Finding The Best Sub-$25 SSD (Adata, Inland, Lexar, Silicon Power)

if you've watched my channel at all you know i work on a lot of budget builds and other small projects and i often find myself in need of cheap 128 gig or 248 ssds i've accumulated a few different models over time but it occurred to me that i really don't have any idea which ones i should be buying so i decided to test out a few different 28 to 25 ssds i got off amazon so that i can hopefully spend my money more wisely in the future and if you're looking to buy a cheap ssd for a project hopefully this will help you as well these are the four drives we're looking at today all of these are two and a half inch say to three six gigabit per second drives and i believe they're all built using 3d nand flash memory but not all of the amazon listings are very clear first up is the lexar ns100 which when i bought it on amazon it was listed for 20.99 the listing shows up to 550 megabytes per second read transfer with write transfer speeds lower i guess we'll have to find out how much lower next we have the inland professional 128 gigabyte which is also listed on amazon at 20.99 this advertises sequential read and write speeds up to 510 megabytes per second and 410 megabytes per second respectively it also says 520 megabytes per second and 450 megabytes per second read and write so i guess we'll just test both moving on we have the silicon power a55 128 gigabyte which was also listed on amazon for 20.99 this model claims to have remarkable transfer speeds that enable faster boot up and improved overall system performance whatever that means this drive also claims to have an slc cache which is a small section of the memory that can be used to temporarily read and write using the slc write method that supposedly can help with read and write speeds but it's not a great caching method for what i've read and i think some of these other drives actually have an slc cache as well they just don't market it the last drive is the adata su-800 which is a little bit unlike the others first it's listed at 24.99 instead of 2099 it claims to have read write speeds up to 560 and 520 megabytes respectively and it has dynamic slc caching as well as a dram cache the dram cache is actually a dram module on the ssd that functions pretty similar to the memory in your computer it essentially gives your drive a place to quickly store and read information and should hopefully help maintain stable speeds over long sequential reads and writes i actually had a fifth drive the kingston 120 gigabyte a400 but it was actually dead on arrival and i didn't want to wait for a return to make this video now just because i happen to get a dead drive it doesn't necessarily mean that the kingston 120 gigabyte a400 is a bad drive but just thought you should know now before we get started with testing i want to be very clear about a few things first of all i'm only using a sample size of one drive so it's very possible that these results could represent outliers rather than the average and expected performance of the drive also i am by no means an expert i know very little about solid state drives and the technology behind them and while i did some research to be more prepared for this video please don't take anything i say as an absolute truth there are tons of other great resources out there so definitely feel free to do your own research and i know what the comment section is probably going to look like at least someone is going to say something along the lines of wtf this guy has no idea what he's talking about and while in many ways i don't know what i'm talking about i think there are many people including myself who have very practical reasons for using small ssds like these a few examples might include a small home server that either doesn't need the extra storage or uses mechanical hard drives or a personal or gaming pc on a tight budget or just a pc for grandma who is really only going to check facebook and maybe download some ransomware for all of the testing i'm connecting the drives to my wife's pc via sata her computer is an hp pre-built with a ryzen 3500 i didn't want to tear down my personal rig and i figured hers would be plenty fast enough to make sure we're getting accurate results we're going to be running three different tests in this video first we're going to reformat each drive and then run crystal disk mark 8 using the default settings after that we'll install windows 10 on each drive and test how long it takes to boot to make sure the installs are identical we'll use clonezilla to duplicate a fresh install to each drive finally we'll run crystaldismarc again but with the drives at a little over 80 capacity to see how well they perform when nearly full to do this i'm just transferring over a list of files to each drive to get the capacity to a little over 100 gigabytes [Music] after testing all of the freshly formatted drives using crystaldiskmark8 we got these results starting with the sequential reads now i'm going to take just a moment here to try and briefly explain some of the terminology but remember i'm not an expert so don't be shocked if i get something wrong here and if i do definitely put in the comments below so i and anyone else can learn from it sequential means we're looking at data that is written sequentially in one spot on the drive so the ssd doesn't have to go looking around in a bunch of different spots this is more representative of how the drive will handle loading large files this one maybe byte is the block size bytes are written to the drive in small sections called blocks that are sort of like pages a small file may take up less than one page or it may take up many pages even if the drive only needs one small bit of information from the block it still has to read that entire block this one maybe by block is very large and helps speed things up for large files but isn't very representative of normal desktop use it does however give us an idea of a more theoretical speed the drive can achieve the queue here stands for qdep which is essentially how many things the drive is being asked to do at once a larger queue can let the drive work on more operations without having to respond improving read and write speeds we also have the thread count which is how many processing threads are accessing the drive at once i think this graph here is measuring the megabyte per second throughput of sequential reads using a very large one that maybe by block size you can see that the a data takes the lead here but all of our drives are pretty comparable in performance with the lexar struggling a bit with a q depth of one and the end line falling behind in both a queue of eight and one we can also measure this test in iops or input output operations per second while the amount of operations performed is pretty crucial to a drive's performance in many applications we won't touch on it much in this video primarily because the iops results are pretty consistently proportional to the megabyte per second results you can see that when we compare the megabyte per second results to the iops results [Music] next we have sequential write speeds using the same block size queue depth and thread count in these tests all of our drives performed nearly identically with the a data just barely maintaining a lead the iops result is once again almost