Workstation build

[temp_anon]

Budget (including currency): 6000

Country: USA

Games, programs or workloads that it will be used for: Python & RStats; Running regex searches; Data analysis on big data

Other details (existing parts lists, whether any peripherals are needed, what you're upgrading from, when you're going to buy, what resolution and refresh rate you want to play at, etc): 

 

I have a $6K budget from my school. I just want to max out on RAM and CPU power. I don't care for graphics or gaming. I am a data analyst and do basic charts. I also use python for some textual analysis. For a large set of texts, my program might run for a week.

[temp_anon]

I am only focused on the PC itself, and maybe I don't need to max out the budget.

 

Also I don't need to worry about peripherals, I have everything I need.

[brob]

Is your work multi-threaded?

[temp_anon]

1 minute ago, brob said:

Is your work multi-threaded?

I can parallelize the work for the python stuff. But mainly I used RStats with a whole lot of RAM

[filpo]

3 hours ago, temp_anon said:

Budget (including currency): 6000

Country: USA

Games, programs or workloads that it will be used for: Python & RStats; Running regex searches; Data analysis on big data

Other details (existing parts lists, whether any peripherals are needed, what you're upgrading from, when you're going to buy, what resolution and refresh rate you want to play at, etc): 

 

I have a $6K budget from my school. I just want to max out on RAM and CPU power. I don't care for graphics or gaming. I am a data analyst and do basic charts. I also use python for some textual analysis. For a large set of texts, my program might run for a week.

if you want the best cpu and RAM for you with no compromises then try this pc partpicker link https://pcpartpicker.com/list/MFtRcb

if you don't need too much ram (like only 128gb instead of 256gb) and you don't need 24 cores but 16 cores is better for you (at a MUCH better price) then go for this pc partpicker link https://pcpartpicker.com/list/YjcjTn
Don't listen to the second suggestion if you REALLY want to spend that money and want something that can do a lot of heavy data sets

[brob]

If you need more memory you consider a Threadripper build. 

 

PCPartPicker Part List

CPU: Intel Core i9-13900K 3 GHz 24-Core Processor  ($569.99 @ Amazon) 
CPU Cooler: ARCTIC Liquid Freezer II 360 56.3 CFM Liquid CPU Cooler  ($139.99 @ Amazon) 
Motherboard: Asus PRIME Z790-P WIFI D4 ATX LGA1700 Motherboard  ($239.99 @ Amazon) 
Memory: G.Skill Ripjaws V 128 GB (4 x 32 GB) DDR4-3200 CL16 Memory  ($289.99 @ Amazon) 
Storage: Western Digital Black SN850X 2 TB M.2-2280 PCIe 4.0 X4 NVME Solid State Drive  ($159.99 @ Amazon) 
Video Card: Asus TUF Gaming EVO OC GeForce GTX 1660 Ti 6 GB Video Card  ($219.99 @ B&H) 
Case: Antec P20CE ATX Mid Tower Case  ($114.99 @ Amazon) 
Power Supply: Corsair RM650x (2021) 650 W 80+ Gold Certified Fully Modular ATX Power Supply  ($104.99 @ Amazon) 
Total: $1839.92
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2023-03-19 16:32 EDT-0400

[brob]

8 minutes ago, Datanerdje said:

4 times 32gb ram is overkill unless your allocating the memory for other things. I dont see that back in his story. 

 

7 hours ago, temp_anon said:

I just want to max out on RAM and CPU power.

 

6 hours ago, temp_anon said:

I can parallelize the work for the python stuff. But mainly I used RStats with a whole lot of RAM

 

[GzeroD]

Answering this well really comes down to your definition of "Large data sets" and "lots of ram". Given what you've mentioned I would Imagine that 128GB is baseline requirement as just holding a few compiled Regular expressions in memory can be several GB of RAM on arrays that are only a couple hundred thousand entries.

With that in mind depending on the size of your expected datasets> You may easily end up on a prosumer platform because of the limitations of how much ram can be added.

 

Also are you sure no GPU or accelerators can be used to help crunch the data you're going through? Sometimes it shocks me to see how fast a CUDA accelerated task will destroy my R9 5950 working through the same problem with a non-cuda enabled version even though I'm just crunching through it with an old M40 I was able to get on eBay for less than $35 early last year. If you have any tasks that may be accelerated with an accelerator grabbing one from a few generations ago will often make sense even if you only use them once in a great while (generally opting for the largest VRAM options to enable you to do as many possible datasets in your price range.)

 

Before we can get into the nuts and bolts of a solid recommendation, if we don't know if your workload requires a workstation platform we can only give half hearted recommendations.

[SorryBella]

4 hours ago, GzeroD said:

Before we can get into the nuts and bolts of a solid recommendation, if we don't know if your workload requires a workstation platform we can only give half hearted recommendations.

I agree. 6 grand is a lot of money and it would be pretty unethical to just cobble the highest end product available for consumer, especially when we dont know how mission critical the data analysis is, how it scales under GP or even tensor cores if available, how big the data set is, is there any IT specialist on site, etc. Im following this thread to see OP's response on it.