Jump to content

Budget (including currency): About 3000-4000$

Country: France

Games, programs or workloads that it will be used for: The purpose for this setup is to do Machine Learning research!

 

Hello there, My brother and I are planning to build a machine learning setup to do research and compete in AI competitions, but we need your help to choose the components.

We will be handling a lot of data so we would like this setup to have at least 256gb of ram (if we can expanded later, better), but the main issue are the cpu and motherboard. 

 

CPU: I personally was looking for the tops CPUs I know (Ryzen 3900X and i9 10900k) but they are limited to 128gb so I guess we will have to go for a server cpu,  but the decision here is simple, we want as many cores as possible, so we can do multiple tasks at the same time.

Motherboard: I do not really have much knowledge in order to choose the right one. but we want to be able to have as mentioned 256gb of ram (or more), slots for m.2 ssds, usb types dont really matter, I think usb 3.0 its ok, for the network card adapter, I dont think we will need a high speed network card like linus uses in their videos haha.

Graphics Card: I think the right choice would be a rtx2080 super or rtx2080Ti. 

 

Question: As far as I know, I think we can use a m.2 ssd to help our ram problem right?.   

 

For the other parts of the setup like power supply, Case, etc. we will like your recommendations and also thoughts on how we can manage to build this setup!

Thank you! 

 

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/
Share on other sites

Link to post
Share on other sites

I don't know much about this grade of hardware, so I cant recommend everything. But it seems you are looking for a thread ripper cpu (or xeon if intel is ahead in this sector of chips)

I am far from an expert in this so please correct me if I’m wrong.

Quote or tag me so I can see your response

 

PSU Tier List

Motherboard Tier List

Graphics Card Cooling Tier List

CPU Cooler Tier List

SSD Tier List

 

PARROT GANG

Mentioned in 7/10/20 WAN Show

Mentioned in 7/15/20 Techlinked

Mentioned in 7/17/20 Techlinked

Mentioned in 7/31/20 WAN Show

Mentioned in 7/31/20 Techlinked

Mentioned in 8/3/20 Techlinked

Mentioned twice in 8/5/20 Techlinked

Mentioned twice in 8/7/20 Techlinked

Mentioned in 8/12/20 Techlinked

Mentioned in 8/19/20 Techlinked

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780646
Share on other sites

Link to post
Share on other sites

You will probably want a Threadripper. Up to 64 cores, up to 1.5TB of ECC RAM Support (also supports non ECC), PCIe 4.0 for fast storage.

Main Rig:-

Ryzen 7 3800X | Asus ROG Strix X570-F Gaming | 16GB Team Group Dark Pro 3600Mhz | Corsair MP600 1TB PCIe Gen 4 | Sapphire 5700 XT Pulse | Corsair H115i Platinum | WD Black 1TB | WD Green 4TB | EVGA SuperNOVA G3 650W | Asus TUF GT501 | Samsung C27HG70 1440p 144hz HDR FreeSync 2 | Ubuntu 20.04.2 LTS |

 

Server:-

Intel NUC running Server 2019 + Synology DSM218+ with 2 x 4TB Toshiba NAS Ready HDDs (RAID0)

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780664
Share on other sites

Link to post
Share on other sites

2 minutes ago, camaron said:

Correct me if Im wrong,  the idea is to used the m.2 not as a storage drive but as a swap partition in the linux config so when the systems ran out of memory , it can use this partition to store temporal data. 

Unless they meant Optane?

