My SSD always does uber high amounts of garbage collection. In this case, does TRIM help much?

[Eug]

I have this SSD in one of my secondary Mac laptops.

https://www.anandtech.com/show/4010/kingston-ssdnow-v-plus-100-review

Anand states that this one does so much garbage collection that there is NEVER a significant performance decrease from repeated writes.

Before fragmentation:

After fragmention:

Basically no difference in speed.

In fact, that's why I bought this drive ages ago. OS X at the time had no TRIM support, so this was a good way to compensate. However, not so long later OS X incorporated TRIM support, so I activated TRIM.

I'm curious though, in this particular case is TRIM actually helping much, perhaps to decrease NAND wear? (I know it makes a huge difference for my Samsung SSDs to keep them running fast.)

---

BTW, in 2017, what are the best SSDs for garbage collection without TRIM to maintain performance? For an external drive with no TRIM, my Samsung 850 EVO becomes almost useless when near full, after repeated writes, due to dramatic performance losses.

 

[UsandThem]

BTW, in 2017, what are the best SSDs for garbage collection without TRIM to maintain performance? For an external drive with no TRIM, my Samsung 850 EVO becomes almost useless when near full, after repeated writes, due to dramatic performance losses.

The Crucial BX300 with MLC maintains performance even when mostly full. Also, with larger SSDs (500 GB 850 EVO for example) dropping in price recently after new model launches (BX300, SanDisk/WD 3d), you could get a larger unit where you don't fill it up so much.

https://www.anandtech.com/show/11766/the-crucial-bx300-480gb-ssd-review-back-to-mlc

 

[Eug]

Thanks, The Crucial BX300 looks interesting. I have an 850 EVO 500 GB that I was using as a USB boot drive without TRIM, but the performance was dropping through the floor once I got significantly over 400 GB. Unfortunately, 1 TB still gets a bit spendy. I don't really need the transfer speed so much as the reduced latency.

 

[UsandThem]

Thanks, The Crucial BX300 looks interesting. I have an 850 EVO 500 GB that I was using as a USB boot drive without TRIM, but the performance was dropping through the floor once I got significantly over 400 GB. Unfortunately, 1 TB still gets a bit spendy. I don't really need the transfer speed so much as the reduced latency.

If I were going to buy a larger drive, say 1TB, I would personally look at the Sandisk Ultra 3D. Sure the Mushkin 1TB goes on sale pretty often in the $250 range, but there are too many "died after 6-9 months" reviews for me to personally buy one when I can spend another $25 for one with better reviews. The Sandisk Ultra 3D is cheaper than the 850 EVO (around $285 vs $350 right now for the Samsung).

https://www.anandtech.com/show/1179...estern-digital-wd-blue-3d-nand-1tb-ssd-review

However, right now at the 500GB size, the 850 EVO dropped in price, so there are several good choices in the $130 - $150 range.

 

[Glaring_Mistake]

I think that what you've seen with your 850 EVO is something similar to what I've seen sometimes when writing a lot to some drives without TRIM (after removing a lot of files) which could both significantly inrease WA/decrease write speeds.
It is not so that it was just slower drives that were affected either but even an 850 Pro had very high WA and low write speeds under those conditions.

Don't think I've seen it happen with a drive using a Marvell controller though so the BX300 and the Blue/Ultra 3D that have been recommended here may likely be good options and do seem to have pretty stable performance in general too.
A 250GB WD Blue 3D I tested did not seem to be have that stable performance.but seeing as pretty much all the reviews of the drive have been conducted with the 1TB drive, not the 250GB it may be that it is either due to the capacity or that the 250GB needs a firmware update.

Finally, it's a pretty old drive but maybe the Intel 730?
Remember someone testing (a few) drives without TRIM and trying to prevent performance to drop too much who found that a drive using a SandForce controller (likely due to its compression) performed pretty well.