I/O CREST w/2x970Pro's WAY SLOW??

Bugs713

macrumors newbie
Original poster
Sep 3, 2018
23
14
Apache Junction, AZ USA
* I/O Crest SI-PEX40129 controller card
* Slot 2
* Two m.2 970 Pro 512MB
* Format RAID/0 HFS boot drive
* 2010 cMP 5,1/12 core 3.46 GHz,
* Boot ROM 140
* High Sierra 10.13.6(17G5019)

Read/Write speeds are only 380/390MBs and should be in the 2000MBs. Any idea what I did wrong? I also tried a single 970EVO 1TB and ran AJA with same low speed results.

Had this same set up in a High Point 7101A with warp speeds in the 5000/6000MBs range.

Thanks!

 
Last edited:

bsbeamer

macrumors 68030
Sep 19, 2012
2,621
1,281
Would be helpful to install DriveDx and post screenshots of the report. Also include firmware version(s) for blade(s).
 

handheldgames

macrumors 68000
Apr 4, 2009
1,706
930
Pacific NW, USA
* I/O Crest SI-PEX40129 controller card
* Slot 2
* Two m.2 970 Pro 512MB
* Format RAID/0 HFS boot drive
* 2010 cMP 5,1/12 core 3.46 GHz,
* Boot ROM 140
* High Sierra 10.13.6(17G5019)

Read/Write speeds are only 380/390MBs and should be in the 2000MBs. Any idea what I did wrong? I also tried a single 970EVO 1TB and ran AJA with same low speed results.

Had this same set up in a High Point 7101A with warp speeds in the 5000/6000MBs range.

Thanks!
The I/O Crest SI-PEX40129 controller card has a 16 lane electrical connection
The board uses an Asmedia 8x PCIe 3.0 PLX chipset, splitting the 8x connection across 2 4x m.2 slots.
In the cMP, the Asmedia chipset runs at 8x, PCIe 2.0,
8 lanes of PCIe 2.0 can deliver a maximum transfer speed of around 3000 MB/s.


The Highpoint SSD-7101 controller card has a 16 lane electrical connection
The board uses an Broadcom 16x PCIe 3.0 PLX chipset, splitting the 16x connection across 4 4x m.2 slots.
In the cMP, the Broadcom chipset runs at 16x, PCIe 2.0
16 lanes of PCIe 2.0 can deliver a maximum transfer speed of around 6000 MB/s.

Your slow speeds sounds wrong and I agree that your system is running toooo slow.

Have you tried the drives one at a time before you built the raid?
Have you tried to rebuild the raid to see if that corrects the issue?
What block size are you running at? I've tried multiple configurations with the Highpoint SSD 7101a, and never experienced a slowdown like you brought up.
Have you run disk utility / first aid?
 
Last edited:

Bugs713

macrumors newbie
Original poster
Sep 3, 2018
23
14
Apache Junction, AZ USA
The I/O Crest SI-PEX40129 controller card has a 16 lane electrical connection
The board uses an Asmedia 8x PCIe 3.0 PLX chipset, splitting the 8x connection across 2 4x m.2 slots.
In the cMP, the Asmedia chipset runs at 8x, PCIe 2.0,
8 lanes of PCIe 2.0 can deliver a maximum transfer speed of around 3000 MB/s.


The Highpoint SSD-7101 controller card has a 16 lane electrical connection
The board uses an Broadcom 16x PCIe 3.0 PLX chipset, splitting the 16x connection across 4 4x m.2 slots.
In the cMP, the Broadcom chipset runs at 16x, PCIe 2.0
16 lanes of PCIe 2.0 can deliver a maximum transfer speed of around 6000 MB/s.

Your slow speeds sounds wrong and I agree that your system is running toooo slow.

Have you tried the drives one at a time before you built the raid?
Have you tried to rebuild the raid to see if that corrects the issue?
What block size are you running at? I've tried multiple configurations with the Highpoint SSD 7101a, and never experienced a slowdown like you brought up.
Have you run disk utility / first aid?
I did try a 970EVO 1tb before I set up the RAID/0 and seen similar low speed.
I did reformat and used APFS instead of HFS, again in Raid/0, same problem.
Block size (chunk size?) Was set at 128mb.

For whatever reason I did not run First Aid, I will be back in town today and will try that and post results.
Thank you!
 

Bugs713

macrumors newbie
Original poster
Sep 3, 2018
23
14
Apache Junction, AZ USA
Installed a new 970EVO 1TB m.2 after formatting it through my iMac (HFS) and still same slow speeds.
Also ran First Aid on the RAID/0 config with no change in speed...
 

