unexpected disk benchmarks are confusing

[shima-pan]

Hello everyone, I know this isn't exactly a linux issue, at least I think, but I can't find relevant info searching anywhere and I had just made this account here earlier so yeah why not. Feel free to ignore this but hopefully someone has some insight or input.

Basically I have 2 hdds, a 2.5" wd black ~1 year old maybe a bit more, and a 3.5" wd blue freshly bought, so not too far apart. The files on them are currently identical (except for a few timeshift backups on the blue). I just benchmarked both, and got some weird results. AFAIK both drives should be the same speed considering the specs on them are basically identical and from what I read performance differs little between the sizes, if anything small=faster... But, as you can see in the attached images, they are not. I had expected the black to be a bit faster but that is not the case. The blue is much more so, +25 avg write, although it seems to fluctuate more in its speeds.

So I'm just wondering, is there any real reason this should be happening? Maybe some setting to tweak or something? First I thought it might be the new sata cable in use on the black, but I also have a SSD on a cable from the same pack which is doing 550/read so I do not think its that.. The blacks, and 2.5's in general from what I've heard, are supposed to be faster though, and 1 year shouldn't degrade a drive that much, so something must be up here.

Also, how does the benchmark in the disks utility work exactly? Is it r/w-ing incrementally along the length of the disk? (and would more/less files on it affect the results?) Because if so it looks to me like around 70% where the blue drastically drops off indicates something around that area that isn't good. And the black being appreciably slower makes me wonder if it isn't starting to go itself..

Am I just overthinking all of this? I just wanna decide which drive to use as main and which as backup, it looks like the blue is a better choice but the large fluctuations and big drop-off are making me wonder if one or both of these drives aren't screwy.

Sorry for the super long post about hard drives lol. I have never had to deal with managing multiple drives before and having all my files in migration is making me cloudy.
Also let me take a moment to say how wonderful it feels to use this OS and get away from windows, many many thanks to all who make this stuff possible :)

[gm10]

You benchmarked two partitions, not two drives. Differences in position and size of those partitions alone can explain the difference you are seeing even if the drives were identical. And why on earth would you trouble yourself to black out the drive's model in each screenshot?

[rui no onna]

You benchmarked two partitions, not two drives.

sda1 and sdc1 should be different drives, no?

@shima-pan
That's expected behavior actually. Something to do with fewer sectors read/write in one spin on the smaller drive, iirc.

You can kinda "short-stroke" the 3.5 WD Blue by partitioning, say, using the first 100-500 GB for Linux, swap and files you want in the fast portion, while using the remainder for bulk data storage. That way, your operating system, programs and frequently used data will be in the fast portion of the drive.

[gm10]

sda1 and sdc1 should be different drives, no?

/dev/sda and /dev/sdc are drives. /dev/sda1 and /dev/sdc1 are partitions on them.

[rui no onna]

/dev/sda and /dev/sdc are drives. /dev/sda1 and /dev/sdc1 are partitions on them.

Yes, those are partitions on the two different drives. Based on the graphs (which look pretty normal to me bar the Blue write slowdown at 70%), both partitions appear to use the full 1TB on each drive.

[shima-pan]

You benchmarked two partitions, not two drives. Differences in position and size of those partitions alone can explain the difference you are seeing even if the drives were identical. And why on earth would you trouble yourself to black out the drive's model in each screenshot?

I sorta realized that distinction after posting last night, but in any case both drives only utilize 1 partition of the same size. Also this is my first real foray into multiple drives and partitions so idk if those strings have anything unique, I figured someone might think that was stupid though lol.

That's expected behavior actually. Something to do with fewer sectors read/write in one spin on the smaller drive, iirc.

You can kinda "short-stroke" the 3.5 WD Blue by partitioning, say, using the first 100-500 GB for Linux, swap and files you want in the fast portion, while using the remainder for bulk data storage. That way, your operating system, programs and frequently used data will be in the fast portion of the drive.

