Plotting dual SSD support, two temporary directories

In the second step of the plotting configurator, two temp folders can be specified. How does it work? Is it possible to speed up plotting when using two physical SSDs? Do I get it right?

I mean this:

Isn’t it better to have those SSDs two?

What is better:

A / One 1024 GB SATA SSD
B / Two 512 GB SATA SSDs

Thank you for the information.

1 Like

The use of the 2nd temp directory just splits where the phases are being written to. If you specify a second location, Data from phase 2 will be read from ssd1 and then written to ssd2 in phase 3 of the plotting process.

1 Like

Also covered here

I understand that it is impossible to say whether using the second directory is better in anything. It also debates how to use this directory.

Can someone please explain to me:

  1. Why would I use / not use the second directory?
  2. How to use / not use the second directory.

Thank you very much for the simple and accurate information without unnecessary assumptions. Does it matter and in what case?

I’m considering buying a second SATA SSD, but I don’t have a m.2 for nvme.

1 Like

@codinghorro: The link to another thread of the forum that you send is also just full of assumptions and no one knows anything. :frowning:

OK, well, run your own tests and report back!

It has been my experience that using a Temp2 drive really only helps when paired with my NVME drive. Although it could just be my particular mashup of equipment, my general lack of full optimization, or even placebo.

I’ve got 3 types of SSD in my rig: 1 x Corsair MP510 1920GB (NVME), 1 x SanDisk Ultra 3D 1TB, and 3 x PNY CS900 500GB. Plotting using the PNY drives for temp space is, expectedly, much slower than the NVME and even the SanDisk. It takes the PNY drives about 20-25% longer to do one plot than my NVME doing 2 simultaneous plots. But somehow they had fantastic Phase 4 times – it’s half that of either of my “better” drives. This got me thinking about Temp 2.

My current plotting configuration has my NVME and my 1TB SanDisk each running staggered parallel plots – up to 3 on the NVME and 2 on the SATA, staggered far enough that they couldn’t run more even if allowed. The NVME fires plots at a regular interval (determined from measured plot phase times), and the SanDisk (SATA) doesn’t fire a second plot until the start of Phase 3.

The PNY drives are set up for Temp 2, and receive the plots from both of my faster plotting drives. The PNYs wrap thing up on those plots while the “better” drives then start new plots (if stagger timer allows. It’s not always immediate, but they’ve got other plots running so it’s not idling).

The only real benefit here is freeing up the NVME to start a new plot sooner. I’ll have to actually look at logs or something to quantify the actual difference, but I’d guess it’s at least one extra plot per day (remember I’m not optimized at all, if you are then expect less impact). The SATA SanDisk doesn’t see much benefit, but I’ve got 3 PNYs and I like to keep things even. It does save a nominal amount of time per plot though, maybe a couple minutes per plot, which is very much welcome in my highly de-optimized setup.