Slow SATA2 internal speed

Associate
Joined
4 Feb 2025
Posts
6
Location
Dewsbury
Hi all

Had some recent problems with my motherboard/cpu so I have swapped it all out, and in the process fitted a bigger 2tb NVME hard drive, I've also added 2 x 2TB transcend SSD disks by SATA as I store a lot of raw footage for editing in resolve.

I took these 2TB external disks from another system that is now going to be repurposed, they are quite new and again I just used them as a footage store - no system files or anything but I noticed the read speed was very slow on them, but I put it down to the original system being old.

I have fitted these to my new system (AMD 5950x, 32gb ram, RTX4060ti etc) and notice after initially seeming fine, they also slow down to a crawl.

When I say crawl, I mean serious crawl - playing back even gopro footage is impossible with it being frame by frame and benchmarking the read speed it drops down to about 10/15mb/s - a mechanical drive could kick its backside.

Issue seems to affect both drives so I'm wondering if the transcend SSD's are frankly rubbish - but I find it unlikely they'd be *this* bad? or am I doing something wrong or another potential cause?

I dont have exact model to hand but they are transcend drives, 2tb each, both identical about a year old, they don't see many r/w operations jusr used for dumping footage on (never deleted). They are connected internally by short sata cables direct to motherboard.

thanks!
Kris
 
Last edited:
I'll let you know, I ran it last night to try and see what the speed was but didn't pay too much attention to the health (sounds counter productive but I was focused on the speed) - speed wasn't much above 10-15mb when it really bogged down (it does seem to come and go)
 
Last edited:
Issue seems to affect both drives so I'm wondering if the transcend SSD's are frankly rubbish - but I find it unlikely they'd be *this* bad? or am I doing something wrong or another potential cause?

I dont have exact model to hand but they are transcend drives, 2tb each, both identical about a year old, they don't see many r/w operations jusr used for dumping footage on (never deleted). They are connected internally by short sata cables direct to motherboard.
A lot of large capacity drives are now using QLC memory and once you exhaust the write cache performance drops off a cliff. However, read shouldn't really be an issue, so it sounds like maybe something isn't right.

speed wasn't much above 10-15mb when it really bogged down (it does seem to come and go)
Does it happen when writing the drive, or just randomly?

Do you have any related errors in the event viewer when they're slow?
 
Do you have the exact drive models?

When you say they initially seem fine and then crawled to a halt, do you mean inside a single read/write they start fast and then slowdown. Or do you mean over a period of days/weeks use, they seem slower?

Playing back go pro footage is too fuzzy of a metric because that can drastically vary in quality and thus size. Ideally we need proper benchmark results, or at least we need to know example file sizes you're attempting to read or write, if they're video files, how long are these files (we can approximate the bitrate from that + filesize). How full are these drives?

Ideally you should run a tool like CrystalDiskInfo and inspect S.M.A.R.T data, this is metadata the drive collects on its own health, how many sectors are having read/write issues and how many have been marked as unusable. This info will tell you if the disk is just on the way out physically, some tools give you a red amber green summary of how bad it is.

Lastly just to double check, are you read/writing to these disks with any other software? Maybe downloading or seeding torrents from them, using them as cache locations for active software like video editing, used for additional pagefiles for windows? Before running tests you should check the performance tab under task manager and look a drive usage and see if there's any apps read/writing secretly (or otherwise). Are the disks in any way using full disk encryption.

Answering the questions above should narrow the suspects. Start with looking at the SMART info first, if half the drive is dead because of hardware problems, you'll want to pull the remaining data off the drive before you lose it. Report back on the above, we'll advise you.
 
Hi all!

apologies, it has been a manic weekend but I managed to get some photos from crystal disk mark, however they didnt to me seem that conclusive as they seemed to behave.

results-disk1.png
results-disk2.png


The characteristic of the issue is it starts quite reasonably quick - so if I play a large file such as a gopro (around 3gb) video or resolve raw clip (can be up to 20gb) performance will be okay for about 20 seconds then slow to a point where it is doing about 1 frame per second, and closing the file and loading another it will be just as bad. If I reboot etc it will be fine again for a few moments.
The drives are both identical transcends:

Drive Model: TS2TSSD220Q

Not sure what to think - the activity is what I'd expect if a buffer/memory was getting full for transfers but if it only applies to writes then rules it out as I use these drives almost exclusively for storage of video files and use during editing i.e reading

TIA!
 
