Overclock.net › Forums › Components › Hard Drives & Storage › SSD › Corsair Force 3 + Asus U3S6, faster readings but slower writings!?
New Posts  All Forums:Forum Nav:

Corsair Force 3 + Asus U3S6, faster readings but slower writings!?

post #1 of 8
Thread Starter 
Hi guys!
Last week I got my Asus U3S6 and happily plugged my ssd to it as my mobo, an Asus P5Q3, has just the Sata II.
So I did a few tests and this is what I got:


ATTO 2.4.6.
As you can see the readings are much faster but on the writings there's even a loss, ~ -13%, which is not quite what I was expecting to see.

Another test run with AS SSD Benchmark in the spoiler: Warning: Spoiler! (Click to show)

At boot this is what I get:


Is that supposed to mean it's running at x1 !?

Also, my Sapphire HD4850 Toxic is now going at x8 as read by GPU-Z during the render test but I'm not sure that's going to be an issue for now unless I'll want to change my gfx card with one which requires an higher bandwidth.



Just to mention, I also have a SB Live! 5.1 and an Asus 1GB nic on my mb.
The 4850 is in the blue slot while the controller in the black one.


Any thoughts about this ? I'm not sure the speeds are barely matching those dreamed when I bought the ssd still being aware of the limits given by my Sata II mobo but now with the controller I had some hope and instead, well, you've seen the results.

How my mb is being such a bottleneck for my ssd? Or the cause could be another? (e.g. firmware)

Thanks, everyone!
Edited by Luc4 - 7/19/13 at 9:19am
post #2 of 8
Quote:
Originally Posted by Luc4 View Post

At boot this is what I get:


Is that supposed to mean it's running at x1 !?

How my mb is being such a bottleneck for my ssd? Or the cause could be another? (e.g. firmware)

Your motherboard isn't listed as one of the compatible models. I'm guessing it needs a little something special for it to work in x4 mode. Unless you're prepared to spend $300-400 on a decent RAID card (which is likely around the same as you'll spend for a low mid-range Haswell upgrade), just stick to the SATA2 Intel ports on your board.
Garnet
(11 items)
 
Lucifiel
(13 items)
 
Metatron
(13 items)
 
CPUMotherboardRAMHard Drive
Intel Core i5-3450S Intel DQ77KB Corsair 16GB DDR3 1600 SO-DIMM Samsung 830 256GB 
Hard DriveHard DriveCoolingOS
Crucial m4 256 mSATA Samsung 840 500GB Intel BXHTS1155LP Windows 7 Ultimate x64 
KeyboardCaseMouse
Logitech K800 Lian Li PC-Q05B Logitech M570 
CPUMotherboardGraphicsRAM
Core i7-860 Gigabyte GA-H55N-USB3 EVGA GTX 460 1GB GDDR5 Kingston 2x4GB DDR3 1333 
Hard DriveOSPowerCase
G.SKILL Phoenix Pro 120GB Windows 7 Ultimate x64 Silverstone ST45SF 450W Silverstone Sugo SG05B 
CPUMotherboardGraphicsRAM
Core i7-860 Biostar TH55B HD MSI GT 240 1GB GDDR5 Kingston 2x2GB DDR3 1333 
Hard DriveOSMonitorPower
Intel X25-M 120GB Windows 7 Ultimate x86 Samsung 22" Antec EA-380D 
Case
Rosewill R101-P-BK 
  hide details  
Reply
Garnet
(11 items)
 
Lucifiel
(13 items)
 
Metatron
(13 items)
 
CPUMotherboardRAMHard Drive
Intel Core i5-3450S Intel DQ77KB Corsair 16GB DDR3 1600 SO-DIMM Samsung 830 256GB 
Hard DriveHard DriveCoolingOS
Crucial m4 256 mSATA Samsung 840 500GB Intel BXHTS1155LP Windows 7 Ultimate x64 
KeyboardCaseMouse
Logitech K800 Lian Li PC-Q05B Logitech M570 
CPUMotherboardGraphicsRAM
Core i7-860 Gigabyte GA-H55N-USB3 EVGA GTX 460 1GB GDDR5 Kingston 2x4GB DDR3 1333 
Hard DriveOSPowerCase
G.SKILL Phoenix Pro 120GB Windows 7 Ultimate x64 Silverstone ST45SF 450W Silverstone Sugo SG05B 
CPUMotherboardGraphicsRAM
Core i7-860 Biostar TH55B HD MSI GT 240 1GB GDDR5 Kingston 2x2GB DDR3 1333 
Hard DriveOSMonitorPower
Intel X25-M 120GB Windows 7 Ultimate x86 Samsung 22" Antec EA-380D 
Case
Rosewill R101-P-BK 
  hide details  
Reply
post #3 of 8
When I saw the ATTO results with the U3S6 card, my first reaction was, that looks like the typical Marvell 912x "SATA III" chipset results.

