Question Both Samsung 990 2TB SSDs producing quite different results in CrystalDiskMark than would be expected

Aug 28, 2023
30
7
35
I recently got 2 2TB Samsung 990 for my computer. When testing the first drive (Drive F) it met the numbers similar to advertised speeds, but the second one (Drive G) ended up being slower than the first one despite being the exact same model. It's making me confused as I can't find a single thing online about such a difference in speed, and while the numbers are still good numbers... 900MB/s difference in SEQ1M Q8T1 read for example seems quite hefty in difference.

As far as the specs of my motherboard go both slots are PCIe 4.0, and I have tested multiple times with CrystalDiskMark and Samsung's tests with same-y results in both. Does it have to do with placement on the motherboard aka closer to CPU = better performance, is it a difference in speed between 2280 vs 22110 slots (though I haven't heard of that being the case before) or is it something else? It would be greatly helpful to hear from someone about this.

I provided one set of results in text below for both drives from CrystalDiskMark.

Drive F:


Drive G:

 
Last edited:
Aug 28, 2023
30
7
35
What specific motherboard?

Screencaps of your CrystalDiskMark, or Samsung Magician, are preferable to the copy/paste text.
My motherboard is the Asus Prime X570-P, and I only went with text as I didn't have my Imgur readily available. I updated the post with the pictures instead.
 

USAFRet

Titan
Moderator
All my specs should be under ever post and comment I leave, but CPU is 5800X3D.
OK. Not all systems under discussion are the same as what is in someones sig.

I asked, because the G series Ryzen CPUs have different performance withe the M2_1 port.

Not the case here.

What does Samsung Magician show?


In reality though...that "difference" is only seen in benchmark tools like this.
 
Aug 28, 2023
30
7
35
OK. Not all systems under discussion are the same as what is in someones sig.

I asked, because the G series Ryzen CPUs have different performance withe the M2_1 port.

Not the case here.

What does Samsung Magician show?


In reality though...that "difference" is only seen in benchmark tools like this.
Ah makes sense, I know some people have multiple computers in their signature and likely many that don't when they do have multiple setups so the question about it makes sense.

Yeah I seen about that with G series when looking at the list of specs for the motherboard. Okay so Samsung Magician roughly says the same as CrystalDiskMark, besides Seq Write speed reporting 100-300MB/s slower on Drive G when already slower than F. Otherwise pretty much identical for both drives.

I know it won't make much a difference in real use case (especially gaming), but it doesn't sit right when it's that much of a difference between 2 of the same SSD models of the same capacity.
 
Ah makes sense, I know some people have multiple computers in their signature and likely many that don't when they do have multiple setups so the question about it makes sense.

Yeah I seen about that with G series when looking at the list of specs for the motherboard. Okay so Samsung Magician roughly says the same as CrystalDiskMark, besides Seq Write speed reporting 100-300MB/s slower on Drive G when already slower than F. Otherwise pretty much identical for both drives.

I know it won't make much a difference in real use case (especially gaming), but it doesn't sit right when it's that much of a difference between 2 of the same SSD models of the same capacity.
Swap the disk+slots.
Does the problem follow the disk or the slot?
 
I'm guessing the slower one is connected to the chipset and other one is CPU which is the likely reason behind the speed difference. As Bob said you can swap the drives to see if it's the drives or the slot, but I'm betting it's the chipset v cpu difference.

edit: note the random performance is basically margin of error between the two
 
Last edited:
Aug 28, 2023
30
7
35
Swap the disk+slots.
Does the problem follow the disk or the slot?
That... would be such a massive hassle involving removing GPU and I don't want to wear out the power connector on the GPU as I already had to remove the GPU so far more than I like (it's a 4090, thus newer port rated for less cable insertions, unless this info is out of date)

I don't know, I will have to really consider it before doing because of that. (Seriously why do they put slots under GPUs)
 
Aug 28, 2023
30
7
35
I'm guessing the slower one is connected to the chipset and other one is CPU which is the likely reason behind the speed difference. As Bob said you can swap the drives to see if it's the drives or the slot, but I'm betting it's the chipset v cpu difference.

edit: note the random performance is basically margin of error between the two
Is there a way to know based on spec sheet of the motherboard what is connected to chipset and CPU? As it wholly could be that since well, one is closer to the CPU and that is the one that is reaching advertised max speeds. Also I seen their comment, but as I said in response to them that would be incredibly annoying to do and would require fully removing the GPU which I really don't want to do right now as I have heard the 40 series GPU power connectors aren't rated for that many cable insertions. I would really have to consider it first.

What are you referring to as "random performance"? Since I am unsure what you are referring to specifically since as for results, they are very consistent ever test that was done.
 

USAFRet

Titan
Moderator
Is there a way to know based on spec sheet of the motherboard what is connected to chipset and CPU?

The M2_1 is CPU, the M2_2 is chipset.
pg 1-2 of the English user manual has a pic of which is which
https://dlcdnets.asus.com/pub/ASUS/...PRIME_X570-P_UM_v3_WEB.pdf?model=PRIME X570-P
 
What are you referring to as "random performance"? Since I am unsure what you are referring to specifically since as for results, they are very consistent ever test that was done.
The bottom two rows in CrystalDiskMark test random performance and they're both basically the same between the two drives which leads me to believe it's the slots they're in rather than the drives themselves since only sequential seems to be impacted.
 
Aug 28, 2023
30
7
35

The M2_1 is CPU, the M2_2 is chipset.
pg 1-2 of the English user manual has a pic of which is which
https://dlcdnets.asus.com/pub/ASUS/mb/SocketAM4/PRIME_X570-P/E17441_PRIME_X570-P_UM_v3_WEB.pdf?model=PRIME X570-P
I meant to reply much sooner, I want to say I appreciate you pointing that out! Thank you. It exactly lines up with how the drive under where my GPU is installed is the M2_1 one and the one lower on the board is the M2_2 slot and explains the performance difference.
 
Aug 28, 2023
30
7
35
The bottom two rows in CrystalDiskMark test random performance and they're both basically the same between the two drives which leads me to believe it's the slots they're in rather than the drives themselves since only sequential seems to be impacted.
Looking at it that makes sense! I appreciate and thank you and USAFRet for help. I was worried something was off, but ended up learning more about PC parts than I knew before! Actually upgrading a majority of my PC parts has been quite the learning experience for more of the finer details, previously I had it custom built at a local shop so I didn't really take in much. So thanks again!
 
  • Like
Reactions: thestryker