Main Rig:-

Ryzen 7 3800X | Asus ROG Strix X570-F Gaming | 16GB Team Group Dark Pro 3600Mhz | Corsair MP600 1TB PCIe Gen 4 | Sapphire 5700 XT Pulse | Corsair H115i Platinum | WD Black 1TB | WD Green 4TB | EVGA SuperNOVA G3 650W | Asus TUF GT501 | Samsung C27HG70 1440p 144hz HDR FreeSync 2 | Ubuntu 20.04.2 LTS |

 

Server:-

Intel NUC running Server 2019 + Synology DSM218+ with 2 x 4TB Toshiba NAS Ready HDDs (RAID0)

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780679
Share on other sites

Link to post
Share on other sites

4 minutes ago, Master Disaster said:

Threadripper. Up to 64 cores, up to 1.5TB of ECC RAM Support

Isn’t the real practical limit 256 gigs ?

because it can’t handle registered ram like the Epyc CPUs

( please correct me if I’m wrong)

Edited by Drama Lama

Hi

 

Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler

hi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780681
Share on other sites

Link to post
Share on other sites

Just now, Drama Lama said:

Isn’t the real practical limit 256 gigs

Yeah but only because thats the physical limit to what you can buy ATM. 8 x 32GB is 256GB. The CPU can support up to 1.5TB.

Main Rig:-

Ryzen 7 3800X | Asus ROG Strix X570-F Gaming | 16GB Team Group Dark Pro 3600Mhz | Corsair MP600 1TB PCIe Gen 4 | Sapphire 5700 XT Pulse | Corsair H115i Platinum | WD Black 1TB | WD Green 4TB | EVGA SuperNOVA G3 650W | Asus TUF GT501 | Samsung C27HG70 1440p 144hz HDR FreeSync 2 | Ubuntu 20.04.2 LTS |

 

Server:-

Intel NUC running Server 2019 + Synology DSM218+ with 2 x 4TB Toshiba NAS Ready HDDs (RAID0)

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780688
Share on other sites

Link to post
Share on other sites

3 minutes ago, Drama Lama said:

Isn’t the real practical limit 256 gigs ?

because it can’t handle registered ram like the Epyc CPUs

( please correct me if I’m wrong)

Ohh now that I'm not 100% sure about. Registered memory belongs to a realm of computing I have zero experience in, I know very little about it.

Main Rig:-

Ryzen 7 3800X | Asus ROG Strix X570-F Gaming | 16GB Team Group Dark Pro 3600Mhz | Corsair MP600 1TB PCIe Gen 4 | Sapphire 5700 XT Pulse | Corsair H115i Platinum | WD Black 1TB | WD Green 4TB | EVGA SuperNOVA G3 650W | Asus TUF GT501 | Samsung C27HG70 1440p 144hz HDR FreeSync 2 | Ubuntu 20.04.2 LTS |

 

Server:-

Intel NUC running Server 2019 + Synology DSM218+ with 2 x 4TB Toshiba NAS Ready HDDs (RAID0)

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780699
Share on other sites

Link to post
Share on other sites

1 minute ago, Master Disaster said:

Ohh now that I'm not 100% sure about. Registered memory belongs to a realm of computing I have zero experience in, I know very little about it.

I think there are no large non registered ram modules ( as far as I know non reg. modules max 32 gb) ( and for reg. modules max 128 gb )

Hi

 

Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler

hi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780720
Share on other sites

Link to post
Share on other sites

Not sure what you want 256gb of RAM for but here

 

PCPartPicker Part List

CPU: AMD Threadripper 3970X 3.7 GHz 32-Core Processor  (€2077.92 @ Amazon France) 
CPU Cooler: Noctua NH-U14S TR4-SP3 82.52 CFM CPU Cooler  (€79.90 @ Amazon France) 
Motherboard: Gigabyte TRX40 AORUS PRO WIFI ATX sTRX4 Motherboard  (€408.40 @ Amazon France) 
Memory: Patriot Viper 4 Blackout 64 GB (2 x 32 GB) DDR4-3200 CL16 Memory  (€254.99 @ Amazon France) 
Memory: Patriot Viper 4 Blackout 64 GB (2 x 32 GB) DDR4-3200 CL16 Memory  (€254.99 @ Amazon France) 
Memory: Patriot Viper 4 Blackout 64 GB (2 x 32 GB) DDR4-3200 CL16 Memory  (€254.99 @ Amazon France) 
Memory: Patriot Viper 4 Blackout 64 GB (2 x 32 GB) DDR4-3200 CL16 Memory  (€254.99 @ Amazon France) 
Storage: Sabrent Rocket Q 2 TB M.2-2280 NVME Solid State Drive  (€249.99 @ Amazon France) 
Video Card: XFX Radeon RX 570 8 GB Video Card  (€159.00 @ Amazon France) 
Case: Deepcool MATREXX 50 ATX Mid Tower Case  (€58.10 @ Amazon France) 
Power Supply: Phanteks Revolt Pro 850 W 80+ Gold Certified Fully Modular ATX Power Supply  (€129.90 @ Amazon France) 
Total: €4183.17
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2020-07-02 23:29 CEST+0200

