Description

This is my first custom PC build and a pretty dramatic upgrade from a 7 year old Dell Studio XPS laptop that I had previous been using at home.

The main uses for this build are:

  • Data science - Statistical analysis of large data sets in R and Python

  • Image editing - Adobe Photoshop (mostly RAW files, some multi-image stitches), some Adobe Illustrator, and maybe some very minimal video editing

  • Gaming - Pretty casual, not a major focus of the build


Picks I was fortunate to have the use of work funds to purchase several of the components that would be directly related to the performance of data analysis (e.g. CPU, SSD, memory, etc.). In addition, I was under a bit of a time crunch to purchase a lot of these parts. Together, this lead to me airing a bit on the side of overkill for some things...

That being said, here was my rationale for certain parts

  • i7-5820K - Both R and Python have ways to take advantages of multiple cores and it seems like Photoshop can use pretty much as many cores as you give it. I could have gone with the i7-5960X, but the i7-5820K seemed like a good compromise between number of cores and single thread speed, not to mention a little less overkill

  • Corsair H110i GT - I was seriously considering the inexpensive Thermalite TRUE Spirit 140 given it's performance seemed to match the best air coolers and even approach that of several good liquid coolers. However, I ultimately settled on the H110i GT for several reasons. 1) I liked the idea of getting most of the weight of the CPU cooler off the motherboard. 2) Almost all air cooling solutions would block my motherboards 1st PCIe slot, which MSI recommended be used with a single graphics card. 3) Peer pressure - nearly all of the builds I was seeing similar to mine were using liquid cooling so I went along with the cool kids... I settled on the H110i over the Kraken X61 due to the price and nearly indistinguishable performance (depending on who you asked). I also picked the GT over the GTX at the suggestion that the GT part manufacturer might be somewhat more reliable than the GTX

  • MSI X99A SLI Plus - Ultimately chosen among several other equally impressive motherboards for it's very high reviews. Had all the specs I was looking for. Besides that, not much to this decision.

  • G.Skill Ripjaw 4, 32G DDR4-2400 kit - My motherboard is rated for DDR-2133, with the ability to overclock up to 3333. Wasn't sure how far I'd want to go with overclocking the memory, so I settled on 2400 instead of paying to go much higher. Went with a 4-stick 32G kit since I'd still have another 4 slots to upgrade to 64G if I wanted. R and Python both store objects in memory so having lots of it certainly wasn't going to hurt

  • Samsung 950 Pro 512 M.2 SSD - I was really interested to try out a NVMe SSD based on it's purported speed over traditional SSDs.

  • WD Black 3TB 7200 RPM - Just for storage

  • EVGA GTX 980 Ti Hybrid - Both R and Python can make use of CUDA and openCL based programming. This can really benefit analysis of high dimensional data, including machine learning, which I frequently use in my work. The benefit of a powerful GPU to image processing and gaming was a nice, but coincidental bonus

  • Fractal Design Define S - Style-wise, I was looking for a pretty minimalist looking case and I liked the simple look of the Define S. I also considered the Corsair 450D, 750D, and NZXT H440, but settled on the Define S for a couple reasons. 1) I knew I wasn't going to need an optical bay or many sized storage drives, so I liked that the Define S completely did away with a drive cage. 2) The size of the Define S was a compromise between the hugeness of the 750D and the somewhat restricted space of the 450D. Seemed good for airflow. 3) Really good price

  • EVGA 850W SuperNova G2 - This PSU is definitely overkill for what I have put together right now, but I thought that if I ever wanted to through in another GPU for SLI, this would be the minimum amount of power needed to make that a possibility.


The build

For the most part, everything went very smooth. The build took me a long time, but mostly because I took my time and not because of any major hang ups. I credit the relatively painless set up to all of the helpful info people have posted to this and other sites, including full length build videos from places like PCPartPicker and Newegg. Here are some individual thoughts in no particular order regarding the build though

  • H110i GT radiator position - I see a lot of people placing their radiators as intakes on the front of their cases, but I was pretty compelled by the reasons to set it as a top exhaust.

  • 980 Ti Hybrid radiator position - I would have liked to place this radiator as a rear exhaust, like many people seem to do. But combined with the top radiator position of the H110i GT, this radiator wouldn't fit as a back exhaust. EVGA recommends that the radiator be placed higher than the height of the graphics card to avoid eventual introduction of bubbles into the pump, so that left only one position left, the highest front intake. I actually didn't find any build examples that used this strategy.

