Budget 41-50k Home PC + ML config by next year

nac

Well-Known Member
Adept
Feb 5, 2008
469
67
116
If it's just few hundreds more as Chetansha says, it's okay. Already we're going over your budget.
 

sunily

Member
Disciple
Jan 1, 2016
42
2
8
My objective is to run machine learning programs down the line and I just read an article from Tim that GTX 580 is not suitable for that.


In initial days, card will be just used for getting graphics because with 2600/3600 I need a GPU. ML Programs will be run after 6 months or more. I am building this system as my old desktop is dead.
 

sunily

Member
Disciple
Jan 1, 2016
42
2
8
It is also confirmed by training institute as well that buy only 10X0 ti card
 

sunily

Member
Disciple
Jan 1, 2016
42
2
8
So, is below list final? Should I go and take local market rates as below price are online (theitdepot).

AMD Ryzen 5 3600 3.6GHz - 17000
Asrock B450 Pro4 AMD AM4 - 6500
Corsair Vegenance 8GB x 2 - 7000
Corsair CX 450W - 3500
LG 24 24MP59G - 10000
Zotac 1050 ti Mini 4gb (Resale) - 6300
Crucial 240GB SSD (optional) - 2400

Total - 52,700/-

Already have Keyboard, Mouse, 3 * 500 GB (internal), 1* 2TB (internal), DVD Writer
Let me know if PSU can take that load of all these or adding higher GPU later. Easy to add 1k now rather than selling CX450 and buying new CX550.
 

nac

Well-Known Member
Adept
Feb 5, 2008
469
67
116
450w is good enough upto 1660. If you're thinking about 2060, yeah go for 550w or 650w for 2070s.
 

sunily

Member
Disciple
Jan 1, 2016
42
2
8
450w is good enough upto 1660. If you're thinking about 2060, yeah go for 550w or 650w for 2070s.
Noted. I stick to 450W then for few years.

Dunn make this optional. Also if you have budget, get Samsung Evo 860. I believe Crucial you have listed in BX500 - entry level not so great SSD.
Noted. Crucial eliminated. Will consider Samsung Evo 860 240GB.

BTW how is internal M.2 SSD like mentioned here :


Does it free SATA port?
And connecting this chip is same as connecting disk?
Can we change boot order in case there are multiple disks and partition it?
 
Last edited:

nac

Well-Known Member
Adept
Feb 5, 2008
469
67
116
M.2 SATA SSD is same as SATA SSD wrt performance. If you want better, you can go for M.2 NVME SSD.

M.2 share lanes with PCIe and SATA. Pro 4 has two m.2 slots, one share lanes with SATA and one share lanes with PCIe.

With already 5 SATA devices you have, you can use PCIe lane shared M.2 slot.
 

b.life

Member
Disciple
Aug 21, 2019
21
7
6
My objective is to run machine learning programs down the line and I just read an article from Tim that GTX 580 is not suitable for that.


In initial days, card will be just used for getting graphics because with 2600/3600 I need a GPU. ML Programs will be run after 6 months or more. I am building this system as my old desktop is dead.
Hi sunily,
To be honest any graphics card can be used for machine learning, the problem is the available ram and the cores for iterating the data, so more ram and faster core and count makes the take quicker. If you are going to test the waters and not serious about machine learning then even a GTX 240 with tiny 1GB ram will be giving you some good results on small data sets(2 or 3 gb). Before 10/12 years machine learning was not this popular and developers ran them on CPU's(clusters) and it worked fine not the fastest but it did work, but now only the latest cards(with more ram and faster core) are recommended because it is faster and nothing more.

Any reason why only nvidia card ?
Hi chetansha,
The nvidia cards support CUDA a processing framework designed and developed by NVIDIA that has the most amount of user friendly libraries for faster implementation of GPU computations and people feel it easier to use it than other libraries. There is nothing preveting a end user from writing their own programs and running them on nvidia or amd or even a decade old VoodoFX cards but the problem is the time and effort required to develop these libraries and their efficiency. Nothing else.