i5 12600KF | Zotac RTX 4080 Gaming trinity | Team Vulcan 2x16GB DDR4 3600 | ASRock Z690M-ITX/ac | WD Black SN850x 2TB

Cooler Master NR200P v2 | ID Cooling Zoomflow 280 XT | SeaSonic Focus SGX-750 | Thermalright 2x140mm + 2x120mm aRGB

LG C2 OLED 48" 120hz | Epomaker TH80 (Gateron Yellow) | Logitech MX Master 3 | Koss Porta Pro Comm

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780736
Share on other sites

Link to post
Share on other sites

34 minutes ago, camaron said:

Budget (including currency): About 3000-4000$

Country: France

Games, programs or workloads that it will be used for: The purpose for this setup is to do Machine Learning research!

 

Hello there, My brother and I are planning to build a machine learning setup to do research and compete in AI competitions, but we need your help to choose the components.

We will be handling a lot of data so we would like this setup to have at least 256gb of ram (if we can expanded later, better), but the main issue are the cpu and motherboard. 

 

CPU: I personally was looking for the tops CPUs I know (Ryzen 3900X and i9 10900k) but they are limited to 128gb so I guess we will have to go for a server cpu,  but the decision here is simple, we want as many cores as possible, so we can do multiple tasks at the same time.

Motherboard: I do not really have much knowledge in order to choose the right one. but we want to be able to have as mentioned 256gb of ram (or more), slots for m.2 ssds, usb types dont really matter, I think usb 3.0 its ok, for the network card adapter, I dont think we will need a high speed network card like linus uses in their videos haha.

Graphics Card: I think the right choice would be a rtx2080 super or rtx2080Ti. 

 

Question: As far as I know, I think we can use a m.2 ssd to help our ram problem right?.   

 

For the other parts of the setup like power supply, Case, etc. we will like your recommendations and also thoughts on how we can manage to build this setup!

Thank you! 

 

Here's a build that ticks all your needs:

 

PCPartPicker Part List

CPU: AMD Threadripper 3960X 3.8 GHz 24-Core Processor  (€1500.99 @ Amazon France) 
CPU Cooler: Noctua NH-U14S TR4-SP3 82.52 CFM CPU Cooler  (€79.90 @ Amazon France) 
Motherboard: ASRock TRX40 Creator ATX sTRX4 Motherboard  (€551.00 @ Amazon France) 
Memory: Corsair Vengeance LPX 256 GB (8 x 32 GB) DDR4-3200 CL16 Memory  (€1299.99 @ Corsair) 
Storage: Samsung 860 Evo 1 TB 2.5" Solid State Drive  (€151.48 @ Cdiscount) 
Storage: Crucial P1 1 TB M.2-2280 NVME Solid State Drive  (€130.72 @ Amazon France) 
Video Card: PNY GeForce RTX 2080 Ti 11 GB Blower Video Card  (€1153.30 @ Amazon France) 
Case: Fractal Design Define R5 ATX Mid Tower Case  (€132.88 @ Alternate) 
Power Supply: Corsair HX Platinum 1200 W 80+ Platinum Certified Fully Modular ATX Power Supply  (€275.00 @ Amazon France) 
Total: €5275.26
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2020-07-02 23:30 CEST+0200