Bugs713

macrumors newbie
Original poster
Sep 3, 2018
23
14
Apache Junction, AZ USA
Again, what does DriveDx report show?
https://binaryfruit.com/drivedx
Sorry...

This is a CCC cloned 970EVO made from the RAID drive~


### SYSTEM INFORMATION ###
Report Timestamp : March 19, 2019 5:56:08 PM MST
Report Timestamp (ISO 8601 format) : 2019-03-19T17:56:08

Application Name : DriveDx
Application Version : 1.8.2.610
Application SubBuild : 1
Application Edition : Standalone
Application Website : https://binaryfruit.com/drivedx
DriveDx Knowledge Base Revision : 9/9

Computer Name : BeanMacPro’s Mac Pro
Host Name : BeanMacPros-Mac-Pro
Computer Model : MacPro5,1

OS Boot Time : 2019-03-19T16:58:47
Time Since Boot : 00h 57m 21s
OS Name : macOS
OS Version : 10.13.6
OS Build : 17G5019
OS Kernel Version : Darwin 17.7.0

SAT SMART Driver Version : N/A
ATA Command Support Tolerance : verypermissive
N of drives in report : 1



### DRIVE 1 OF 1 ###
Last Checked : March 19, 2019 5:54:58 PM MST
Last Checked (ISO 8601 format) : 2019-03-19T17:54:58

Advanced SMART Status : OK
Overall Health Rating : GOOD 100%
SSD Lifetime Left Indicator : GOOD 100%
Issues found : 0

Serial Number : S467NX0KA04062N
WWN Id :
Volumes : MaCHD
Device Path : /dev/disk0
Total Capacity : 1.0 TB (1,000,204,886,016 Bytes)
Model Family : Samsung 970 EVO Series SSD
Model : Samsung SSD 970 EVO 1TB
Firmware Version : 2B2QEXE7
Drive Type : SSD

Power On Time : 362 hours (15 days 2 hours)
Power Cycles Count : 251
Current Power Cycle Time : 1.0 hours



=== DEVICE CAPABILITIES ===
S.M.A.R.T. support enabled : yes
DriveDx Active Diagnostic Config : Samsung 970 EVO Series SSD config [ssd.nvme.samsung.970.evo.01]
Sector Logical Size : 512
Sector Physical Size : 512
Physical Interconnect : PCI-Express (PCIe)
Logical Protocol : NVM-Express (NVMe)
Removable : no
Ejectable : no
NVMe Revision Supported : 1.30
PCI Vendor Id : 0x144d
Thermal Throttling Supported : no
Volatile Write Cache Supported : yes
Maximum Data Transfer Size : 512 Pages
Log Page Attributes : 0x3
Format NVM Attributes : 0x5
Optional Admin Command Support : 0x0
Optional NVM Command Support : 0x5f (Compare, Wr_Unc, DS_Mngmt, Wr_Zero, Save/Sel_Feat, +)
Number of Power States Supported : 4
Error Log Page Entries : 63
Keep Alive Support : 0
Total NVMe Capacity : 1,000,204,886,016 (1.0 TB)
Unallocated NVMe Capacity : -
Number of NVMe Namespaces : 1
I/O Path : IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU1@7/IOPP/PXS2@0/IOPP/pci-bridge@0/IOPP/pci144d,a808@0/IONVMeController/IONVMeBlockStorageDevice@1
[Known device ]: yes
[Drive State Flags ]: 0x0


=== CURRENT POWER CYCLE STATISTICS ===
Data Read : 27.9 GB
Data Written : 36.4 GB
Data Read/Write Ratio : 0.77
Average Throughput (Read) : 37.9 MB/s
Average Throughput (Write) : 65.6 MB/s

Operations (Read) : 198,521
Operations (Write) : 323,960
Operations Read/Write Ratio : 1
Throughput per operation (Read) : 147.5 KB/Op
Throughput per operation (Write) : 117.7 KB/Op

Latency Time (Read) : 0 ns
Latency Time (Write) : 0 ns
Retries (Read) : 0
Retries (Write) : 0
Errors (Read) : 0
Errors (Write) : 0


=== PROBLEMS SUMMARY ===
Failed Indicators (life-span / pre-fail) : 0 (0 / 0)
Failing Indicators (life-span / pre-fail) : 0 (0 / 0)
Warnings (life-span / pre-fail) : 0 (0 / 0)
I/O Error Count : 0 (0 / 0)


