Hi all,
It has been a while since I have been here, but in my last thread I was given some very good advice. Unfortunately for me, I didn't really take it to heart and bought a Dell R810. The R810 will be sold, as I am a recent college grad and it would not be financially responsible for me to use it. Not only that, but it is overkill for my needs.
Admitted mistake aside, I own a Lenovo S30. I want to upgrade it to make it into a versatile machine. In specific, I am hoping to do some machine learning on it. It will also be used for general development. Keep in mind that as a recent grad I have yet to figure out what to do in my free time, so my main goal is to make a well rounded system rather than something process specific.
My budget is around $200 currently, but will be expanded. I know that my list of ideas will be more than my budget, but I am just planning ahead.
Noise and energy shouldn't be an issue. I plan on using the 2011 v2 series, which I have heard is relatively power efficient.
As for special software, there really isn't any. I plan on running TensorFlow CPU initially (not budget for a gpu), but that is well supported. Considering that the system base is already here, I don't have much choice.
Current specs:
E5-2620 v2
2x4gb ddr3 ECC, UDIMMS, 1600
Generic 500gb hdd, 7200rpm
GT 710
Potential goals:
E5-2660 v2, E5-2670 v2, or E5-2680 v2. Is there much reason to spend the ~$80 more for the 2680 v2 over the 2660 v2? Will it really speed up my machine learning projects that much?
32gb ddr3 ECC, RDIMMS or 64gb ddr3 ECC, RDIMMS. Getting 64gb would cost roughly $100 more currently. I also may be able to get the 32gb (8x4gb) from the R810 I have, meaning that it would (could) cost even less. How important is the speed? Am I correct in thinking that RDIMMS will perform as well as UDIMMS? It is my understanding that UDIMMS cost more due to compatibility with non-server systems. As for memory compatibility, will pretty much any ddr3 ECC RDIMMS work? I am hesitant to buy without knowing, and my ram knowledge is lacking (feel free to point me at articles, I don't expect to be spoon fed!).
I will be getting an ssd for a boot drive at the very least. Any recommendations storage wise? While a PCI-E ssd would be ideal, I am am not sure it is financially feasible.
The video card will have to stay the same for a while. My main desktop (read: old gaming computer) has a GTX 670 4gb that I may use in the future, which is better than the GT 710 currently in my workstation (clearly). If there are relatively low cost recommendations for video cards, I am open to ideas (especially if it offsets costs elsewhere, such as not needing a better CPU). But a GTX 1080ti is far too expensive for me.
Any other recommendations?
It has been a while since I have been here, but in my last thread I was given some very good advice. Unfortunately for me, I didn't really take it to heart and bought a Dell R810. The R810 will be sold, as I am a recent college grad and it would not be financially responsible for me to use it. Not only that, but it is overkill for my needs.
Admitted mistake aside, I own a Lenovo S30. I want to upgrade it to make it into a versatile machine. In specific, I am hoping to do some machine learning on it. It will also be used for general development. Keep in mind that as a recent grad I have yet to figure out what to do in my free time, so my main goal is to make a well rounded system rather than something process specific.
My budget is around $200 currently, but will be expanded. I know that my list of ideas will be more than my budget, but I am just planning ahead.
Noise and energy shouldn't be an issue. I plan on using the 2011 v2 series, which I have heard is relatively power efficient.
As for special software, there really isn't any. I plan on running TensorFlow CPU initially (not budget for a gpu), but that is well supported. Considering that the system base is already here, I don't have much choice.
Current specs:
E5-2620 v2
2x4gb ddr3 ECC, UDIMMS, 1600
Generic 500gb hdd, 7200rpm
GT 710
Potential goals:
E5-2660 v2, E5-2670 v2, or E5-2680 v2. Is there much reason to spend the ~$80 more for the 2680 v2 over the 2660 v2? Will it really speed up my machine learning projects that much?
32gb ddr3 ECC, RDIMMS or 64gb ddr3 ECC, RDIMMS. Getting 64gb would cost roughly $100 more currently. I also may be able to get the 32gb (8x4gb) from the R810 I have, meaning that it would (could) cost even less. How important is the speed? Am I correct in thinking that RDIMMS will perform as well as UDIMMS? It is my understanding that UDIMMS cost more due to compatibility with non-server systems. As for memory compatibility, will pretty much any ddr3 ECC RDIMMS work? I am hesitant to buy without knowing, and my ram knowledge is lacking (feel free to point me at articles, I don't expect to be spoon fed!).
I will be getting an ssd for a boot drive at the very least. Any recommendations storage wise? While a PCI-E ssd would be ideal, I am am not sure it is financially feasible.
The video card will have to stay the same for a while. My main desktop (read: old gaming computer) has a GTX 670 4gb that I may use in the future, which is better than the GT 710 currently in my workstation (clearly). If there are relatively low cost recommendations for video cards, I am open to ideas (especially if it offsets costs elsewhere, such as not needing a better CPU). But a GTX 1080ti is far too expensive for me.
Any other recommendations?
