Would this laptop be futureproof for bioinformatics work

Question

[johndoe]

Hi, Would want to buy a laptop under 600 Euros 700 Dollars, that would be somewhat futureproof.

Will be doing some alignment work and multiplex amplicon based sequencing analysis. Not full genome analysis and for that I would use a workstation/server/cloud based solution. Want to be able to test pipelines and scripts.

Currently looking at either a Dell Inspiron 14 with i3 (dual-core) processor or i5 (quad-core) processor all other specs are the same the price difference is 419 to 569 Euro.

DELL Inspiron 14 3480/3481 Notebook i5-82650U/i3-7020U
8GB DDR4-2666 SO-DIMM RAM (would add another 8GB to second slot) (16GB Total)
DELL Inspiron 14 3480/3481 Notebook
512 GB SSD PCIe NVMe
2 x USB 3.0
1 x USB 2.0

Would I be losing alot of function with the slower dual core?

If you guys have another idea, I would be open to it. Like I said I have that budget and would like to get to 16 GB RAM, for alignment work.

Thanks for the advice in advance.

EDIT::::: Dell says the max RAM would be 16 GB but some Websites claim that 32 GB would be possible. Any info about this, if I could upgrade to 32 GB that would be a reason to buy the Laptop, otherwise hearing that 32 GB or at least that possibility would "futureproof" the laptop has got me thinking. ANYONE have an idea if 32 GB will work and if there are any cons to doing that.

[GenoMax]

Want to be able to test pipelines and scripts.

For that base version should be fine as well. Upgrade the memory now because you are unlikely to do that in future (personal experience). Technology keeps getting better and cheaper over time. You then loath to spend money on upgrades.

Getting i5 version for more money makes sense. A quad-core CPU will allow you to do more multi-tasking.

[johndoe]

I guess futureproof means for me that the next three years I won't have to buy something new.

[GenoMax]

Please use ADD COMMENT/ADD REPLY when responding to existing posts to keep threads logically organized.

To answer your question: yes this should be future proof using your definition of 3 yrs (with 16G RAM and i5 CPU) to test your scripts/pipelines.

16 GB RAM, for alignment work.

With that much RAM you are not going to be able to align human genome sized data sets. Most aligners need ~30G of free RAM for human genome alignments. Only bwa (which has lower memory requirements) may be able to run in 16G. Please take that into account.

[Gordon Smyth]

Subread can go as low as 8Gb RAM for full genome alignments using a gapped index:

https://doi.org/10.1093/nar/gkz114

[GenoMax]

Good to know. Will keep that in mind.

Answer 1

[Kevin Blighe]

Nothing is truly 'future proof', particulary in tech.

I would definitely upgrade the RAM to 16GB, and get a hard-disk with at least 1TB if possible (unless you expect to use the cloud for storage).

I have been using a HP laptop for 5 years which has 4 CPU cores, 16GB RAM, and 2TB hard-disk. The disk filled up long ago, so, I have another 5TB available in the cloud. Otherwise, it still functions very well, albeit after 2 fan replacements and many lost keys.

Kevin

[JC]

I also use a 4yo HP laptop, I have to change the screen a few months ago but everything else is fine, it has 8-cores i7, 16 GB RAM, 256 GB HD.

Only my "Shift" key is worn-out.

[Kevin Blighe]

You're lucky / Tienes suerte. I have lost 7 keys. Each time that I start the laptop, I disable the keypad manually via xinput list | xinput float XY, and connect a USB keyboard. My CPU is actually:

AMD A10-7300 Radeon R6, 10 Compute Cores 4C+6G × 4

[Jean-Karim Heriche]

I am still working on my mid-2012 MacBook Pro with everything in working order after almost 7 years, including the original battery which still allows me more than two hours of videoconference. Also MacOS seems to have a particularly good way of managing memory as I've occasionally been able to do computations on its 8 GB RAM that ran out of memory on a 16 GB Linux workstation.

Answer 2

[Mensur Dlakic]

In my experience, the biggest obstacle to "futureproof" bioinformatics machines is memory. I bought 4 desktops back in 2011 in a 6-month span. Two of them were identical:

Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz 4 physical CPUs with 2 threads each; 16 Gb RAM

The other two had the same processor Intel(R) Core(TM) i7 CPU 860 @ 2.80GHz 4 physical CPUs but with 16 Gb and 4 Gb memory, respectively. The only one of those 4 that I am not still using is with 4 Gb RAM. Back in those days most common configurations were 4 Gb, and a "recommended" configuration was 8 Gb for bioinformatics applications. Getting twice the recommended memory size kept those computers useful enough even today, though I could never assemble a (meta)genome with them.

Your demands for this machine seem to be relatively small, but I echo the others who suggested that 16 Gb is a must. If you can get 24-32 Gb without a crazy jump in price, that would probably extend your useful window well beyond 3 years.

By the way, whatever your RAM is in the end, make sure that the swap is at least twice the RAM size. I follow this rule even for 64 Gb machine, though it may seem wasteful to commit that much disk space to swap. Disk space is much cheaper - including on SSDs - than physical memory.

[Mensur Dlakic]

It is much easier to make a case for little or no swap when one has 512 or 128 Gb vs. 16 Gb physical memory, so I am not going to argue with your choices for swap size. The point was that even some users who have machines with low memory still don't see value of having large swaps, and I think that's a mistake. Although a concept of low memory is relative, I would consider 8 Gb and below to be low for bioinformatics purposes.

In my experience nothing ever gets done if a machine is swapping, on both big servers and small local machines.

That's true, assuming the machine is swapping throughout the whole run. Yet sometimes there is a peak of memory need, usually at the very beginning while loading in the data, that can't be surmounted without enough swap space, and after that the run can proceed fully in RAM. This didn't necessarily happen in bioinformatics application, but I have numerous examples where I could handle a dataset with 32 Gb RAM and 80 Gb swap, while users with 64 Gb RAM and 16 Gb swap couldn't.

[colindaven]

Interesting, why do you still bother with swap ? In my experience nothing ever gets done if a machine is swapping, on both big servers and small local machines. I do always have some swap, maybe 4GB, but only as a defensive measure. To elaborate, I did have a 512GB server which someone else configured without any swap. It crashed hard a few times if those 512GB were completely used.

[JC]

I agree in the swap size, in my personal server, I have a 128 GB SSD and 2 TB HD, and because I have 128 GB in RAM, I just put 8 GB in swap after checking some recommendations. It is 3 years old and never crashed because memory/swap problems.