SSD with ambiguous transfer rate - what do you think?

Hi! I've been looking for a SSD with PATA interface which does its
best to saturate ATA133 bandwith. What do you think about such an
offer:
http://tinyurl.com/nxuomz
Did anybody any benchmarks for this product, like e.g. CrystalDiskMark
http://tinyurl.com/2r3mqy

The confusion is especially about its ambiguous description of a
transfer rate
"120M read and 80M write speed. Pls. note this is the potential speed
of the disk but in real world, performance will be capped by the
limitation of PATA interface.
Notice: PATA interface, although support 133MB/s by specification, can
never go above 65MB/s in real application. Our disk will run at the
full capacity of PATA, at around 65MB/s. There is no disk can break
that barrier."

Please, also look at their site:
http://www.ssdfactory.com/en/product...5/index-1.html
looks convincing, but the assertion about PATA's limitation already
NOT!
If ATA100 interface limited transfer to 66MB/s so why were these
benchamrk results:
http://www.dvnation.com/benchmarks-Solidata-SSD.html
http://www.dvnation.com/benchmarks-Photofast-SSD.html
(please look at the bottom of page where the results for PATA devices
are put)

Thanks in advance for any useful remarks!

Alojzy Zakalec alojzy.zakalec gmail.com wrote:

Hi! I've been looking for a SSD with PATA interface which does
its best to saturate ATA133 bandwith.

Sounds dubious/fishy.

The confusion is especially about its ambiguous description of a
transfer rate
"120M read and 80M write speed. Pls. note this is the potential
speed of the disk but in real world, performance will be capped
by the limitation of PATA interface.
Notice: PATA interface, although support 133MB/s by
specification, can never go above 65MB/s in real application.
Our disk will run at the full capacity of PATA, at around
65MB/s. There is no disk can break that barrier."

Looks like they are saying "if you want speed, forget it". It might
be OK if you absolutely must have quiet and low-power and if you
cannot do SATA.

I would look for and read reviews on Newegg, too. Good luck.


--
Big front wheel skates (144mm front wheel, four 80mm trailing
wheels). Great for rough street skating.
http://www.flickr.com/photos/27532210@N04/3056505603/

On 30 Lip, 17:32, John Doe wrote:
Looks like they are saying "if you want speed, forget it". It might
be OK if you absolutely must have quiet and low-power and if you
cannot do SATA.

I would look for and read reviews on Newegg, too. Good luck.

But when you look at Newwgg's SSD PATA offers:
http://tinyurl.com/kwt56p
then there are also only brands with poor speed.

If have doubts if in PATA's worlds bigger transfers are possible then
take into consideration the HongKong company Solidata products, e.g.
P-2:
http://www.solidatum.com/EnProduct.a...e=P(2.5%20PATA)

How do you think, can the shop selling their products:
http://www.solidata.hk/onlineshopitems/p2.htm
be fully trusted (to pay money a priori)?

Alojzy Zakalec alojzy.zakalec gmail.com wrote:

How do you think, can the shop selling their products:
http://www.solidata.hk/onlineshopitems/p2.htm
be fully trusted (to pay money a priori)?

The massive amount of spam selling counterfeit products comes from
Hong Kong, pouring into USENET through Google Groups. Hong Kong is
where the counterfeit products you find on eBay are sold from.

Big companies from world democracies have outsourced manufacturing
of their products to the communist Chinese (using their slave
labor pool) to make namebrand products, and now the Chinese have
turned around and are royally screwing those same big companies.
Poetic justice IMO.

Just make sure it's a namebrand product...
(kidding)

Alojzy Zakalec wrote:
Hi! I've been looking for a SSD with PATA interface which does its
best to saturate ATA133 bandwith. What do you think about such an
offer:
http://tinyurl.com/nxuomz
Did anybody any benchmarks for this product, like e.g. CrystalDiskMark
http://tinyurl.com/2r3mqy

The confusion is especially about its ambiguous description of a
transfer rate
"120M read and 80M write speed. Pls. note this is the potential speed
of the disk but in real world, performance will be capped by the
limitation of PATA interface.
Notice: PATA interface, although support 133MB/s by specification, can
never go above 65MB/s in real application. Our disk will run at the
full capacity of PATA, at around 65MB/s. There is no disk can break
that barrier."

