How do you know if an SSD is "DRAM-Less"?

Format _C:

2[H]4U
Joined
Jun 12, 2001
Messages
3,885
I bought a cheap Team 1TB SSD and to put in perspective this appears as a tad bit faster then a 7.2K RPM 2.5" Spinner so needless to say I want to replace it.

Since it is so slow I'd assume this is DRAM-Less?

I want to know if this Samsung 870 EVO or This Samsung 870 QVO
drive has a DRAM Cache

Also why on the subject of cache-less Vs cached SSDs how can you tell this? Because they all say write "up to xxx MB/S as well as a different number for read speed

In my Amazon SSD order history I have the following drives and if someone doesn't mind can they tell me if they are cached or cache less

# 1 Western Digital 500GB WD Blue SN570
# 2 fanxiang S101
# 3 Crucial P2 1TB 3D NAND NVMe PCIe
# 4 Kingston SA400S37/240G
# 5 Samsung 860 EVO 500GB (MZ-76E500B/AM)

Thank You in advance LoL! :)
 
Last edited:
As an Amazon Associate, HardForum may earn from qualifying purchases.
Mostly by doing some research and reading reviews/ Watching video reviews. And hope that the manufacturer has not made changes to the product since those reviews.

You could make it easier for people to answer your questions quickly by putting the product name in your list instead of just numbering them 1-5.

Lastly, there is almost no way to know from the listing as they don't want to advertise when they don't have it, but if you look at there product line stay with the top tear as stuff below probaly doesn't have DRAM.
 
You could make it easier for people to answer your questions quickly by putting the product name in your list instead of just numbering them 1-5.
My bad I just fixed that 1 - 5 now has the make and model next to the numbet
 
Not clicking through a bunch of links for which you couldn't bother to type out the product names.

I know that the Samsung 870 Evo and Crucial MX500 have a DRAM cache and TLC NAND. The 870 QVO uses QLC NAND, so should be avoided for an OS/app volume.

Is your system not recent enough to utilize a m.2 NVMe SSD for OS/apps?
 
I'm guessing a drive benchmarking tool might tell you.
In my case I only have the hdparm suite (Linux), which gives you 2 speed figures - one is "timing cached reads", the other is "timing buffered disk reads".
I remember Windows also having tools that distinguish between these two. "Burst" and "sustained" IIRC is one way to call them.

I'm not sure if this is a good way to do it - if anyone of you guys has a DRAM-less drive and Linux, please check your hdparm -tT numbers.
For reference, both my nvme SSD and my SATA SSD are like 12 000 MB/s "cached", ~450 MB/s "buffered". These have DRAM.

If this is a red herring, lemme know so I can edit this post.
 
Not clicking through a bunch of links for which you couldn't bother to type out the product names.
I fixed that
is your system not recent enough to utilize a m.2 NVMe SSD for OS/apps?
The system is my oldie but goodie Dell Precision M4800

so SATA only unless a MiniPCIe to M.2 exists that can be bootable as well?

I was given basically a base model M4800 I upped it to a 15.6 FHD Vs the OEM 720P one which was a ROYAL PITA as I Also had to swap the LVDS video cable
I then got 32GB of DDR3 ram for it
 
You say that the Team SSD is only a little faster than a hard drive. Are you seeing this in real life usage, and if so, doing what? Or is this a benchmark number of some sort?

I'd expect the Team unit to handily beat any hard drive, except maybe doing a very large write (as in gigabytes).
 
Also why on the subject of cache-less Vs cached SSDs how can you tell this?

I've got an Inland SSD that's just like this. I found out it was DRAM less looking at reviews once its performance really pissed me off. In benchmarks, I can get good numbers if the concurrency is really ramped up, but sequential is garbage. Works fine as a temporary os drive for random stuff though.
 
In short, if it's really cheap (price wise), it's an SLC cache :)
^^THIS^^

When comparing seemingly similar drive models, the cheaper of the bunch will be the dram-less models, since the budget-cruncher crowd is who they are aimed at, whereas the higher spec'd & higher-priced models usually include a sizable amount of dram, which obviously costs the mfgr moar to make, and that will always be reflected in the selling price.

The other thing to consider is that the dram-less models will use cheaper, older controller chips too, which further reduces not only their build costs as well as their retail prices too, whereas the higher-tier units will have the more recent & higher performing controllers...
 
You say that the Team SSD is only a little faster than a hard drive. Are you seeing this in real life usage, and if so, doing what? Or is this a benchmark number of some sort?

I'd expect the Team unit to handily beat any hard drive, except maybe doing a very large write (as in gigabytes).
Real life usage
Windows 10 loads in under a minute so that is pretty fast but still seems slow for doing things like changing files or seeking through long and/or large video files with ZoomPlayer

I do have some Crystal Disk Mark benchmarks and the numbers don't seem to be that bad but for larger operations as the mouse pointer can freeze for a few seconds sometimes!

So I'm probably going to just get a Samsung 870 EVO and call it a day LoL.



CrystalDiskMark_TEAM_SSD_03-13-2023-2-00PM (Real World Performance Mode)CrystalDiskMark-TEAM_SSD_03-13-2023-2-19PM (Default + Mixed Mode)
 
Flash (SSD and NVMe beside Intel Optane) are organized in large blocks (several MB) build from pages. As a delete action can only be applied to blocks not pages or single datacells, a data update means read block, erase block, write updated block. This mean effective write is much larger than raw data write (affects endurance) and that writing to a non empty/ secure deleted SSD is quite slow especially on steady write and mixed read/write loads.

