ADATA Explains Changes with XPG SX8200 Pro SSD

[randomUser]

...only to find out that Kingston laser-etched all chips to be unidentifiable, in good spirits of shady Aliexpress counterfeit chip sellers ))))


There are two important quotes from Toms that are missing, which would make things clearer for people who weren't following the issue:
1) Adata has reached to clarify that the Redditor measured performance with the new SSD configuration connected to the PCH and compared performance to the originally-shipping SSD connected to the CPU.
2) Adata has clarified that the endurance rating remains the same.

Pretty sure that with either variant or configuration the actual performance will be within the margin of error(~10% sequential). I did own both SX8200 and I do own an SX8200Pro, and as far as I know, the only noticeable difference is random R/W, and only in benchmarks. Not something you'll even notice in daily use.


You are paying for a product that fits the spec, not for tiny letters on tiny chips.
AFAIK, either an official spec on their website or product datasheet even mentions which 2262 gen it has and who's NAND do they slap on it.
Not every company has means of manufacturing their own NAND and controllers.

I also have SX8200 and SX8200 Pro. There is no noticeable difference in performance, however, there is noticeable difference in degradation. non Pro version seems to degrade much much faster when given many little files to write to it.
sx8200 480GB version, 9,2TB total host writes @ 92% health
sx8200 pro 1TB version, 8,2TB total host writes @ 99% health
both are living similar life in terms of how are they used and what is being done with them, so results are comparable.

 

[Kaczorowaty]

Sooo, there is another version of SX8200 Pro.

ADATA SX8200 Pro v7 [ SM2262ENG + Micron 64L TLC ]

 

[TheLostSwede]

I think people who are savvy should understand that 3rd parties like Adata have no control over the parts they use in their products since they produced none of it. They usually just buy off the shelf products. But for Adata to quietly make those changes and with obvious degrade to performance, I feel they could have handled the situation a lot better. In my opinion, they can't get away with this bad business decision. They've lost the credibility that they have built up over the last few years, and unlikely to get it back any time soon.

That's not quite true though, as Adata actually make their own DRAM modules and SSDs and do in fact manufacture for a lot of other companies.
This is also why they have their own branded memory, even though it's made by someone else.
So in other words, Adata has control of what they're making, but maybe not of what parts are available to them at all times.

 

[cellar door]

Here is mine, looks to be like yet another version

 

[silentbogo]

I think there is another version, similar to v5, but uses Micron 96L TLC. I call it v1.5. But we can call it v6. I have two reports of it.

Mine is the same. 1TB version bought at the very end of 2019.

I've managed to find a couple of reviews of variants w/ 2262G controller, and while results vary slightly, they aren't too far off from "The better version". Also, it's worth reading a conclusion for both.

Review - ADATA XPG SX8200 Pro 1TB M.2 NVMe SSD

M.2 SSD nowadays is capable of offering even faster performance than a regular SATA SSD at a very competitive price point and the ADATA XPG SX8200 Pro which

www.tech-critter.com

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

 

[Athlonite]

Hmm somewhat different version again using 96 layer Intel nand chips

Hmm who's lying to whom Aida64 seem to think it's Micron Nand

 

[ThrashZone]

Hi,
Best to not test adata stuff and just enjoy
If it tests 500 read/ write lower than spec's who cares right

 

[Sir Alex Ice]

While ADATA may not have perfect control on what parts they find on the market to make their products, it still is a dick move not to tell up front that the controller or flash memory have changed.

 

[maxfly]

Hmm somewhat different version again using 96 layer Intel nand chips
View attachment 190719

Hmm who's lying to whom Aida64 seem to think it's Micron Nand

View attachment 190721

Apparently Aida only has a database it goes by it doesnt actually read the controller or nand information. So the smi_nvme_flash_id tool or looking at the drive itself are the only ways to find out what controller and nand you have. Altho i couldnt read the nand on my 1tb drive. I think the nand information is under the warranty sticker side on 1tb drives. So the smi id tool may be the only way to get your nand info on those unless you dont mind voiding your warranty

Anyhow i finally had time to check the two sx8200 pros ive got left.
Bought one on 3-1-2019 a 512gb, sm2262en/micron 64L TLC. Bought on 4-3-2020 1tb sm2262eng/micron 64L TLC.
I somehow managed to avoid the crappy sm2262g controller and samsung 96L NAND on both drives despite there being a year between each purchase and the capacity being different.