 

It's way over your budget, as you can see. What are you planning to do with it? Kaggle competitions? If so, a way more sensible build would be:

 

PCPartPicker Part List

CPU: AMD Ryzen 7 3700X 3.6 GHz 8-Core Processor  (€310.00 @ Amazon France) 
CPU Cooler: Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler  (€64.54 @ Amazon France) 
Motherboard: Gigabyte X570 AORUS PRO ATX AM4 Motherboard  (€289.99 @ Amazon France) 
Memory: Kingston HyperX Fury 128 GB (4 x 32 GB) DDR4-3200 CL16 Memory  (€539.90 @ Amazon France) 
Storage: Samsung 860 Evo 1 TB 2.5" Solid State Drive  (€151.48 @ Cdiscount) 
Storage: Crucial P1 1 TB M.2-2280 NVME Solid State Drive  (€130.72 @ Amazon France) 
Video Card: PNY GeForce RTX 2080 Ti 11 GB Blower Video Card (2-Way SLI)  (€1153.30 @ Amazon France) 
Video Card: PNY GeForce RTX 2080 Ti 11 GB Blower Video Card (2-Way SLI)  (€1153.30 @ Amazon France) 
Case: Fractal Design Meshify C ATX Mid Tower Case  (€115.88 @ Alternate) 
Power Supply: Corsair RM (2019) 850 W 80+ Gold Certified Fully Modular ATX Power Supply  (€145.94 @ TopAchat) 
Total: €4055.05
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2020-07-02 23:37 CEST+0200

 

You want to max out on GPU to allow for quick iterations. 128gb should be enough for all competitions, and a 3700x should be more than enough since you'll be GPU (and VRAM) bound most of the time.

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
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780749
Share on other sites

Link to post
Share on other sites

51 minutes ago, Aereldor said:

Not sure what you want 256gb of RAM for but here

 

PCPartPicker Part List

CPU: AMD Threadripper 3970X 3.7 GHz 32-Core Processor  (€2077.92 @ Amazon France) 
CPU Cooler: Noctua NH-U14S TR4-SP3 82.52 CFM CPU Cooler  (€79.90 @ Amazon France) 
Motherboard: Gigabyte TRX40 AORUS PRO WIFI ATX sTRX4 Motherboard  (€408.40 @ Amazon France) 
Memory: Patriot Viper 4 Blackout 64 GB (2 x 32 GB) DDR4-3200 CL16 Memory  (€254.99 @ Amazon France) 
Memory: Patriot Viper 4 Blackout 64 GB (2 x 32 GB) DDR4-3200 CL16 Memory  (€254.99 @ Amazon France) 
Memory: Patriot Viper 4 Blackout 64 GB (2 x 32 GB) DDR4-3200 CL16 Memory  (€254.99 @ Amazon France) 
Memory: Patriot Viper 4 Blackout 64 GB (2 x 32 GB) DDR4-3200 CL16 Memory  (€254.99 @ Amazon France) 
Storage: Sabrent Rocket Q 2 TB M.2-2280 NVME Solid State Drive  (€249.99 @ Amazon France) 
Video Card: XFX Radeon RX 570 8 GB Video Card  (€159.00 @ Amazon France) 
Case: Deepcool MATREXX 50 ATX Mid Tower Case  (€58.10 @ Amazon France) 
Power Supply: Phanteks Revolt Pro 850 W 80+ Gold Certified Fully Modular ATX Power Supply  (€129.90 @ Amazon France) 
Total: €4183.17
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2020-07-02 23:29 CEST+0200

Thanks you bro!, I really appreciate it :).

