Photofast G-Monster V4 ZIF SSD User Review

[BASpro]

Hello Nando4:

I would like to compare some of the drive details between my production unit and your pre-production unit. I executed an 'hdparm -I /dev/hda' on my drive and posted the results below. Could you do the same on your preproduction unit? In particular, I wanted to compare firmware revision and DMA cycle time between the two units. 120ns is UDMA-2 mode timing and the write results I get are bang on with UDMA-2 speed - I'm pulling at straws here. Thanks.

ATA device, with non-removable media
Model Number: GMonster-50IDE V4 SSD 128GB
Serial Number: ************
Firmware Revision: VATP1281
Transport: Parallel, ATA8-APT
Standards:
Used: unknown (minor revision code 0x0039)
Supported: 8 7 6 5
Likely used: 8
Configuration:
Logical max current
cylinders 16383 16383
heads 16 16
sectors/track 63 63
--
CHS current addressable sectors: 16514064
LBA user addressable sectors: 234881024
LBA48 user addressable sectors: 234881024
device size with M = 1024*1024: 114688 MBytes
device size with M = 1000*1000: 120259 MBytes (120 GB)
Capabilities:
LBA, IORDY(cannot be disabled)
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
DMA: sdma0 sdma1 sdma2 mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5 udma6
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* SMART feature set
Security Mode feature set
* Power Management feature set
* Write cache
Look-ahead
* Host Protected Area feature set
* WRITE_BUFFER command
* READ_BUFFER command
* NOP cmd
* DOWNLOAD_MICROCODE
SET_MAX security extension
* 48-bit Address feature set
* Mandatory FLUSH_CACHE
* FLUSH_CACHE_EXT
Security:
Master password revision code = ******
supported
not enabled
not locked
frozen
not expired: security count
supported: enhanced erase
2min for SECURITY ERASE UNIT. 2min for ENHANCED SECURITY ERASE UNIT.
HW reset results:
CBLID- above Vih
Device num = 0 determined by the jumper
Checksum: correct

All the best, Mass

[nando4]

Originally Posted by BASpro

Hello Nando4:

I would like to compare some of the drive details between my production unit and your pre-production unit. I executed an 'hdparm -I /dev/hda' on my drive and posted the results below. Could you do the same on your preproduction unit? In particular, I wanted to compare firmware revision and DMA cycle time between the two units. 120ns is UDMA-2 mode timing and the write results I get are bang on with UDMA-2 speed - I'm pulling at straws here. Thanks.

I have firmware VATE1273. Note too mine has 32nm Micron flash as shown in the internals photo. I too got writes at one point down to 30MB/s after heaps of reimaging and benchmarking and needed to do the AS-Cleaner to correct. All that Tony Trim does is consolidates free space, then AS-Cleaner writes a file with 0xFF (nulls) that fill the empty space to effectively do a manual GC/Trim. I'd also advise ensuring you don't have any other devices on the same PATA channel to ensure your seeing full PATA performance during testing.

My hdparm output below:

# hdparm -i /dev/sda | cut -c1-76

/dev/sda:

Model=GMonster-18ZIF V4 SSD 128GB , FwRev=VATE1273, SerialNo=20
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=0
BuffType=DualPortCache, BuffSize=16384kB, MaxMultSect=16, MultSect=?16?
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=234881024
IORDY=yes, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: sdma0 sdma1 sdma2 mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5 udma6
AdvancedPM=no WriteCache=enabled
Drive conforms to: unknown: ATA/ATAPI-1,2,3,4,5,6,7

* signifies the current active mode

[BASpro]

Originally Posted by nando4

I too got writes at one point down to 30MB/s after heaps of reimaging and benchmarking and needed to do the AS-Cleaner to correct. All that Tony Trim does is consolidates free space, then AS-Cleaner writes a file with 0xFF (nulls) that fill the empty space to effectively do a manual GC/Trim. I'd also advise ensuring you don't have any other devices on the same PATA channel to ensure your seeing full PATA performance during testing.

Thanks. There's only one IDE device installed.

My last straw is As-cleaner; although, write speeds greater than 34MB/s were never achieved from the first time I installed the drive. If As-cleaner doesn't improve the speeds, then I'll simply accept that there is a limitation in my computer chipset that is preventing full speed. It's still a great product and glad I purchased it.

I tried As-clearner and it will not start up - "The application failed to initialize..." error pops up. Do you know what is preventing it from starting?

Thanks. All the best, Mass

[nando4]

Originally Posted by BASpro

I tried As-clearner and it will not start up - "The application failed to initialize..." error pops up. Do you know what is preventing it from starting?

Not worth figuring out what the problem is. Can do the same with the workaround below.

Workaround manual GC

1. Run PerfectDisk, do "Consolidate Disk Space"
2. Download nulls.bin and copy to c:\test. nulls.bin is a 1k file of 0xFF (nulls)
3. Save code below as c:\test\cleaner.bat and run c:\test\cleaner.bat

cleaner.bat

@echo off
: Backup nulls.bin
set a=0
copy nulls.bin junk%a%.bin

:loop
:: Repeat,  doubling the size of the NULLs file on each loop until run out of
:: space. Effectively appends NULLS to the remaining free space on the SSD.

set /a b=%a% + 1
copy junk%a%.bin + junk%a%.bin junk%b%.bin /b
set /a a=%b% 
goto loop

4. It will eventually fail with "not enough disk space" type error, upon which repeat steps 2-3 in directory "c:\test2" to fill remaining space.
5. Once complete delete c:\test and c:\test2 directories.

[BASpro]

Originally Posted by nando4

Not worth figuring out what the problem is. Can do the same with the workaround below.