Please, also look at their site:
http://www.ssdfactory.com/en/product...5/index-1.html
looks convincing, but the assertion about PATA's limitation already
NOT!
If ATA100 interface limited transfer to 66MB/s so why were these
benchamrk results:
http://www.dvnation.com/benchmarks-Solidata-SSD.html
http://www.dvnation.com/benchmarks-Photofast-SSD.html
(please look at the bottom of page where the results for PATA devices
are put)

Thanks in advance for any useful remarks!

Buy an Intel X25-E SATA SSD, and use a SATA to PATA adapter ?

*******

There are various brands of these, and the SATA connector on this,
plugs directly into a SATA 2.5" or 3.5" device. For SATA 1.8"
you'd likely need a hard-to-find adapter. On the other side
is the IDE ribbon cable interface. I'm not aware of any
web pages that compare the performance of these, so you may
end up buying a few and testing them.

http://www.newegg.com/Product/Produc...82E16812200156

http://www.startech.com/item/IDE2SAT...d-Adapter.aspx

(Pictures...)

http://www.startech.com/Share/Galler...SAT.Alarge.jpg

http://www.startech.com/Share/Galler...SAT.Blarge.jpg

(Power for the adapter and the drive connected to it, is via a
floppy connector.)

http://www.startech.com/Share/Galler...SAT.Clarge.jpg

(Power cable included. This is something to watch for - some products
lack necessary cables, and the cables may be hard to find or
require an additional purchase.)

http://www.startech.com/Share/Galler...SAT.Dlarge.jpg

Paul

On 30 Lip, 21:33, Paul wrote:
There are various brands of these, and the SATA connector on this,
plugs directly into a SATA 2.5" or 3.5" device. For SATA 1.8"
you'd likely need a hard-to-find adapter. On the other side
is the IDE ribbon cable interface. I'm not aware of any
web pages that compare the performance of these, so you may
end up buying a few and testing them.

http://www.newegg.com/Product/Produc...82E16812200156

But how can its size fit into a laptop!

Besides I doubt it'd allow for a greater speed than 66MB/s... Why?
Because probabely such a SATA2IDE adapter has the same controler on
board that the SSD PATA in question has and bottlenecks its throuput!
This is why the same brand's similar items (but one is PATA and the
second SATA) have threefold difference of speed!

Alojzy Zakalec wrote:
On 30 Lip, 21:33, Paul wrote:
There are various brands of these, and the SATA connector on this,
plugs directly into a SATA 2.5" or 3.5" device. For SATA 1.8"
you'd likely need a hard-to-find adapter. On the other side
is the IDE ribbon cable interface. I'm not aware of any
web pages that compare the performance of these, so you may
end up buying a few and testing them.

http://www.newegg.com/Product/Produc...82E16812200156

But how can its size fit into a laptop!

Besides I doubt it'd allow for a greater speed than 66MB/s... Why?
Because probabely such a SATA2IDE adapter has the same controler on
board that the SSD PATA in question has and bottlenecks its throuput!
This is why the same brand's similar items (but one is PATA and the
second SATA) have threefold difference of speed!

Well, I don't have enough hardware here to dispute that claim.
The claim does not make sense though.

To give an example, my previous primary computer was a P4C800-E
with Intel Southbridge. Back when Intel Southbridges had IDE
interfaces on them, they never offered to operate at 133MB/sec.
The design was intended for 100MB/sec operation. Read and write
performance differed, because of how a write strobe signal
was created for the ribbon cable. The best the chip could
do on write, was 88.9 MB/sec instead of the theoretical
100MB/sec. I could verify that, using sustained transfer
disk benchmarks. And the burst rate reproduced the expected
numbers, so I was able to verify the information presented
in the Intel datasheet for the Southbridge.

So I know I have at least one interface here, that can do
88.9MB/sec writes. And that is a 100MB/sec interface, and
not even attempting to do 133MB/sec.

So I'm not buying there is some kind of 65MB/sec barrier,
that nobody has ever noticed before on their hardware.
There can be plenty of performance bottlenecks in hardware
implementations, such as bridging a storage interface
to an internal PCI bus, that can cause funny behaviors.
But I refuse to believe, that every motherboard chipset
produced, suffers from such problems.

I'll see if I can cook up a test case, and get back to you
later.

Paul

On 30 Lip, 19:56, John Doe wrote:
Alojzy Zakalec alojzy.zakalec gmail.com wrote:

How do you think, can the shop selling their products:
* *http://www.solidata.hk/onlineshopitems/p2.htm
be fully trusted (to pay money a priori)?

The massive amount of spam selling counterfeit products comes from
Hong Kong, pouring into USENET through Google Groups. Hong Kong is
where the counterfeit products you find on eBay are sold from.