  • Other fan placement - The front of the case should have room for 3x 140mm fans, but the combination of 1 radiator and 2x 140mm fans was quite tight. In fact I really had t push up on the radiator and the bottom screws clamp down on the housing of the mounting slots rather than the mounting slots themselves because of how high the radiator is mounted.

  • Cable management - I've heard this is a bit of an art, but I was fairly pleased with how the cabling turned out in my build. Though my palette for fine cabling is not that refined yet

  • EVGA 980 Ti Hybrid LED - At first, the LED lighting on my GPU didn't work. I wasn't super concerned with how the internal lighting of my build turned out, but it eventually annoyed me enough to try to figure out. In the end, it turned out that the power cable underneath the GPU shroud wasn't connected. Simply plugging it in did the trick

  • OS installation - Had a slight surprise with the choice between MBR and GPT formatting which , at the time of the build, I didn't know about. Eventually got past this, but don't actually know what I did to get windows to finally format the SSD correctly...

  • One slight unresolved issue is, occasionally, when I power the PC on, it will start up for a second, then everything will stop completely for another second, and then the PC will resume a normal startup. Some googling has suggested that I may have some of the case wires plugged into the motherboard incorrectly but I haven't looked in to it any further yet.


Performance

  • All speeds and temps seem to suggest that everything is installed properly.
  • I posted the Fire Strike 1.1. and UserBenchmark scores in the photos

Overall, I'm quite happy with how everything turned out. I had a lot of fun putting everything together and am very satisfied with the results!

Comments

  • 46 months ago
  • 4 points

the placement of the gpu rad's fan is making it exhaust toward the front of your case and the hot exhaust from the gpu is being pulled back in by the front intake fans. this in not a good setup. all of your fans on anyone side of the case should all be moving air in the same direction.

  • 46 months ago
  • 2 points

The intention was to have the fan on the radiator in pull so that the slot would still be acting as an intake into the case. If I have it the other way around, that was an assembly error on my part, so thanks for pointing it out.

Edit: Yup, you were right. Fixed!

  • 46 months ago
  • 1 point

anytime :) enjoy the new build.

  • 46 months ago
  • 2 points

Great part selection. Looks very nice. The CPU and GPU should work great for your needs.

  • 46 months ago
  • 1 point

Thanks!

  • 46 months ago
  • 1 point

You're welcome.

  • 46 months ago
  • 2 points

Dang, I really like how the braided tubes combine with the sleeve on the psu cables, very nice!

  • 46 months ago
  • 1 point

Thanks! That was a lucky accident since I really didn't spend much time thinking about how it would end up looking.

  • 46 months ago
  • 2 points

This is a really awesome build! :-)

  • 46 months ago
  • 2 points

Thanks!

  • 46 months ago
  • 2 points

Very nice build! +1

As pureblackfire mentioned change the fan on rad so all fans are intaking air on front panel

Other than that enjoy this beasty build!

  • 46 months ago
  • 1 point

Thanks!

  • 46 months ago
  • 2 points

"Gaming - Pretty casual, not a major focus of the build"

buys one of the most expensive gaming GPUs

  • 46 months ago
  • 1 point

Casual = 1 980 ti

Slightly formal = 4-way SLI with Titan Xs

  • 46 months ago
  • 1 point

Fancy = 10-way SLI-Fire with GTX R10 1090s

  • 46 months ago
  • 1 point

I wonder if a 10 way SLI PC even exists....

  • 46 months ago
  • 1 point

No, it isn't possible, even if there was a motherboard with 10 pci express slots.

  • 46 months ago
  • 1 point

It's funny to imagine a 5 foot long SLI bridge, though.

  • 46 months ago
  • 1 point

Hah well like I said I was interested in trying out GPU computing. It also helped that I didn't have to pay for that part.

  • 46 months ago
  • 2 points

Hey another Python dude!

What type of work are you currently doing?

Bioinformatics!?!

  • 46 months ago
  • 2 points

Yup, exactly! I work with a lot of Illumina-based sequencing data. Currently, my R usage is still higher than Python, but that may be changing.

  • 46 months ago
  • 1 point

Awesome!

I'm like, the exact opposite. I pretty much only use Python right now but I'm probably gonna move more into R.