Workaround manual GC
...

Nando4:

Thanks for the GC. I haven't executed the clean yet, because it takes a long time; I'll post my results once I complete it.

I had one more question: I like to set up my drive exactly as you set up yours - for testing the speed. Could you let us know how many partitions are set up, their size for each, the offset starting point for each, and the allocation unit size for each partition. Thanks.

All the best, Mass

[nando4]

Originally Posted by BASpro

I had one more question: I like to set up my drive exactly as you set up yours - for testing the speed. Could you let us know how many partitions are set up, their size for each, the offset starting point for each, and the allocation unit size for each partition.

You are certainly on the right track for why you may be seeing lower write performance. If your partitions are not NAND page aligned, the SSD controller's writes will occur across two NAND pages for each logical 4KB block used on your partition, quite likely halving your write performance as you are seeing. See my G-Monster V4 ZIF drive/partition info for my setup. Notable points:

- NAND page size is 4KB
- erase block size is 512KB (for Runcore ProIV)
- ntfs partition using default 4kb allocation

Since I used the same image to test both SSDs, I ensured the partitions were aligned to a 512KB boundary (divides into 4KB). If you do a fresh Win7 install, it will automatically align to a SSD-aware 1MB boundary so need to mess around with partition offsets.

[BASpro]

Originally Posted by nando4

... If your partitions are not NAND page aligned, the SSD controller's writes will occur across two NAND pages for each logical 4KB block used on your partition ....

Hello Nando4:

OK, here are my new results (small changes in speed):

--------------------------------------------------
CrystalDiskMark 2.2 (C) 2007-2008 hiyohiyo
Crystal Dew World : http://crystalmark.info/
--------------------------------------------------

Sequential Read : 85.043 MB/s
Sequential Write : 38.246 MB/s
Random Read 512KB : 77.890 MB/s
Random Write 512KB : 26.593 MB/s
Random Read 4KB : 11.180 MB/s
Random Write 4KB : 2.814 MB/s

Test Size : 100 MB
Date : 2010/01/26 23:29:08


Below are the partitions that are set up. I was unable to exactly match the starting offset of the second partition, DISKPART would simply not implement the offset (6,923,776k) that I tried to enter - the OS added the additional 524,288 bytes to the value shown; nevertheless, the partitions are all aligned:

Partition Disk #0, Partition #0 (4kb allocation)
Partition Size 6.60 GB (7,089,422,336 bytes)
Partition Starting Offset 524,288 bytes

Partition Disk #0, Partition #1 (4kb allocation)
Partition Size 39.06 GB (41,943,040,000 bytes)
Partition Starting Offset 7,090,470,912 bytes


Here are the specs of my drive:

/dev/hda:

Model=GMonster-50IDE V4 SSD 128GB, FwRev=VATP1281, SerialNo=2009
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=0
BuffType=DualPortCache, BuffSize=0kB, MaxMultSect=16, MultSect=off
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=234881024
IORDY=yes, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: sdma0 sdma1 sdma2 mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5 udma6
AdvancedPM=no WriteCache=enabled
Drive conforms to: unknown: ATA/ATAPI-1 ATA/ATAPI-2 ATA/ATAPI-3 ATA/ATAPI-4 AT
A/ATAPI-5 ATA/ATAPI-6 ATA/ATAPI-7

* signifies the current active mode


It seems this is the best that can be had. I will confirm with the manufacturer the significance of 'BuffSize=0kB' and 'MultSect=off' as shown in my drive specs above - these values show differently on your drive specs.


Nando4: thanks for all your hard work!

All the best, Mass

[nando4]

Originally Posted by BASpro

It seems this is the best that can be had. I will confirm with the manufacturer the significance of 'BuffSize=0kB' and 'MultSect=off' as shown in my drive specs above - these values show differently on your drive specs.


Nando4: thanks for all your hard work!

Alignment + manual GC has improved your writes a little. Still nowhere near 60-90MB/s as advertised nor matching my result of 73MB/s at UDMA5 speeds. Your read speed is quite good.

Good point noting the BuffSize=0kb. That would indicate to me that your 64MB onboard cache, used for write caching, is not recognised and perhaps not working. The likely reason for the slower write performance. You could see from my hdparm output that I had 16MB cache active on the pre-release product. The multisect value likely nothing to be concerned about.. my Runcore too has it set to 1 as shown here, but it has 32MB cache active.

FYI: netbooknews got 80MB/s sequential writes with their G-Monster V4 IDE on a UDMA6 interface here.

[BASpro]

Originally Posted by nando4

... noting the BuffSize=0kb...

The manufacturer confirms that the BuffSize displays as 0kb because the production unit was configured to not display this information to the ATA interface - hence appearing as if it doesn't exist. Hmmmm, if the ATA interface believes a cache doesn't exist, will it still attempt to burst write?

Next steps:
1. Test the drive on my workstation - it will take a few days to get an interface converter; I'll post my results when completed.
2. Do you know of any bit, registry or driver setting that impacts IDE write caching/bursting?

Thanks.
All the best, Mass

[nando4]

Originally Posted by BASpro

Next steps:
1. Test the drive on my workstation - it will take a few days to get an interface converter; I'll post my results when completed.
2. Do you know of any bit, registry or driver setting that impacts IDE write caching/bursting?

Oh.. there is one thing that could be the culprint. Ensure you disable power savings mode. A 2510P users reported 31MB/s writes here with a OCZ Vertex with powersavings on, 65MB/s writes with powersaving off. I did the original performance tests running Rmclock locked to the highest multiplier and XP Power setting on "Minimal Power Management".

Final random guess: could disable then enable the IDE port as shown here, but do not do the Scan New Hardware.