Review Adata XPG SX8200 Pro Review: Go Pro on a Budget

[Admin]

Today we tear into the XPG SX8200 Pro, Adata’s highest-performing NVMe M.2 SSD yet.

Adata XPG SX8200 Pro Review: Go Pro on a Budget : Read more

 

[popatim]

Did you run into any issues witht he drive being detected on restarts like several forum users reported?

 

[Khadijahp]

Though an limited edition, for comparision purposes the Intel's Ryzen killer Core i9 9900KS should had been included, instead of, or beyond the overclocked 9900K.

 

[Mezoxin]

Did you run into any issues witht he drive being detected on restarts like several forum users reported?

had the drive for quite a while on my system as my main drive never experienced such issues

 

[mac_angel]

I thought you did a review of this drive a long time ago.

 

[Mezoxin]

I thought you did a review of this drive a long time ago.

yes this is an old review from feburary-april i think , it got somehow reposted , i know this because i read this review before buying mine

 

[mac_angel]

lol, same. I have the 1tb one and I love it.

 

[Gillerer]

"Pro" is just a word. You can't draw conclusions from Samsung's use of it to mean "MLC" / "more consistent performance", and apply it to other manufacturers. (You could just as well draw conclusions from AMD's past use of it to mean "partially disabled silicon".)

This simply has an updated controller and firmware compared to the original SX8200, so different - but not necessarily better in all cases - performance characteristics. It's not a different class of product to SX8200 (non-Pro), but rather a later revision that replaces it.

 

[AlistairAB]

Did you run into any issues witht he drive being detected on restarts like several forum users reported?

Have bought several of these, never had issues showing up in the BIOS. I did have that problem with the Intel equivalent once (the Intel 760p or whatever it was called). Anyways that stuff is motherboard dependent, and has nothing to do with the SSD.

 

[plotinusredux]

SM2262EN drives are showing significant slowdowns off the x570 chipset compared to the off the CPU, though we're having trouble getting a response on it. For me (3700x MSI ACE x570 ADATA SX8200PNP) even at Q=1, sequential read is dropping from 2794 MB/s to 2422 MB/s in Crystal. Random read performance is only dropping a little, but random writes are also affected (226 MB/s to 199 MB/s at Q=1, 1451 MB/s to 1041 MB/s at Q=32).

This seems to only be happening with the SM2262EN--see "X570 + SM2262(EN) NVMe Drives" reddit thread where people have tested specifically SM2262EN drives and ones with other controllers in the same system.

If you could test the ADATA and one with another controller on an x570 motherboard, maybe we could get SMI or AMD's attention.

 

[Zpxkma]

Anyone else receive one of these ssds with different dram and nand? My first sx8200 pro had micron/nanya and the second one I ordered came with Unic²/Samsung(sec).

https://imgur.com/a/XujJcFn

View: https://imgur.com/a/XujJcFn

 

[mtdew332]

Anyone else receive one of these ssds with different dram and nand? My first sx8200 pro had micron/nanya and the second one I ordered came with Unic²/Samsung(sec).

https://imgur.com/a/XujJcFn

View: https://imgur.com/a/XujJcFn

Yes mine seems to be different from review units also, but it doesn't seem to affect performance.

 

[Zpxkma]

Yes mine seems to be different from review units also, but it doesn't seem to affect performance.

Yup. That's what I have on my 1TB drive, ADATA branded chips.

 

[Co BIY]

Was just about to buy a XPG SX8200 hundred in 2TB based on the great review in Toms and the good pricing but am concerned about a change in production from "Made in Taiwan" to "Made in China".

The picture in the review and shows a SSD "Made in Taiwan" but when I went to Adata's website the product photo for the 2TB (and all the others) shows photos that say "Made in China".

Is Tom's getting review hardware that is different than the production model or have they made other changes to the product with the change in production location ?

I am buying as few "Made in China" products as I can (Not always easy). The XPG no longer makes the cut for me even if the performance and materials are the same as reviewed.

Some others have noticed a reduction in performance.
https://www.overclock.net/forum/355-ssd/1741060-nvme-nand-dram-bait-switch.html

I think Tom's should remove it from the top spot in the recommendations until the new product is retested.

 

[seanwebster]

Was just about to buy a XPG SX8200 hundred in 2TB based on the great review in Toms and the good pricing but am concerned about a change in production from "Made in Taiwan" to "Made in China".

The picture in the review and shows a SSD "Made in Taiwan" but when I went to Adata's website the product photo for the 2TB (and all the others) shows photos that say "Made in China".

Is Tom's getting review hardware that is different than the production model or have they made other changes to the product with the change in production location ?

I am buying as few "Made in China" products as I can (Not always easy). The XPG no longer makes the cut for me even if the performance and materials are the same as reviewed.

It should be the same/similar hardware just built in a different manufacturing facility. It could just be a stock photo. Can't know unless you have one in your hands.

Some others have noticed a reduction in performance.

https://www.overclock.net/forum/355-ssd/1741060-nvme-nand-dram-bait-switch.html

I think Tom's should remove it from the top spot in the recommendations until the new product is retested.

