Jump to content

Nvidia Tesla K80 Suddenly Very Cheap

z99o

I've recently discovered that a formerly very expensive card ($8000 in 2017!) for machine learning and neural network training in servers, the Nvidia Tesla K80, is now $350 on amazon and selling between $150 and $250 used on ebay. Is this an insane untapped deal? Some research into this card indicates that you need your own cooling solution. It also has no video output, so for gaming, would you be able to utilize this card? I'd imagine with how much computing power it has, it would run any modern game very well, (if it's possible). Has anyone tried this out or figured out a way to get this card working in a desktop, even for the original purpose of neural networks?

Link to comment
Share on other sites

Link to post
Share on other sites

3 minutes ago, Electronics Wizardy said:

Its relly not that fast or great of a deal. Its basically  gtx780ti class  card,nd missing many newer fetures, and gonna be. pain to use in a newer build. For most users there are better cards for the money.

reality strikes again!

Bethesda PC:   R7 3700X  -  Asrock B550 Extreme 4  -  Corsair Dominator Platinum RGB 16GB@3.6GHz -  Zotac AMP Extreme 1080TI -  Samsung 860 Evo 256GB  -  WD Blue 2TB SSD -  500DX  -  Stock cooling lul  -  Rm650x

CrumpleBox V3:  Xeon X5680  -  Asus X58 Sabertooth  -  DDr3 16GB@1.33Ghz  -  Gigabyte 1660s -  TT smart RGB 700W  -  

Cooler Master Storm Trooper  -  120GB Samsung 850 Pro   -  LTT Edition Chromax NH-D15 ?

 

CrumpleBox 3 ROTF: I5-6400  -  MSI B150m Mortar  -  16GB 2133Mhz Vengeance Pro RGB  -  Strix 1070Ti - GTX 1070 FE  -  Adata 128GB SSD  -  Fractal Design Define C  -  Gammaxx 400V2  -  Cooler Master silent pro gold 1000W

CrumpleBox 2: i7-7820x - MSI X299 Raider - 32GB Thermaltake Toughram 3.6Ghz - 2x Sapphire Nitro Fury - 128GB PCie Adata SSD - O11 Dynamic - EVGA CLC 360 - Corsair RM1000X

 

Perhiperals:  Gateway 900p60 monitor  -  Dell 1024x768@75  -  Logi. G403 Carbon  -  Logi. G502  -  SteSer. Arctis 5  -  SteSer. Rival 110 - Corsair Strafe RGB MK.2

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

29 minutes ago, Electronics Wizardy said:

Its relly not that fast or great of a deal. Its basically  gtx780ti class  card,nd missing many newer fetures, and gonna be. pain to use in a newer build. For most users there are better cards for the money.

Is it good for folding in anyway? I know the cards specialized for tasks besides gaming.

AMD blackout rig

 

cpu: ryzen 5 3600 @4.4ghz @1.35v

gpu: rx5700xt 2200mhz

ram: vengeance lpx c15 3200mhz

mobo: gigabyte b550 auros pro 

psu: cooler master mwe 650w

case: masterbox mbx520

fans:Noctua industrial 3000rpm x6

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

45 minutes ago, Letgomyleghoe said:

Is it good for folding in anyway? I know the cards specialized for tasks besides gaming.

Folding@home doesn't really support quadro/tesla cards because that is not what the majority of people will be using

Bethesda PC:   R7 3700X  -  Asrock B550 Extreme 4  -  Corsair Dominator Platinum RGB 16GB@3.6GHz -  Zotac AMP Extreme 1080TI -  Samsung 860 Evo 256GB  -  WD Blue 2TB SSD -  500DX  -  Stock cooling lul  -  Rm650x

CrumpleBox V3:  Xeon X5680  -  Asus X58 Sabertooth  -  DDr3 16GB@1.33Ghz  -  Gigabyte 1660s -  TT smart RGB 700W  -  

Cooler Master Storm Trooper  -  120GB Samsung 850 Pro   -  LTT Edition Chromax NH-D15 ?

 

CrumpleBox 3 ROTF: I5-6400  -  MSI B150m Mortar  -  16GB 2133Mhz Vengeance Pro RGB  -  Strix 1070Ti - GTX 1070 FE  -  Adata 128GB SSD  -  Fractal Design Define C  -  Gammaxx 400V2  -  Cooler Master silent pro gold 1000W

CrumpleBox 2: i7-7820x - MSI X299 Raider - 32GB Thermaltake Toughram 3.6Ghz - 2x Sapphire Nitro Fury - 128GB PCie Adata SSD - O11 Dynamic - EVGA CLC 360 - Corsair RM1000X

 

Perhiperals:  Gateway 900p60 monitor  -  Dell 1024x768@75  -  Logi. G403 Carbon  -  Logi. G502  -  SteSer. Arctis 5  -  SteSer. Rival 110 - Corsair Strafe RGB MK.2

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

23 minutes ago, GrockleTD said:

Folding@home doesn't really support quadro/tesla cards because that is not what the majority of people will be using

I know they support the k80, just don’t know if it would be worth it over a 590 or whatnot.

AMD blackout rig

 

cpu: ryzen 5 3600 @4.4ghz @1.35v

gpu: rx5700xt 2200mhz

ram: vengeance lpx c15 3200mhz

mobo: gigabyte b550 auros pro 

psu: cooler master mwe 650w

case: masterbox mbx520

fans:Noctua industrial 3000rpm x6

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

21 hours ago, z99o said:

I've recently discovered that a formerly very expensive card ($8000 in 2017!) for machine learning and neural network training in servers, the Nvidia Tesla K80, is now $350 on amazon and selling between $150 and $250 used on ebay. Is this an insane untapped deal? Some research into this card indicates that you need your own cooling solution. It also has no video output, so for gaming, would you be able to utilize this card? I'd imagine with how much computing power it has, it would run any modern game very well, (if it's possible). Has anyone tried this out or figured out a way to get this card working in a desktop, even for the original purpose of neural networks?

The K80 is really slow for modern standards. Only benefit it has is the power to do full FP64, which none of the non-100 chips have in later chips.

Even for ML purposes it's slow, since the sheer lack of raw power will make it run slower than a 1660, even tho it has way more vram.

19 hours ago, Letgomyleghoe said:

I know they support the k80, just don’t know if it would be worth it over a 590 or whatnot.

it won't. It will be less powerful and consume more power.

FX6300 @ 4.2GHz | Gigabyte GA-78LMT-USB3 R2 | Hyper 212x | 3x 8GB + 1x 4GB @ 1600MHz | Gigabyte 2060 Super | Corsair CX650M | LG 43UK6520PSA
ASUS X550LN | i5 4210u | 12GB
Lenovo N23 Yoga

Link to comment
Share on other sites

Link to post
Share on other sites

12 minutes ago, igormp said:

The K80 is really slow for modern standards. Only benefit it has is the power to do full FP64, which none of the non-100 chips have in later chips.

Even for ML purposes it's slow, since the sheer lack of raw power will make it run slower than a 1660, even tho it has way more vram.

Thank you this is what I was tryna figure out, thought it might be better for folding, but wasn’t quite sure.

AMD blackout rig

 

cpu: ryzen 5 3600 @4.4ghz @1.35v

gpu: rx5700xt 2200mhz

ram: vengeance lpx c15 3200mhz

mobo: gigabyte b550 auros pro 

psu: cooler master mwe 650w

case: masterbox mbx520

fans:Noctua industrial 3000rpm x6

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×