Hm, interesting. I will have to look into that re:sectors
Actually considering I have two drives equal size maybe I will just partition both this way and use the the second half of both for backup storage, maybe mirroring each first half to the other drive's second half. Unless there is some reason you only suggested that with the blue. OS is on a SSD but size on that is limited. Anyways ty for input. Also reminded me to make a thread about swap partitions sometime today.

Based on the graphs (which look pretty normal to me bar the Blue write slowdown at 70%)

Ugh I knew there was something weird about that.. I also had a weird issue with that drive(or partition, w/e) not mounting after a power outage, and I had to repair the filesystem from disks utility which seemed to fix it.. I'm behind a surge protector though.. Idunno.

Anyways thank you for the input thus far everyone.

Oh wait, I just remembered another weird thing. After first formatting the blue there was something like 50gb already in use (yellow block at start of partition), whereas when I did the exact same format with the black there was only ~15gb. I never did find out what that was all about, but I can't imagine why it even should do that. What a waste of valuable first sectors. Someone said it could be some reserve cache but then why would both not be the same if I made them the same in gparted.. Idk, too many questions, not trying to be annoying.

[rui no onna]

Actually considering I have two drives equal size maybe I will just partition both this way and use the the second half of both for backup storage, maybe mirroring each first half to the other drive's second half. Unless there is some reason you only suggested that with the blue.

Why Blue? Based on your graph, first 50% of the 3.5" Blue ranges 160-200 MB/s read and 130-150 MB/s write. Meanwhile, first 50% of the 2.5" Black ranges 130-150 MB/s read and 90-110 MB/s write. For the 3.5" Blue it only drops to 140 MB/s read and 100 MB/s write at 70%.

If you're after performance, I'd be more inclined to use the 3.5" Blue over the 2.5" Black.

[shima-pan]

Actually considering I have two drives equal size maybe I will just partition both this way and use the the second half of both for backup storage, maybe mirroring each first half to the other drive's second half. Unless there is some reason you only suggested that with the blue.

Why Blue? Based on your graph, first 50% of the 3.5" Blue ranges 160-200 MB/s read and 130-150 MB/s write. Meanwhile, first 50% of the 2.5" Black ranges 130-150 MB/s read and 90-110 MB/s write. For the 3.5" Blue it only drops to 140 MB/s read and 100 MB/s write at 70%.

If you're after performance, I'd be more inclined to use the 3.5" Blue over the 2.5" Black.

I think I may not have been clear with what I meant, but I understand what you meant now so its okay.

But yeah, it does seem that even at 50% the blue is basically faster than the black from the start. I still think this is all a bit weird but who knows, it could probably be one of a number of things and anyways as long as it isn't a sign of some kind of defect I can live with having one run a bit slower. The black was basically in 24/7 use in a laptop for that year or so, maybe its not too far-out to think it could just be ageing.

Tbh I am more weirded out at this point by the 50gb of ghost data on the blue after format. Maybe I'll try reformatting it and see if I can get that number down somehow, and maybe run a bunchmark on the empty drive to compare how that effects results.

Anyways, I guess I'm just gonna use the blue. Black has the long warranty which woulda been nice for heavy use but no big deal, the stability control technology and all that is mostly wasted on a desktop most likely and I probably wouldn't send in a dead drive with all my data on it anyways :P

[rui no onna]

But yeah, it does seem that even at 50% the blue is basically faster than the black from the start. I still think this is all a bit weird but who knows, it could probably be one of a number of things and anyways as long as it isn't a sign of some kind of defect I can live with having one run a bit slower. The black was basically in 24/7 use in a laptop for that year or so, maybe its not too far-out to think it could just be ageing.

Nothing to do with ageing. HDDs don't really slow down with age although it might seem that way since HDDs typically get filled up and on Windows, become more fragmented. Really just the characteristics of the drive. Despite both being 1TB 7200 RPM, they probably have different platter densities.

Even new, they just normally perform like that. Heck, I remember owning 1TB 3.5" green drives (5200-5900 RPM) that were just as fast or faster compared to 1TB 2.5" 7200 RPM drives. That was quite a while back, though. I'm mostly using SSDs now with my only HDDs being external USB or in the NAS.