exactly proportional to the megabyte per second result and this trend continues throughout the rest of the benchmarks so i'll just be skipping those results for the rest of the video i will post a link however in the description to a google sheet with all the raw data so you can check those results out if you're interested we'll go ahead and move on to the random reads where things start to get more interesting here the data is randomly stored in smaller four cubic byte blocks across the whole drive which is often what you would find in a real world scenario with a large q depth the a data runs away with this one followed by the inland and then with the silicon power and lexar at around or less than half of the speed of the a data as far as i'm aware though a queue depth of 32 isn't often seen in desktop applications if you know that your application or use case takes advantage of a high q depth then this info will probably be pretty helpful but otherwise a q depth of one is going to be a more realistic benchmark with this the a data actually comes up just short of the silicon power drive the lexar really struggled in this regard and i would imagine that application and file load types on the lexar would be noticeably slower to the user compared to the a data or silicon power with random writes the results are similar to sequential writes the adata su-800 is slightly in the lead here and the lexar falls a bit behind but nothing really sticks out this next graph just shows both of our sequential and random read speeds no new data is here but i thought seeing them all together might help with comparison after installing windows 10 on each drive and then doing an initial boot to make sure everything was good i ran three boot tests starting the timer on the first frame that the power led was on and stopping it when the clock on the windows login screen was rendered i'm showing the median result of each drive here the result isn't very exciting or shocking the lexar a data and inland all managed almost identical times of around 21.5 seconds while the silicon power took about a half second longer at 22 seconds that's possibly within a margin of error but all three tests on the silicon power were 22 seconds or longer so it was a pretty consistent result this is slightly odd considering the silicon power had pretty good read speeds but i think the next set of results might explain why it fell slightly behind i transferred an identical assortment of files over to each drive to get the capacity to a little above 80 this is to see how much performance degradation there was when the drive was close to being full and this is where things definitely get shaken up at least with random reads looking at random reads with a four kilobyte block size and a q depth of one we see that the a data while still maintaining a lead is over 20 slower than when the drive was nearly empty the silicon powered drive took the biggest hit while once in the lead with this test it is now over 40 slower falling behind the inland which actually managed not to slow down at all the lexar even recorded a slight increase in performance although still being the slowest drive in the test and this increase is probably within a margin of error but could also be due to some mistake i made with my methodology with the less realistic queue depth of 32 we see a very similar result the a data and silicon powered drives take a hit and the lexar and inland drives maintain almost identical performance before we wrap things up i thought it would be fun to compare my results to the advertised speed of each drive's amazon listing the lexar stated that it could hit read speeds of up to 550 megabytes per second with lower write speeds looking at the best case scenario of a sequential read result the lexar barely achieved its target hitting 551 megabytes per second and the write speed was indeed lower at 464 megabytes per second so i would say these advertise speeds are accurate the inland advertised reads and writes up to 510 megabytes per second and 410 megabytes per second respectively it barely fell short of its read target hitting 508 megabytes per second on its best result but hit as high as 480 megabytes per second with sequential write the silicon power gave us no indication of actual speeds other than describing them as remarkable i technically made remarks about them so i guess their marketing is pretty spot on here while the a data managed to take the lead in almost all of our benchmarks it unfortunately didn't hit its advertised write speed of 520 megabytes per second but managed to just barely make it above its read target of 560 megabytes per second at 561 megabytes per second [Music] out of curiosity and just for fun after i finished testing i opened up each of the drives and of course our a data drive is the only one that includes a dram module that you can see here so no surprises just looking at the benchmark data it's clear that the adata su-800 is the winner here but that's a little unfair considering the four dollar price difference and the dram cash you get for it so here's my recommendation if you're looking at buying a small ssd to use as a boot drive for your own or someone else's personal computer spend the extra money to get a dram cash model this adata su-800 works well and is as far as i can tell the only 128 gigabyte ssd with a dram cache on amazon however if you don't think you need the drm cache and you're looking at the offering of very cheap 20 128 gigabyte ssds on amazon all i can say for sure is maybe avoid the lexar ns 100 128 gigabyte the silicon power a55 seemed to have the upper hand until we started getting closer to full capacity where it began to struggle with random reads and although the inland was behind in the clean crystal disc mark test it was only by slim margin and its performance was much more consistent the hit to random read performance on the silicon power a55 is massive so this makes me lean towards the inland here but i'll let you make that decision once again i'll leave a link to all of the raw test data in the description below if you want to check that out i have no idea what longevity looks like with these but as i continue to use them i will add any updates either to the description a pinned comment or both so maybe check those places out before making any purchase decisions if you have any experience with these drives or other similar models or if you have some sort of cool fun project using a small ssd leave a comment below i'd love to hear about it and if you want to support the channel a like is always super helpful and you can maybe even consider subscribing to see more content down the road for now though thanks for watching stay curious and i'll see you in the next one [Music] you

Channel: Hardware Haven

Views: 99,382

Keywords: pc building, old computers, old pc, budget pc, budget builds, Computer restoration, cheap nas, budget nas, 128gb ssd, 128gb ssd vs 256gb ssd, best ssd for budget gaming pc, best ssd 2022, cheap ssd for pc, adata su800, lexar ns100 ssd review, adata su800 ssd review, silicon power ssd review, silicon power a55 review, inland ssd review, 128gb ssd review 2022, are 128gb ssds good, should i get a 128gb ssd, kingston 128gb ssd review, dram cache ssd cheap, cheap dram ssd

Id: _ug_x_F-p4g

Length: 13min 44sec (824 seconds)

Published: Thu Feb 10 2022