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780901
Share on other sites

Link to post
Share on other sites

51 minutes ago, igormp said:

Here's a build that ticks all your needs:

 

PCPartPicker Part List

CPU: AMD Threadripper 3960X 3.8 GHz 24-Core Processor  (€1500.99 @ Amazon France) 
CPU Cooler: Noctua NH-U14S TR4-SP3 82.52 CFM CPU Cooler  (€79.90 @ Amazon France) 
Motherboard: ASRock TRX40 Creator ATX sTRX4 Motherboard  (€551.00 @ Amazon France) 
Memory: Corsair Vengeance LPX 256 GB (8 x 32 GB) DDR4-3200 CL16 Memory  (€1299.99 @ Corsair) 
Storage: Samsung 860 Evo 1 TB 2.5" Solid State Drive  (€151.48 @ Cdiscount) 
Storage: Crucial P1 1 TB M.2-2280 NVME Solid State Drive  (€130.72 @ Amazon France) 
Video Card: PNY GeForce RTX 2080 Ti 11 GB Blower Video Card  (€1153.30 @ Amazon France) 
Case: Fractal Design Define R5 ATX Mid Tower Case  (€132.88 @ Alternate) 
Power Supply: Corsair HX Platinum 1200 W 80+ Platinum Certified Fully Modular ATX Power Supply  (€275.00 @ Amazon France) 
Total: €5275.26
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2020-07-02 23:30 CEST+0200

 

It's way over your budget, as you can see. What are you planning to do with it? Kaggle competitions? If so, a way more sensible build would be:

 

PCPartPicker Part List

CPU: AMD Ryzen 7 3700X 3.6 GHz 8-Core Processor  (€310.00 @ Amazon France) 
CPU Cooler: Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler  (€64.54 @ Amazon France) 
Motherboard: Gigabyte X570 AORUS PRO ATX AM4 Motherboard  (€289.99 @ Amazon France) 
Memory: Kingston HyperX Fury 128 GB (4 x 32 GB) DDR4-3200 CL16 Memory  (€539.90 @ Amazon France) 
Storage: Samsung 860 Evo 1 TB 2.5" Solid State Drive  (€151.48 @ Cdiscount) 
Storage: Crucial P1 1 TB M.2-2280 NVME Solid State Drive  (€130.72 @ Amazon France) 
Video Card: PNY GeForce RTX 2080 Ti 11 GB Blower Video Card (2-Way SLI)  (€1153.30 @ Amazon France) 
Video Card: PNY GeForce RTX 2080 Ti 11 GB Blower Video Card (2-Way SLI)  (€1153.30 @ Amazon France) 
Case: Fractal Design Meshify C ATX Mid Tower Case  (€115.88 @ Alternate) 
Power Supply: Corsair RM (2019) 850 W 80+ Gold Certified Fully Modular ATX Power Supply  (€145.94 @ TopAchat) 
Total: €4055.05
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2020-07-02 23:37 CEST+0200

 

You want to max out on GPU to allow for quick iterations. 128gb should be enough for all competitions, and a 3700x should be more than enough since you'll be GPU (and VRAM) bound most of the time.

Thanks you bro!, I really appreciate it :).

It is not just for kaggle, we will do things like nlp research that has massive datasets, but the main goal is to try as many things as possible at the same time, also, I think because this set up is very expensive, we will go for 64gb of ram and then we will upgrade it later.

Link to comment
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13780913
Share on other sites

Link to post
Share on other sites

42 minutes ago, camaron said:

Thanks you bro!, I really appreciate it :).

It is not just for kaggle, we will do things like nlp research that has massive datasets, but the main goal is to try as many things as possible at the same time, also, I think because this set up is very expensive, we will go for 64gb of ram and then we will upgrade it later.

If you're not in a hurry, I'd recommend you to wait for the new releases this year. New AMD CPUs along with Ampere GPUs (which will have more tensor cores). Worst case scenario, current hardware gets a drop in price :)

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
https://linustechtips.com/topic/1217238-machine-learning-setup/#findComment-13781030
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

×