Quad NVMe card and Windows 10 Striping

[TeleFragger]

Ok so I had a dell t5810 (got off of craigslist) and had questions over the last few years.. upgrade cpu, etc.. but then started acting up and it sat dead... replaced mobo and PSU and it still randomly shuts off or reboots..
whatever..

moving on..

I got a Lenovo P710 off of craigslist and moved my cpu (e5-2678 v3) and bought a 2nd and also moved over my 160gb ddr4 memory, zotac gtx 1060 and game warzone kinda ok.
Since I had that quad NVMe card in my old Dell, I didnt have any storage other than a few SSD's...
the other day I was looking at that Dell card and after google, gave it a try and boom
the Dell Card works in my Lenovo!!!!!!

I was all happy of my self.. hah.. I have all 4 nvme drives as a stripe set in Windows 10. (win11 unsupported for my cpu's ugh)

I was benchmarking my drives with Anvil Benchmark and the numbers between my 500gb ssd and nvme (even though striped it said 1 card) were about the same.
So came up here and started searching and while i didnt find what I was looking for... i found this thread
https://hardforum.com/threads/anyon...erformance-at-high-4tbish-capacities.2020479/

big point here is that I am looking at the benchmarks that The_Heretic and Zepher put up and my numbers are not good at all..


so my question.. I know long winded.. I am sorry..

Is it bad to stripe 512gb nvme drives together as in performance? I had 512gb ssd's separate and what a pain that is...

Crucial MX500 SSD

NVMe

Side note - I like Anvil's Benchmark

thanks for the insight..

 

[Zepher]

Can you move the data off the stripe and make one of the NVME drives it's own volume and see what it benches by itself on that controller in CrystalDiskMark?

 

[TeleFragger]

I like how you think... I lack that common sense!!! hah.. will give it a try later.. as I have to go pick my kid up from mma in a bit.

 

[TeleFragger]

All I can find out about this card
DELL JV6C8 Ultra Speed Drive Quad X16 Pcie To M.2 Adapter

Seems that the top 4 numbers are faster, but the bottom 4 numbers are not...
I do not know what these numbers mean...

WDC-SN730

Samsung-PM961

Samsung-PM981

 

[TeleFragger]

looks like to me it is worth striping them...
guess I will restripe them! thx

 

[mwroobel]

Is the slot you are plugging the x4 m.2 card into x16/x16? If no, you will likely get better than single drive but less than four drive. All of this said, while synthetic benchmarks will increase you will find your real-world response almost identical to a single drive. I have tried various stripes with SATA600, SATA M.2 and NVME and while the synthetics can be great, actual differences in startup time, loading times and other perceived tasks gave approximately 0%-20% improvements versus investment in the x4 card, the additional drives, the additional complexity, the fact that a single failure blows your drive, etc.

 

[TeleFragger]

Is the slot you are plugging the x4 m.2 card into x16/x16? If no, you will likely get better than single drive but less than four drive. All of this said, while synthetic benchmarks will increase you will find your real-world response almost identical to a single drive. I have tried various stripes with SATA600, SATA M.2 and NVME and while the synthetics can be great, actual differences in startup time, loading times and other perceived tasks gave approximately 0%-20% improvements versus investment in the x4 card, the additional drives, the additional complexity, the fact that a single failure blows your drive, etc.

in bios i had to set card to x4x4x4x4 but is in a pcie16x slot

on the costs..
i had all of this on hand as my dell died and had the nvme card in it with drives in it... so in theory, i was going to sell it but found a use for it...

for the failure, i dont keep anything on it that I can lose so I understand that.

 

[Denpepe]

Well I'm not realy a specialist on these things, but your drives are an odd bunch, I always tought it was best to use mostly the same drives for best results.

I do know that raiding a bunch of SSD's together usually adds some latency but that should not matter much for bulk storage.

Seems that the top 4 numbers are faster, but the bottom 4 numbers are not...

Also, I don't know what you mean by this, the raid numbers in the first post are a decent chunk faster then the individual drives you posted later

 

[Beaflag VonRathburg]

Well I'm not realy a specialist on these things, but your drives are an odd bunch, I always tought it was best to use mostly the same drives for best results.

I do know that raiding a bunch of SSD's together usually adds some latency but that should not matter much for bulk storage.

Neither am I, but... Generally best practice is to use the same drives while having them all on the same firmware. Mixing; different drives, models, manufacturers, firmware, and then adding Windows raid system on top of it is asking for issues. It might work great for years, but it's not the best idea. Depending on what you're storing on the drives and if / how you implement backups this maybe great for you OP, but that's something you'd need to decide.