Regards
B.Life
 

sunily

Member
Disciple
Jan 1, 2016
42
2
8
Hi sunily,
To be honest any graphics card can be used for machine learning, the problem is the available ram and the cores for iterating the data, so more ram and faster core and count makes the take quicker. If you are going to test the waters and not serious about machine learning then even a GTX 240 with tiny 1GB ram will be giving you some good results on small data sets(2 or 3 gb). Before 10/12 years machine learning was not this popular and developers ran them on CPU's(clusters) and it worked fine not the fastest but it did work, but now only the latest cards(with more ram and faster core) are recommended because it is faster and nothing more.

Regards
B.Life
Thanks b.life. So my decision to go with 1050ti 4GB is good for entry level ML programming. Even if we don't get serious about it then it is not much damage to pocket.
I have gone through some ML lessons and to be honest those are way beyond my scope as I am not into programming/algorithms/mathematics. Hence not sure I or my daughter can handle it.
It is just I wanted to go with industry trend as well along with her and get basic understanding about it.
 

b.life

Member
Disciple
Aug 21, 2019
21
7
6
Thanks b.life. So my decision to go with 1050ti 4GB is good for entry level ML programming. Even if we don't get serious about it then it is not much damage to pocket.
I have gone through some ML lessons and to be honest those are way beyond my scope as I am not into programming/algorithms/mathematics. Hence not sure I or my daughter can handle it.
It is just I wanted to go with industry trend as well along with her and get basic understanding about it.
I am sorry to hijack this thread but please dont loose hope on machine learning, To say the truth it is actually not that difficult to get started in it. Recently someone I know wanted to start in the machine learning and I collected some of these links to make the task easier. If you dont mind me saying, these days most of beginner tutorial on the web, in machine learning are not started with hello world equivalent but with those fancy multi-gigabyte fancy pants scare the hell away programs.

1. https://www.guru99.com/artificial-intelligence-tutorial.html
2. https://www.ibm.com/downloads/cas/GB8ZMQZ3 a good book on machine learning from IBM
3.https://towardsdatascience.com/simple-machine-learning-model-in-python-in-5-lines-of-code-fe03d72e78c6 very simple example, this site has some good tutorials
4.https://vas3k.com/blog/machine_learning/ very easy to follow tutorial and this blog has very interesting reads
5.https://machinelearningmastery.com/machine-learning-in-python-step-by-step/ another good tutorial and please read the comment section also
6.https://elitedatascience.com/machine-learning-projects-for-beginners good overall resource about machine learning

Once again I am sorry to have hijacked your thread. If you need any help let me know, I will do what I can, machine learning is a very enjoyable field and beginners getting turned off by fancy talk is not ok.

Regards
B.Life
 

sunily

Member
Disciple
Jan 1, 2016
42
2
8
I am sorry to hijack this thread but please dont loose hope on machine learning, To say the truth it is actually not that difficult to get started in it. Recently someone I know wanted to start in the machine learning and I collected some of these links to make the task easier. If you dont mind me saying, these days most of beginner tutorial on the web, in machine learning are not started with hello world equivalent but with those fancy multi-gigabyte fancy pants scare the hell away programs.

1. https://www.guru99.com/artificial-intelligence-tutorial.html
2. https://www.ibm.com/downloads/cas/GB8ZMQZ3 a good book on machine learning from IBM
3.https://towardsdatascience.com/simple-machine-learning-model-in-python-in-5-lines-of-code-fe03d72e78c6 very simple example, this site has some good tutorials
4.https://vas3k.com/blog/machine_learning/ very easy to follow tutorial and this blog has very interesting reads
5.https://machinelearningmastery.com/machine-learning-in-python-step-by-step/ another good tutorial and please read the comment section also
6.https://elitedatascience.com/machine-learning-projects-for-beginners good overall resource about machine learning

Once again I am sorry to have hijacked your thread. If you need any help let me know, I will do what I can, machine learning is a very enjoyable field and beginners getting turned off by fancy talk is not ok.

Regards
B.Life
Thanks again for motivation and providing links. I will pm you if I need more information.