 

[Athlonite]

@maxfly yeah i thought it strange I might suggest that aida team incorporate some form of software to actually read the hardware instead of relying on a misinformed database of product descriptions

 

[fb020997]

Mine is the same. 1TB version bought at the very end of 2019.
View attachment 190674

I've managed to find a couple of reviews of variants w/ 2262G controller, and while results vary slightly, they aren't too far off from "The better version". Also, it's worth reading a conclusion for both.

Review - ADATA XPG SX8200 Pro 1TB M.2 NVMe SSD

M.2 SSD nowadays is capable of offering even faster performance than a regular SATA SSD at a very competitive price point and the ADATA XPG SX8200 Pro which

www.tech-critter.com

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

We have the same exact SSD. I bought mine around March 2020.

 

[s3thra]

I seem to have another SM2262ENG + Micron 64L TLC:



I bought it in Nov 2019 IIRC.

 

[Athlonite]

I seem to have another SM2262ENG + Micron 64L TLC:

View attachment 192793View attachment 192794

I bought it in Nov 2019 IIRC.

That's an 512GB Adata XPG SX8200 Pro so it's a different beast altogether we're talking about the 1TB version

 

[s3thra]

That's an 512GB Adata XPG SX8200 Pro so it's a different beast altogether we're talking about the 1TB version

Ah, my bad.

 

[TheBacon]

So having the ENG on the chip is better than the G right? It's sooo confusing revisions with different performances.

Also the product box does have a version, the 7.2 on a review of the G chip, 7.3 is again ENG, etc...

 

[GabrielLP14]

My new SX8200 Pro just arrived today, and it came with with the new Samsung V6 6th V-NAND (128-Word Layers)

I'll be checking pSLC Caching volume and recovery times, but i think it might be a Smaller Static pSLC Cache.
Fortunately it came with Awesome Dies. I'd just hoped it was the 256Gb instead of the 512Gb for better performance

So having the ENG on the chip is better than the G right? It's sooo confusing revisions with different performances.

Also the product box does have a version, the 7.2 on a review of the G chip, 7.3 is again ENG, etc...

No they're both the same controller, but different solder joints, one is BGA and the other SM2262G means TFBGA

 

[fb020997]

My new SX8200 Pro just arrived today, and it came with with the new Samsung V6 6th V-NAND (128-Word Layers)
View attachment 226728

I'll be checking pSLC Caching volume and recovery times, but i think it might be a Smaller Static pSLC Cache.
Fortunately it came with Awesome Dies. I'd just hoped it was the 256Gb instead of the 512Gb for better performance


No they're both the same controller, but different solder joints, one is BGA and the other SM2262G means TFBGA

Another one??? Is that the… 8th or 9th revision?
it would be cool to have all of them and do a, say, PCMark run or some real life test on the same machine, just to see how the performances are affected by the component choices.

 

[GabrielLP14]

Another one??? Is that the… 8th or 9th revision?
it would be cool to have all of them and do a, say, PCMark run or some real life test on the same machine, just to see how the performances are affected by the component choices.

11th, there are even others

Another one??? Is that the… 8th or 9th revision?
it would be cool to have all of them and do a, say, PCMark run or some real life test on the same machine, just to see how the performances are affected by the component choices.

i'm planning to do it, a more in depth review, i did a 30 min video reviewing my SX8200 Pro, i was gonna do a deep dive into the main different NANDs topology and specs of each drive. But it would be an hour long video rofl.

 

[fb020997]

11th, there are even others


i'm planning to do it, a more in depth review, i did a 30 min video reviewing my SX8200 Pro, i was gonna do a deep dive into the main different NANDs topology and specs of each drive. But it would be an hour long video rofl.

