Jump to content

First PC build, need advice on parts selected

Budget (including currency): $4000-$5000 (R60000 - R70000)

Country: South Africa

Games, programs or workloads that it will be used for: Machine Learning (Training deep learning models), Gaming is not essential, but ML and gaming PC parts go hand-in-hand. 

PC_Build.thumb.PNG.ee610b66cb63c61bfbc2783ee873565f.PNG 

 

The RTX 3090 (if I can eventually get one around MSRP) is essential for training large models, due to its 24gb of memory. So unless there are better GPUs with CUDA support (basically just NVIDIA GPU's) and large memory (above 16gb) at around the same price as the 3090, the 3090 is the core component.

 

Looking for advice on the best parts to complement the RTX 3090 without bottlenecking it or overspending on unnecessary parts.

 

Don't really know too much about overclocking yet, but it would be nice to future proof this ability to speed up certain components if necessary. 

 

P.S. I chose the 1000W Superflower PSU since it is cheaper than the other branded 750-850W PSU's  (corsair, cooler master, etc...) in my country. Is it still overkill to have a 1000W PSU for this build? I read something about running a PSU at less than 60% of its max wattage is not great for the PSU, looking for some confirmation regarding this statement.

 

Also, any advice on if all the components will fit well in side the case and what fan configuration I should use would be appreciated.

 

Thanks in advance.

 

Link to comment
Share on other sites

Link to post
Share on other sites

3 minutes ago, MikeFish12345 said:

Budget (including currency): $4000-$5000 (R60000 - R70000)

Country: South Africa

Games, programs or workloads that it will be used for: Machine Learning (Training deep learning models), Gaming is not essential, but ML and gaming PC parts go hand-in-hand. 

PC_Build.thumb.PNG.ee610b66cb63c61bfbc2783ee873565f.PNG 

 

The RTX 3090 (if I can eventually get one around MSRP) is essential for training large models, due to its 24gb of memory. So unless there are better GPUs with CUDA support (basically just NVIDIA GPU's) and large memory (above 16gb) at around the same price as the 3090, the 3090 is the core component.

 

Looking for advice on the best parts to complement the RTX 3090 without bottlenecking it or overspending on unnecessary parts.

 

Don't really know too much about overclocking yet, but it would be nice to future proof this ability to speed up certain components if necessary. 

 

P.S. I chose the 1000W Superflower PSU since it is cheaper than the other branded 750-850W PSU's  (corsair, cooler master, etc...) in my country. Is it still overkill to have a 1000W PSU for this build? I read something about running a PSU at less than 60% of its max wattage is not great for the PSU, looking for some confirmation regarding this statement.

 

Also, any advice on if all the components will fit well in side the case and what fan configuration I should use would be appreciated.

 

Thanks in advance.

 

swich the 970 evo plus for the 980 pro

10400f

16 gb 4000

1650 super

z490 msi 

p400a digital

 

 

games i play: krunker (sometimes), valorant, and csgo

 

Link to comment
Share on other sites

Link to post
Share on other sites

6 minutes ago, MikeFish12345 said:

P.S. I chose the 1000W Superflower PSU since it is cheaper than the other branded 750-850W PSU's  (corsair, cooler master, etc...) in my country. Is it still overkill to have a 1000W PSU for this build? I read something about running a PSU at less than 60% of its max wattage is not great for the PSU, looking for some confirmation regarding this statement

There should be really no problem with running a PSU at or below 60% when computers are at idle they are well below this anyway, and AFAIK it is just a less efficient at the lower power.

Link to comment
Share on other sites

Link to post
Share on other sites

12 minutes ago, MikeFish12345 said:

Also, any advice on if all the components will fit well in side the case and what fan configuration I should use would be appreciated.

You also shouldn't have any problem fitting all of the parts in, my only worry would be with the graphics card but that case supports 435mm and that graphics card is only 335mm so there is even extra space to have the front mount radiator.

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, MikeFish12345 said:

Budget (including currency): $4000-$5000 (R60000 - R70000)

Country: South Africa

Games, programs or workloads that it will be used for: Machine Learning (Training deep learning models), Gaming is not essential, but ML and gaming PC parts go hand-in-hand. 

PC_Build.thumb.PNG.ee610b66cb63c61bfbc2783ee873565f.PNG 

 

The RTX 3090 (if I can eventually get one around MSRP) is essential for training large models, due to its 24gb of memory. So unless there are better GPUs with CUDA support (basically just NVIDIA GPU's) and large memory (above 16gb) at around the same price as the 3090, the 3090 is the core component.

 

Looking for advice on the best parts to complement the RTX 3090 without bottlenecking it or overspending on unnecessary parts.

 

Don't really know too much about overclocking yet, but it would be nice to future proof this ability to speed up certain components if necessary. 

 

P.S. I chose the 1000W Superflower PSU since it is cheaper than the other branded 750-850W PSU's  (corsair, cooler master, etc...) in my country. Is it still overkill to have a 1000W PSU for this build? I read something about running a PSU at less than 60% of its max wattage is not great for the PSU, looking for some confirmation regarding this statement.

 

Also, any advice on if all the components will fit well in side the case and what fan configuration I should use would be appreciated.

 

Thanks in advance.

 

Go for 2x32gb sticks, or even 4x32gb at once. If you're serious about ML and want to make proper use of that 3090, 32gb of ram won't cut it.

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

1 hour ago, resolve_beta said:

swich the 970 evo plus for the 980 pro

Why?

geometry is hard
b550 > x570

Link to comment
Share on other sites

Link to post
Share on other sites

33 minutes ago, igormp said:

Go for 2x32gb sticks, or even 4x32gb at once. If you're serious about ML and want to make proper use of that 3090, 32gb of ram won't cut it.

Okay thanks, I will take that into consideration. I will start off with 1x32gb stick so I can add 3 more later if need be. Will the 1x32gb perform worse than having 2x16gb sticks, since I won't be running a dual channel configuration?

 

 

What speed and CAS latency should I go for when looking for RAM. Is the higher speeds that necessary?

Link to comment
Share on other sites

Link to post
Share on other sites

40 minutes ago, Downkey said:

Why?

based on what he working on he needs fast speeds and the price gap isnt much for his budget

10400f

16 gb 4000

1650 super

z490 msi 

p400a digital

 

 

games i play: krunker (sometimes), valorant, and csgo

 

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Pigeon_Milk said:

You also shouldn't have any problem fitting all of the parts in, my only worry would be with the graphics card but that case supports 435mm and that graphics card is only 335mm so there is even extra space to have the front mount radiator.

Thanks, I was planning on going for a top-mounted radiator so I won't be utilizing that space.

Link to comment
Share on other sites

Link to post
Share on other sites

Another question I have is regarding the GPU sag. Should I get a GPU support bracket for the case or will the motherboard and case be able to support the GPU just fine?

 

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, MikeFish12345 said:

Okay thanks, I will take that into consideration. I will start off with 1x32gb stick so I can add 3 more later if need be. Will the 1x32gb perform worse than having 2x16gb sticks, since I won't be running a dual channel configuration?

 

 

What speed and CAS latency should I go for when looking for RAM. Is the higher speeds that necessary?

AFAIK, it will perform worse for games.

 

For ML or any data wrangling, it doesn't matter. You can go for the cheapest 3200 or 3600MHz stick, it should be good enough and you won't notice a difference going further than that.

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

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

×