There is no reduction in performance nor any bait and switch. Adata packages the NAND themselves but they may not always have enough supply. Adata is a very large company so it has variuos suppliers to help fulfil demand. While the NAND packages, in that case, may be UNIC, its most likely IMFT 256Gb 64L TLC as it should be for the product. The performance numbers look fine to me. Performance figures vary depending on the benchmark, the system used, and configuration settings. That user has an X570 motherboard and is probably even using the device in the PCH connected M.2 slot, which adds latency = reduced throughput.

 

[Catbot]

Hi everybody,

I just bought an SX8200Pro after being happy with this review.
However I made an effort to at least compare this new purchase with the things I already had to get some idea if it is an upgrade. So I ran the atto benchmark on the target device with 3 different SSDs - and my results for the 8200 Pro 2TB model are extremely similar to the ones in this report.
The piece I got says Made in China printed in it, so this indicates that the change of manufacturing location does not impact performance.

However there is something I wish this review would have done - it is going to larger IO sizes. What I have observed is that beyond 4MB IOs speeds are dropping - which they do not on a samsung 970 Evo.

I am adding for reference my 2 atto runs with the SX8200 Pro 2TB, and the 970 Evo 1TB below:
ATTO Benchmark Runs (imgur.com)

 

[seanwebster]

Hi everybody,

I just bought an SX8200Pro after being happy with this review.
However I made an effort to at least compare this new purchase with the things I already had to get some idea if it is an upgrade. So I ran the atto benchmark on the target device with 3 different SSDs - and my results for the 8200 Pro 2TB model are extremely similar to the ones in this report.
The piece I got says Made in China printed in it, so this indicates that the change of manufacturing location does not impact performance.

However there is something I wish this review would have done - it is going to larger IO sizes. What I have observed is that beyond 4MB IOs speeds are dropping - which they do not on a samsung 970 Evo.

I am adding for reference my 2 atto runs with the SX8200 Pro 2TB, and the 970 Evo 1TB below:
ATTO Benchmark Runs (imgur.com)

That has to do with the SLC caching policies in the firmware. When the devices SLC cache is full, the Silicon Motion SM2262EN controller will only recover 4-6GB worth of cache within a few seconds after the workload ends and it is left to idle.

I let my 500GB and 1TB samples idle for up to half an hour and noted that only 6GB of the cache was still all that recovered. So, because SLC recovery isn't prioritized like it is on other controllers like Phison's E12S/E16 where it will recover until the whole cache is available again, you are experiencing this slowdown. Other SSDs behave like this, too. Leaving hot data in the cache can improve performance as the SLC is faster access.

If you run Trim (Windows Optimize) and let it sit for overnight or maybe all day, you should get a clean graph. I haven't tested the exact time window as that would take too long.

 

[steiNetti]

BE AWARE!

ADATA silently changed the controller in the latest revision from SM2262EN to SM2262G. This not only compromised performance, but worse it makes this NVMe incompatible with basically ALL GIGABYTE Z390 boards and most of their X570 lineup (even though they list the same SKU in their vendor compatibility list and there is no way to descipher the versions before opening the box)!

This article needs revisiting and there needs to be put a warning on top of it! The SX8200 PRO (PNP) sold in 2020 is not the same model as reviewed here, even though they both have the same SKU!

PS: I can confirm that this is real as I have received a SM2262G (you can read it on the warranty sticker and it is not working in any of my GIGABYTE boards).

Sources:
View: https://www.reddit.com/r/hardware/comments/jiwyut/adata_has_made_a_secret_revision_to_the_sx8200/

ADATA XPG SX8200 Pro 1TB NVMe SSD Review

In our ADATA XPG SX8200 Pro review, we see how this NVMe SSD performs which is especially interesting given the drive can have different controllers onboard

www.servethehome.com

 

[Deleted member 14196]

I wish this website would stop trying to sell these so hard. I am never going to buy adata

 

[Zpxkma]

That user has an X570 motherboard and is probably even using the device in the PCH connected M.2 slot, which adds latency = reduced throughput.

I'm "that user" and no. I have it in the 2nd M.2 slot which uses CPU lanes. The board is an X570 ACE, that has the third m.2 slot connected to the PCH.

 

[seanwebster]

I wish this website would stop trying to sell these so hard. I am never going to buy adata

We no longer recommend it. I've dropped it from our Best SSDs list. I do not recommend products that differ from the original unless they improve said product.

https://www.tomshardware.com/news/adata-and-other-ssd-makers-swapping-parts

I'm "that user" and no. I have it in the 2nd M.2 slot which uses CPU lanes. The board is an X570 ACE, that has the third m.2 slot connected to the PCH.

Sorry to break it to you, but the second M.2 goes through the PCH, not direct to CPU - only the first slot is direct to CPU. Check your manual, it shows clearly.

Both the SM2262 and SM2262EN work on X570, it’s the platform I test most drives on nowadays. If it doesn’t, I would blame the mobo manufacturer / AMD for a poor NVMe supporting UEFI. Newest UEFI revisions have much better compatibility and support.