Last edited:
I don't think it's a cache problem because they primarily effect writes, and playback of footage is reads.

I suspect 2 possible causes, they're overheating and throttling, or the drives have hardware problems and are failing. the fact that they seem to operate ok as "externals" but then were repurposed for internal drives where temps are typically hotter, hints to me it's probably the former. You can check potential problems both with an app called CrystalDiskInfo, made by the same people.

Untitled.png


Make sure you first select the right disk at the top, or under the "Disk" menu. the health status on the top left will summarize all the SMART data the drive has collected about itself over its lifetime into a simple good/bad.

The temperature by default is read out every 10 minutes which is too slow to test, but you can refresh it instantly with the F5 key. If you know how to trigger slow down reliably say by playing back footage then do that. Keep refreshing the temp every 1~2 seconds. See how high it climbs, note the temp at which the speed drop occurs during playback of your file. The app may even trigger an alarm if it gets too high, to warn you. Once speeds drop, likely so will the temps, so we're interested in the peak value it reaches.

The steps above will generally rule in or out bad drive health, and overheating+throttling as potential issues, once we are sure of the cause, we can suggest fixes.
 
Last edited:
To be honest it was slow pretty much instantly with a temperature of 30 degrees c

Could it have anything to do with free space? The second drive actually seems a bit better but despite them both being same design has more free space on it.

Task manager shows the first drive (D:) as 100% utilised when trying to play, but other drive shows 7%

faulty drive maybe after all
 
Last edited:
Ok I've done a quick crash course on the QLC memory type these drives use, it's a known trade off with this memory it seems. QLC NAND memory packs storage tighter per memory block, the benefit is that you get much larger drives for cheaper. The problem is that as they start to get full, the controller on the drive has more work to do to shuffle data around during read/writes. This is standard behavior for SSDs as they have a complex and unintuitive way of reading and writing data to avoid corruption and wear that comes with Nand memory. So even reads can get messy internally on the drive, severely affecting speeds. It's not "faulty" per se, these are just design/price tradeoffs.

The take away from my research is that ideally users should over provision SSD, that for QLC Nand specifically you should target about 70% usage of your drive if you if you want to retain some semblance of the marketed performance characteristics. Past that expect pretty significant degradation in read/writes, especially larger sustained ones. For drives of this size you will probably run into slowdown past about 1.5Tb of usage.

Honestly for your use case I would recommend getting a classic HDD to install internally. Put all the long term storage on there, the oldest files that are least likely to be worked on. Then see how performance is like on the SSDs with more room free on them. Thing is, speed wise SSDs are just a bit meh when they're behind SATA interfaces, they're limited by the SATA bandwidth so often aren't much faster than HDDs.
 
Thank you so much for the detailed reply, it's not what i want to hear obviously but it's given me enough to go away and do some homework and unfortunately it seems you are right - so what I've done is balanced the free space on each and i think I might have to revise my storage strategy, I do wonder if black magic raw could perform okay being pulled from a 7200rpm drive, can hardly be worse can it...?
 
Thank you so much for the detailed reply, it's not what i want to hear obviously but it's given me enough to go away and do some homework and unfortunately it seems you are right - so what I've done is balanced the free space on each and i think I might have to revise my storage strategy, I do wonder if black magic raw could perform okay being pulled from a 7200rpm drive, can hardly be worse can it...?

No problem, I enjoyed learning it. My advice is HDDs for archiving files you want to save but don't need to access frequently. Put your OS on your NVMe if you haven't already because it will keep your OS super responsive. Then use a workflow where you put files ready for editing on your faster C: and edit in place, when you're done, archive those files to your HDD.

There really no serious benefit now a days to SSDs over SATA the trade offs are bad. With minor exceptions.
 
Back
Top Bottom