=== IMPORTANT HEALTH INDICATORS ===
ID NAME RAW VALUE STATUS
7 Data Units Written 12,313,368 (5.7 TB) 100% OK
14 Media and Data Integrity Errors 0 100% OK


=== TEMPERATURE INFORMATION (CELSIUS) ===
Current Temperature : 39
Power Cycle Min Temperature : 35
Power Cycle Max Temperature : 39
Lifetime Min Temperature : 35
Lifetime Max Temperature : 39
Recommended Min Temperature : 5
Recommended Max Temperature : 85
Temperature Min Limit : 5
Temperature Max Limit : 85


=== DRIVE HEALTH INDICATORS ===
ID | NAME | TYPE | UPDATE | RAW VALUE | VALUE | THRESHOLD | WORST | LAST MODIFIED | STATUS
2 Composite Temperature Life-span online 39 (39 °C) 61 15 61 3/19/19 5:54 PM 54.1% OK
3 Available Spare Pre-fail online 100 100 10 100 - 100% OK
5 Life Percentage Used Life-span online 0 100 0 100 - 100% OK
6 Data Units Read Life-span online 4,701,255 (2.2 TB) 100 0 100 3/19/19 5:54 PM 100% OK
7 Data Units Written Life-span online 12,313,368 (5.7 TB) 100 0 100 3/19/19 5:54 PM 100% OK
8 Host Read Commands Life-span online 70,360,874 100 0 100 3/19/19 5:54 PM 100% OK
9 Host Write Commands Life-span online 91,227,049 100 0 100 3/19/19 5:54 PM 100% OK
10 Controller Busy Time Life-span online 149 (149 minutes) 100 0 100 3/19/19 5:54 PM 100% OK
11 Power Cycles Life-span online 251 100 0 100 - 100% OK
12 Power On Hours Life-span online 362 (362 hours) 100 0 100 3/19/19 5:54 PM 100% OK
13 Unsafe Shutdowns Life-span online 119 100 0 100 - 100% OK
14 Media and Data Integrity Errors Pre-fail online 0 100 0 100 - 100% OK
15 Error Information Log Entries Pre-fail online 0 100 0 100 - 100% OK
16 Warning Composite Temperature Time Pre-fail online 0 (0 minutes) 100 0 100 - 100% OK
17 Critical Composite Temperature Time Pre-fail online 0 (0 minutes) 100 0 100 - 100% OK
101 Temperature Sensor 1 Life-span online 39 (39 °C) 61 15 61 3/19/19 5:54 PM 54.1% OK
102 Temperature Sensor 2 Life-span online 53 (53 °C) 47 15 47 3/19/19 5:54 PM 37.6% OK



=== DRIVE ERROR LOG ===
error log is empty
[doublepost=1553043930][/doublepost]The actual RAID/0 (2x970PRO 512GB) drives are in my High Point 7101A card, but seeing the same low speeds no matter which m.2(s) I install in the I/O Crest.
 

Bugs713

macrumors newbie
Original poster
Sep 3, 2018
23
14
Apache Junction, AZ USA
Been having bad luck last few months with bad electronics!?? My first High Point 7101A died after 30 days of use, got a replacement through Amazon. Thought I'd try this card do to lack of faith in HPT and would be happy with almost warp speed that this card would provide, but....

The replacement 7101A is back to being my boot RAID/0 and doing great, still kinda worried, but really feel I got a bad first card as this one behaves differently. I have thought that maybe it is my cMP killing these cards, but just don't believe that at this point. I have worked on computers and build electronics for last 30 years, I know the 'Laws" of safe handling etc, so can't blame myself either, though I gladly would if I was the problem...

Gonna return the I/O Crest for refund and move forward.

Thank you guys for the help!!
 

bsbeamer

macrumors 68030
Sep 19, 2012
2,621
1,281
Do not believe High Sierra required permissions for full disk access for CCC, but double check in Menu > Carbon Copy Cloner to see if permissions need to be granted. Believe this is a new change only in Mojave and can prevent a true clone from completing properly.

Screen Shot 2019-03-22 at 11.44.52 AM.png
 
  • Like
Reactions: Bugs713

handheldgames

macrumors 68000
Apr 4, 2009
1,706
930
Pacific NW, USA
Thanks, will check tonight when home.
CCC has issues with APFS / Mojave / NVMe.

It's an apple bug, CCC just aggravates it.

It happens with other disk apps as well. It's an Apple bug.

The issue only happens on NVMe. SATA / SATA Express is not effected.

Not every system experiences the issue, its based on the loading of the pcie bus.
 
  • Like
Reactions: crjackson2134