New OWC Express 4M2 enclosure performance results

Has anyone tested this in RAID 5? Or are you running in RAID 5?

It would appear that each blade gets ~700MB/s of bandwidth.

So, getting the highest performance blades to populate this doesn't seem to make a lot of sense.

So, I could install 4 x Intel 660p which tops out @ "up to" 1800 MB/s both reading & writing and still have plenty of headroom per drive?

Intel product page: https://www.intel.com/content/www/us/en/products/memory-storage/solid-state-drives/consumer-ssds/660p-series.html

I was considering the Corsair MP510 series ( https://www.corsair.com/us/en/Categories/Products/Storage/M-2-SSDs/Force-Series-MP510/p/CSSD-F1920GBMP510) which are quite a bit faster (up to 3480 Read/2700 Write) than the Intels, but from what I'm reading, it wouldn't really make a lot of difference.

Am I drawing the proper conclusion from the result presented?

Thanks.

Cheers,
Jon

Yes it looks like you are correct. One extra is RAID 4 is faster for these configurations.

Yes it looks like you are correct. One extra is RAID 4 is faster for these configurations.

Do IOPS figure into any performance calculation in our discussion?

I've always assumed more is better, but given the limitations of the enclosure, I'm not sure it matters.

Thanks.

Cheers,
Jon

IOPS is not something I have paid attention to, as there are not many apps for Mac OS X where that is critical. Happy to see any user posts on I/O's per second information.

Hi, speed tests via Aja attached. Shouldn't the Thunderbolt 3 (MBP 4,3) speeds be faster? We were hoping to get closer to 2500 MBps.

- macOS Mojave 10.14.2
- OWC Express M42
- 4x Samsung 970 EVO 256GB SSD
- Dedicated bus
- SoftRAID configured RAID 0

- Mac Pro 6,1 (Thunderbolt 2 - see attached)
- Macbook Pro 4,3 (Thunderbolt 3 - see attached)...

Hi, speed tests via Aja attached. Shouldn't the Thunderbolt 3 (MBP 4,3) speeds be faster? We were hoping to get closer to 2500 MBps.

- macOS Mojave 10.14.2
- OWC Express M42
- 4x Samsung 970 EVO 256GB SSD
- Dedicated bus
- SoftRAID configured RAID 0

- Mac Pro 6,1 (Thunderbolt 2 - see attached)
- Macbook Pro 4,3 (Thunderbolt 3 - see attached)...

Hi, speed tests via Aja. Shouldn't the Thunderbolt 3 (MBP 4,3) speeds be faster? We were hoping to get closer to 2500 MBps write speeds.

- macOS Mojave 10.14.2
- OWC Express M42
- 4x Samsung 970 EVO 256GB SSD
- Dedicated bus
- SoftRAID configured RAID 0

- Mac Pro 6,1 (Thunderbolt 2 = 1100 MBps write / 1300 MBps read
- Macbook Pro 4,3 (Thunderbolt 3 = 1600 MBps write / 2000 MBps read

Hi!

Haven't had a chance to test it with your requested method – but I will put what I have for now.

I am using with MacBookPro 13,2 (Thunderbolt 3) and 3 x WD Black™ NVMe™ SSD 1TB. The results when it comes to speed are a little bit disappointing. Not sure what I will get with 4th. WD Black.

I will get back soon with results AJA results.

Hi!

I've managed to do some new test with OWC Express 4M2 with 3x WD Black™ NVMe™ SSD 1TB.

Here are the results from AJA Lite.
Read – 2111 MB/s
Write – 1771 MB/s

It seems that 4M2 might reach 2800 MB/s with four of WD Blacks.

Greetings from Warsaw!

Getting maximum throughput on a box designed for generic NVMe blades is complicated by a lot of factors. In the OWC Thunderblade, which pretty much maximizes thunderbolt 3 throughput, everything was tuned to maximize performance. You cannot "daisy chain" the Thunderblade to another device, as that will drop performance dramatically.

Are you sure you are testing with a Thunderbolt 3 40Ghz cable?

I am pasting four results from the M2/4 Express with 4 512GB Samsung 970 Pro NVMe blades.

1. Thunderbolt 3, 40Mhz .5 meter cable
2. Thunderbolt 3, 20Mhz 1 meter cable (all 1m cables are 20Mhz)
3. Thunderbolt dock, using Thunderbolt 3, .40Mhz 5 meter cables
4. USB C .5 meter cable

The differences in performance are just from the cable.

Getting maximum throughput on a box designed for generic NVMe blades is complicated by a lot of factors. In the OWC Thunderblade, which pretty much maximizes thunderbolt 3 throughput, everything was tuned to maximize performance. You cannot "daisy chain" the Thunderblade to another device, as that will drop performance dramatically.

Are you sure you are testing with a Thunderbolt 3 40Ghz cable?

I am pasting four results from the M2/4 Express with 4 512GB Samsung 970 Pro NVMe blades.

1. Thunderbolt 3, 40Mhz .5 meter cable
2. Thunderbolt 3, 20Mhz 1 meter cable (all 1m cables are 20Mhz)
3. Thunderbolt dock, using Thunderbolt 3, .40Mhz 5 meter cables
4. USB C .5 meter cable

The differences in performance are just from the cable.

Hmm. Not seeing those speeds with the 256GB EVO’s. We are using the supplied cable

My test unit has the 970 EVO Pro's, perhaps that is the difference.

Either that, or the 2018 MacMini is much faster over Thunderbolt 3.

The posted results show what you can expect with the enclosure, though. Now it is figuring what the bottleneck is in your setup.

1. Thunderbolt 3, 40Mhz .5 meter cable
2. Thunderbolt 3, 20Mhz 1 meter cable (all 1m cables are 20Mhz)
3. Thunderbolt dock, using Thunderbolt 3, .40Mhz 5 meter cables
4. USB C .5 meter cable

This is interesting.
Thank you.

 

I was just wondering if you have the same issue with OWC Express 4M2 – I mean the fan of the enclosure is always working – even if the enclosure is not attached via Thunderbolt to any computer. And the only way to turn the enclosure off is to take out the power cable from it. Is it right? It was supposed to be that way?

No.

When the system sleeps, after about 30 seconds to a minute, the fan should turn off. It does not require detaching the Thunderbolt cable, or power cable.

I tested this on a 2018 Mini just now.