11 revisions……
Well, this could be a great ad:

“SX8200 Pro...now turned up to 11*!”

*11 revisions. Disclaimer: Adata reserves the right to modify specs, components and performance at any moment.

Yes, there’s a teeeeeeny tiny asterisk between 1 and !, as with every “good” ad would be XD

 

[bug]

11 revisions……
Well, this could be a great ad:

“SX8200 Pro..now turned up to 11*!”

*11 revisions. Disclaimer: Adata reserves the right to modify specs, components and performance at any moment.

Yes, there’s a teeeeeeny tiny asterisk between 1 and !, as with every “good” ad would be XD

Except that by the time you buy it, it's already turned up to 12 or 13!

 

[GabrielLP14]

11 revisions……
Well, this could be a great ad:

“SX8200 Pro..now turned up to 11*!”

*11 revisions. Disclaimer: Adata reserves the right to modify specs, components and performance at any moment.

Yes, there’s a teeeeeeny tiny asterisk between 1 and !, as with every “good” ad would be XD

Rofl

Except that by the time you buy it, it's already turned up to 12 or 13!

i don't doubt it, i bet in the near future we'll see these SSDs with Micron's FortisFlash B47R..... but i'm gonna be honest i think this revision with V6 had a much better performance than the original Micron's FortisFlash B16A / B17A, the only problem is that apparently ADATA capped the sustained Write Speeds:
i'd recommend watching this video, PS: it's in portuguese (since i'm from Brazil), but nevertheless check out in "21:00" (21 minutes) in which i test it's pseudo-SLC Caching. Which was Hybrid (Static + Dynamic) and i also tested how much GBs it recovered in some idle rounds.

The problem was that ADATA apparently capped the sustained write speeds, maybe because of better endurance, since the NAND was running in 325.5 MHz (650 MT/s) [Bus between controller and NAND] instead of the 800 MT/s (400 MHz) that each controller channel can handle. Or even to make chia minners look away from this SSD. But with 16 dies of V6, this drive SHOULD have a higher sustained write speeds compared to the first revision of SX8200 Pro

 

[bug]

Rofl


i don't doubt it, i bet in the near future we'll see these SSDs with Micron's FortisFlash B47R..... but i'm gonna be honest i think this revision with V6 had a much better performance than the original Micron's FortisFlash B16A / B17A, the only problem is that apparently ADATA capped the sustained Write Speeds:
i'd recommend watching this video, PS: it's in portuguese (since i'm from Brazil), but nevertheless check out in "21:00" (21 minutes) in which i test it's pseudo-SLC Caching. Which was Hybrid (Static + Dynamic) and i also tested how much GBs it recovered in some idle rounds.

The problem was that ADATA apparently capped the sustained write speeds, maybe because of better endurance, since the NAND was running in 325.5 MHz (650 MT/s) [Bus between controller and NAND] instead of the 800 MT/s (400 MHz) that each controller channel can handle. Or even to make chia minners look away from this SSD. But with 16 dies of V6, this drive SHOULD have a higher sustained write speeds compared to the first revision of SX8200 Pro

Sustained writes is what causes a NVMe SSD to throttle. If you're copying that much data, it's probably faster if it starts slower and prevents throttling than starting full speed and choking along the way.

 

[GabrielLP14]

Sustained writes is what causes a NVMe SSD to throttle. If you're copying that much data, it's probably faster if it starts slower and prevents throttling than starting full speed and choking along the way.

Correct, and this "heat spreader" helps with nothing, it's just asthetics

 

[bug]

Correct, and this "heat spreader" helps with nothing, it's just asthetics

I never tested it, but I suspect most heat spreaders are just for show. No heat spreader works without air flow and with proper air flow, you may not need the heat spreader at all. But we're drifting.

 

[GabrielLP14]

I never tested it, but I suspect most heat spreaders are just for show. No heat spreader works without air flow and with proper air flow, you may not need the heat spreader at all. But we're drifting.

Indeed, i didn't had a thermal camera at the moment to test it, but after seeing after the review, the heat doesn't really dissipates, in this particular drive is 100% asthetics