this format.

  • $PRICE <item name> (<person interested in buying it>)
    • reason for proposed purchase
    • link to a store page or quote with the price

these are just ideas, don't feel shy about putting something with a wacky price tag down

  • $716 8x14TB SAS Renewed HDDs (rsk)
    • we can use these to turn cyberdelia into a massive NAS (~84TB with RAIDZ2 and a minimal root filesystem), which can then be used to serve public package mirrors (e.g. all ~5TB of Debian mirrors) and provide general storage for members
    • https://www.amazon.com/gp/product/B0D5SP2487

Possible Setup 1 (this is just to test the waters) Ill have better ones There will be many possible builds depending on what other hardware people want to use or have (technically I am sleeping rn but this build was on my mind)

  • $179 ASUS Prime Z490-P LGA 1200 (Intel® 10th Gen) ATX Motherboard (daddykeystroke)
    • I am ASUS enjoyer (People say they changed with the logo I have not noticed quality drop myself). Cheap, effective, works with 10th gen intel. (ONLY 2 PCIE 16 slots) this is never enough and the p100 faces the wrong way if I remember right.
    • https://www.amazon.com/ASUS-Prime-Z490-P/dp/B07ZT3F95V


API SERVICES If we set up a rig we can get google one and openai accounts so we can set up our own agents to do activites for the benefit of society. This might be against everything we stand for.

  • $20/month rate limit Openai api
    • The comparable local model would be at minimum llama 3.2 11B (about as good as chatgpt 4o on everything but logically stepping through problems) (logic can be improved with prompting techniques consisting of long prompts since context length doesn't matter anymore). We need at least a 24gb of VRAM (we'd be using about 22gb of it a long prompt will push it over)
    • https://openai.com/api/pricing/
  • $20/month Google One with Gemini subscription