To improve or uphold performance you can use trim and a dram write cache. Main problem then is a possible data loss/corrupted filesystem on a crash during write or firmware initiated background garbage collection. So your options are:

- Intel Optane (can adress single datacells, not affected by this problem, very expensive)
- slower SSD or without Dram cache or Dram cache protection (desktop SSD like the Evo)
- fast SSD with Dram cache and powerloss protection ("enterprise ssd", more expensive):

The risk of SSDs without plp is not huge but existent. Beside ZFS Raid that can repair some problems based on checksum errors, plp is mandatory for databases or ZFS Slogs and in general a good idea for all non-Raid SSD arrays.
 
Last edited:
For awhile now if a client wanted to go QLC or dram-leas NVME/Sata SSD I always advised get more ram like 32gb and purchase a license to primo cache to offset the performance iv tested retested and found it keeps the NVME drives above 1200mbs and sata at close to 1100mbs during my testing.
 
I bought a cheap Team 1TB SSD and to put in perspective this appears as a tad bit faster then a 7.2K RPM 2.5" Spinner so needless to say I want to replace it.

Since it is so slow I'd assume this is DRAM-Less?
*snip

I found this a while back, might be useful and I hope it can be continually updated. Sorry I can't recall where I got it, no intention of taking away the original author(s)' hard work put into it.

 
Last edited:
As an Amazon Associate, HardForum may earn from qualifying purchases.
Nice table, would be nice if you could find the link for it, as I would like to download it into a spreadsheet for future reference :)

but as usual...

"cheap is as cheap does"

"You only get what you pay for"

"coulda, shoulda, woulda"

So if you're happy with budget-level performance, then stay with what you have.

Otherwise, next time, buy a better drive like the WD SN850X or Sammy 990Pro & be happierz :D
 
*snip

I found this a while back, might be useful and I hope it can be continually updated. Sorry I can't recall where I got it, no intention of taking away the original author(s)' hard work put into it.


Thank You my Team "EX2" is DRAM-Less as well as being just a standard 2.5" SATA III SSD so this explains the lackluster performance
I should have sprung for the Samsung EVO SSD first I will in the future and this will become a storage drive

I learned my lesson research parts before I buy them not after I buy them LoL!
 
Nice table, would be nice if you could find the link for it, as I would like to download it into a spreadsheet for future reference :)

but as usual...

"cheap is as cheap does"

"You only get what you pay for"

"coulda, shoulda, woulda"

So if you're happy with budget-level performance, then stay with what you have.

Otherwise, next time, buy a better drive like the WD SN850X or Sammy 990Pro & be happierz :D
It's an iframe, you can get to the actual spreadsheet from there and print, export, etc.
 
*snip

I found this a while back, might be useful and I hope it can be continually updated. Sorry I can't recall where I got it, no intention of taking away the original author(s)' hard work put into it.


Is this the public Google Sheet somebody released a while back? I used to have it on my google sheets but it's gone.
 
Is this the public Google Sheet somebody released a while back? I used to have it on my google sheets but it's gone.
I am not able to recall where I got it as I'd like to give credit to the author or the website I got it from. Link was in my notes where it sat until I saw this thread.
Hopefully someone will bring more info.

*edit:
Okay, kept googling until I found it, saving it now on my notes lol:
https://www.reddit.com/r/NewMaxx/comments/dhvrdm/ssd_guides_resources/
 
Last edited:
Iv purposely bought QLC+DRAM -less SSDs to do my testing as iv shown before these drives are the Kingston NV2 1TB drives.
Been testing with Huge HDDS as well that one gets a dedicated Nvme as L2 Cache and some dram from the system for builds.
 

Attachments

  • C uncached.png
    C uncached.png
    24 KB · Views: 0
  • Cached.png
    Cached.png
    25 KB · Views: 0
Iv purposely bought QLC+DRAM -less SSDs to do my testing as iv shown before these drives are the Kingston NV2 1TB drives.
Been testing with Huge HDDS as well that one gets a dedicated Nvme as L2 Cache and some dram from the system for builds.
I wish I could get storage that fast in this old lappy (Dell Precision M4800) it is SATA (SATA III or 6Gbps) only but I wonder does a MPCIe to M.2 adapter exist I know I probably won't be able to boot from it though so that mostly defeats the point of having a fast drive.
 
I wish I could get storage that fast in this old lappy (Dell Precision M4800) it is SATA (SATA III or 6Gbps) only but I wonder does a MPCIe to M.2 adapter exist I know I probably won't be able to boot from it though so that mostly defeats the point of having a fast drive.
While great at benchmarks, you may find that in "real use", a SATA SSD isn't all that different from a modern NVMe, especially given the other technology restrictions (older) of your platform.
 
For awhile now if a client wanted to go QLC or dram-leas NVME/Sata SSD I always advised get more ram like 32gb and purchase a license to primo cache to offset the performance iv tested retested and found it keeps the NVME drives above 1200mbs and sata at close to 1100mbs during my testing.
I am willing to bet though most of those clients are not doing anything intense enough they would notice the performance on a DRAM-Less SSD anyways? And if they are, and can afford 32GB of ram, why not just spend a little more for an SSD with DRAM and not worry about 3rd party software at all.
 
While great at benchmarks, you may find that in "real use", a SATA SSD isn't all that different from a modern NVMe, especially given the other technology restrictions (older) of your platform.
Very much so, the access times of an SSD even on SATA would be night and day from spinning rust drive.
 
I thought the platters were glass in modern (as in made in the past 20 years or so) hard drives and glass can't rust so spinning rust is a misnomer LoL!
Glass can't hold a magnetic charge. But iron oxide sure can (but only some forms, brown rust may not be?). I'm specifically not claiming that hard drives use iron oxide in the coating of the platters, but they conceptually could.
 
Back
Top