Solidata seems to be the standalone company, there is their site to
judge oneself:
http://www.solidatum.com/enoverview.asp?title=overview

And whats more I think manufacturing SSDs is very simple undertaking,
you buy few ready chips (controler, cache, flash memories, and maybe
converter to IDE) and assemble them on circuit board wtih primitive
logic and interface connector (SATA or IDE).
No extremely advanced technology is needed, a pack of students from
technical univeristy, or even hobbyists-enthusiasts, easily can do the
job! (And I suspect this is the case of "SSDfactory"
http://www.ssdfactory.com/en/aboutus.html )

On 31 Lip, 05:45, Paul wrote:
So I'm not buying there is some kind of 65MB/sec barrier,
that nobody has ever noticed before on their hardware.
There can be plenty of performance bottlenecks in hardware
implementations, such as bridging a storage interface
to an internal PCI bus, that can cause funny behaviors.
But I refuse to believe, that every motherboard chipset
produced, suffers from such problems.

I'll see if I can cook up a test case, and get back to you
later.

It would be very welcome indeed, Paul!
Thanks in advance

Alojzy Zakalec wrote:
On 31 Lip, 05:45, Paul wrote:
So I'm not buying there is some kind of 65MB/sec barrier,
that nobody has ever noticed before on their hardware.
There can be plenty of performance bottlenecks in hardware
implementations, such as bridging a storage interface
to an internal PCI bus, that can cause funny behaviors.
But I refuse to believe, that every motherboard chipset
produced, suffers from such problems.

I'll see if I can cook up a test case, and get back to you
later.

It would be very welcome indeed, Paul!
Thanks in advance


Well, the results are interesting, and not what I was expecting.

My current motherboard has a VIA chipset (VT8237S), and the
ATA is supposed to support UDMA6 (133MB/sec). What I don't know,
is where the controller is bridged in the Southbridge. In years past,
it may have been bridged to the PCI bus, in which case there would be
a capacity limit caused by the PCI bus (110-120MB/sec at most).

I dug out a seven year old PC, and pulled a Maxtor UDMA6 drive
out of it. The drive is 7200RPM, but has dreadful transfer
rate for sustained transfer. It only does about 45MB/sec or so
on sustained read, near the beginning of the disk.

My reason for using that disk, is to benchmark burst transfer
speed (not sustained, which is head limited). I compared it
to my current generation Seagate IDE drive, which is limited
to UDMA5. These are read-only burst test results, intended
to highlight "bottlenecks in the plumbing".

HDTach3040 HDTune255
Seagate UDMA5 94.3MB/sec 78.9MB/sec Theoretical max 100MB/sec on cable
Maxtor UDMA6 119.6MB/sec 98.2MB/sec Theoretical max 133MB/sec on cable

So my first problem, is the utilities do not agree on the results.

None of the results are as bad as 65MB/sec. The above are read
tests, and I don't have a benchmark tool that will do writes.
(I have WinTune97 that runs under Win98SE, which I could dig
out, but I don't know if its results are any more trustworthy.
I think it leaves the OS file cache enabled.)

The results are still lower than I would have liked. I don't
think writing these benchmark tools is that easy, due to the
opportunity to make mistakes.

The fact that DMA transfer is involved in moving the data from
the Southbridge port, into system memory, should really allow
the thing to run at full speed. There are no headers on the
data coming across the cable -- it should just be 512 bytes
per sector. So the data should be able to stream into memory.
It all depends on how the controller logic block is bridged
to the system buses, as to how close to theoretical max
it can get.

SATA uses packets for transfer, so there is some kind of header
on each packet. The results I've seen so far, still don't seem
to make it all the way to 300MB/sec.

I have one other motherboard that would be worthy of testing,
but it is retired, and is not in a computer case at the moment.
And I have no spare power supplies. So it would be a bit of
work to set that up. I don't even have a table top to put it
on !

The other option, is to find a PCI Express card with IDE connector
on it, and evaluate the performance of that. The reason I suggest
that, is PCI Express x1 has 250MB/sec of bandwidth, and perhaps that
would no longer limit a UDMA6 IDE interface. But if your computer
doesn't have such an interface, again, this would only be of
academic interest.

http://www.jmicron.com/PDF/JMB363/JMB363.pdf

BYTECC PCIe SATA II 300 + PATA Raid Card Model BT-PESAPA
(Uses a JMB363)

http://c1.neweggimages.com/NeweggIma...283-005-04.jpg

Paul