How would you rate the SciPy stack relative to R? In what tasks do you think R would be a superior option?

  • 46 months ago
  • 2 points

If I'm being honest, I would say a lot of the reason I prefer R at the moment is simply because I learned it first. Whenever I need to do a new task, I always think about how I would do it in R and usually resort to Python if I can't do it otherwise or my R code is too slow, so it because a self reinforcing preference.

That being said, there are a couple things that I like about R over Python at the moment.

1) The "data frame" class in R really lends itself well to statistical analysis. I know there are equivalents in NumPy and pandas, but to me they often feel cumbersome compared to the fluidity of R data frames

2) Package management seems much simpler in R and, for the most part, all possible packages you could want are going to be hosted and well supported on CRAN and Bioconductor. Having to deal with multi leveled python modules that may or may not be preinstalled based on what distribution you use has always frustrated me a bit

3) The R studio IDE is really nice compared to Spyder or iPython notebooks. While I really like the markdown style of iPython (R studio also has a markdown language, but its a little less flexible), I like the ability to move back and forth between a console and script I'm editing

4) Hadley Wickham. He seems like a pretty large figure in the principles of computational data analysis and writes great packages for R. People have ported his work to Python, it's great to have the source.

Anyway, again, a lot of these things mainly apply because I'm most comfortable in R, but I'm sure you can make a compelling argument for Python for each of these points as well!

  • 46 months ago
  • 1 point

Thank you for your input.

I think I see where you are coming from with points 1 and 2. Always remember to document the version number of any extra Python packages that were used!!

Good luck on your work, and hopefully you won't run out of RAM!

  • 46 months ago
  • 2 points

That could be the mantra of a new religion; "May you always have enough RAM!"

  • 46 months ago
  • 2 points

Very nice build, part selection and pics. It makes my day when I see a nice build on this site that's not specifically for gaming. I'm contemplating building a similar machine for video editing. Enjoy!

  • 46 months ago
  • 1 point

Thanks! As someone already pointed out, it could be considered a gaming build in disguise, but it is true that the purpose of the build was indeed for things other than gaming. Video editing did cross my mind too when I was picking things out, but only for occasional GoPro footage.

  • 46 months ago
  • 1 point

Yeah, depending on the video software you could see much faster rending speeds w/ 1080p and 4K video with the 980 TI vs. something like a 970, especially if you add several effects to your clips.

  • 46 months ago
  • 2 points

What was the last image of? Solid build. +1

  • 46 months ago
  • 1 point

Just realizing that you posted it in performance section.

  • 46 months ago
  • 1 point

The of the two screenshots? The first is of 3DMarks Firestrike benchmarking and the second is UserBenchmark.com's benchmarking

  • 46 months ago
  • 1 point

Thanks!

  • 46 months ago
  • 1 point

Very, very nice!

  • 46 months ago
  • 1 point

How fast is that Samsung 950 Pro ?? Trying to justify getting one..

Very Nice Build..

Thanks for Sharing !!

  • 46 months ago
  • 1 point

I went from a $499 off the shelf ASUS laptop to a custom built I7 skylake and the Samsung 950 pro NVMe M.2 for my C:\ drive. I also have 100MB FIOS. I have no bottlenecks during normal usage, email, web, steam gaming, there is no waiting at all. Every click is answered immediately. Maybe it's the combo of top notch parts i have assembled, but I do believe the M.2 plays a big part in the fastest computer that I have ever used.

Grover

  • 46 months ago
  • 1 point

Check out my build.

  • 46 months ago
  • 1 point

Really fast! I have an 850 Pro in a laptop and the 950 Pro has about 3x the sequential read speed. However, these two SSDs are so fast, that it's really hard to perceive that speed boost outside of bench mark stats. Feels good though!

  • 42 months ago
  • 1 point

Looks great!! Check out my data science build at: https://pcpartpicker.com/user/tenuki/builds/#view=MPjcCJ parts should be here this weekend, will post pictures then. I jumped the gun on 'completed' lol. I'll probably be running it headless, to freak out all the gamers.. ;) CUDA is a great great thing.

[comment deleted]
  • 46 months ago
  • 1 point

Thanks! I wasn't expecting a lot in terms of style (I don't really have a great eye for that), but I was pleasantly surprised with how cool it ended up looking!