Many of us here know that the Marvell 912x series SATA chipsets cannot provide true SATA 6Gb/s performance, since they only have one PCIe lane connected to the chipset, at 5Gb/s max, not 6Gb/s. I was not familiar with the U3S6 card, so I did a quick search, and checked it on 'ASUS web site.

I couldn't tell what SATA chip was on that card, so I checked the driver downloads, and found this:

Marvell 9123 Controller Driver V1.0.0.1027 for Windows 32/64bit XP & Windows 32/64bit Vista & 32/64bit Windows 7.(WHQL)

There it is, the Marvell 9123 SATA chipset! No surprise really, just more of the same old marketing BS. thumbsdownsmileyanim.gif

Sorry Luc4, but you will never get full SATA 6Gb/s speeds from the U3S6 card, it is impossible. I don't care if that card has a PCIe X 4 bridge chip, which is probably more useful for the USB 3.0 ports than the SATA ports.

Yes, as it shows in your boot picture, the Marvell chipset is connected to one PCIe lane, and is running at 5.0 Gb/s. The "Bridge chip", really likely a multiplexing chip, is just a switch sending signals from multiple inputs to fewer outputs, two at most. Even the picture that ASUS provides describing how the U3S6 works, shows the output of the bridge chip is connected to ONE PCIe 2.0 lane, into the Marvell SATA chip.

Even if the bridge chip did combine the inputs from four PCIe lanes into one signal, the output of the bridge chip is still one PCIe lane, and its speed is no faster than 5Gb/s.

How do they get away with this?

You're not using the Marvell SATA driver now, just the standard Windows msahci driver. You might get a bit more speed using the Marvell driver, but it won't be a big increase, and could be worse in some ways too.

Your mother board is not the problem, and new SSD firmware will not make a difference, it is the U3S6 card with the Marvell 9123 chipset that is the main limitation.
post #4 of 8
Thread Starter 
Thank you for your very exhaustive reply Parsec.

Thru AIDA64 I get this:
Quote:
In uso @ x4 (PLX Technology ExpressLane PEX 8608 8-Lane 8-Port PCI Express 2.0 Switch --> PLX Technology ExpressLane PEX 8608 8-Lane 8-Port PCI Express 2.0 Switch --> Marvell 88SE9120 SATA 6Gb/s Controller, PLX Technology ExpressLane PEX 8608 8-Lane 8-Port PCI Express 2.0 Switch --> NEC uPD720200 USB 3.0 Host Controller)
Which is a bit frustrating as it doesn't seem to be @ x4 ... anyway.
Not sure if the chip is the 9120 or 9123, can check manually eventually.

Yesterday I remembered I did not install the Marvell driver for sata but only the usb so took the disc and installed.
It was the 1.0.0 1042 18/7/2010.

The results were poor compared with the windows driver. I guess the driver that came with the disc is even older than Windows' one.

Yesterday:

This morning:

I really have no idea why the speeds even dropped that much, the system was in "idle", about this last above pic I've tried a few times and the results were always the same, speeds dropped unexplainably.

I've so then updated the drivers and got this:
Quote:
Windows driver, it's the same pic of the 1st post.

So overall the solution with the 1.2.0 1035 drivers seems a bit better.

If anyone should ever need those drivers I've found them here: http://www.station-drivers.com/page/marvell_sata.htm, the EXEs are compressed 7zip files which it's possible to decompress using 7zip itself (to avoid malicious code) and then to install them just let windows search for updated drivers in the main directory where you have decompressed them.



Quote:
Originally Posted by parsec View Post

...
There it is, the Marvell 9123 SATA chipset! No surprise really, just more of the same old marketing BS. thumbsdownsmileyanim.gif
...

Your mother board is not the problem, and new SSD firmware will not make a difference, it is the U3S6 card with the Marvell 9123 chipset that is the main limitation.

I'm not sure I should feel like scammed by Asus about it, honestly when some time ago (months) I've been reading some reviews this card didn't seem to have any unhappy surprise, I should have checked again before ordering it. Price was even appealing 29€ compared to the usual 39~40 I've seen previously. I wonder how it really works with those models on the compatibility list, on their forum there was also somebody that was using it with a P35.


Do you think/know there's some other valid better controller to replace this? I could do with the readings speeds but in this situation what bothers me most is the very low writing speed.sadsmiley.gif
Edited by Luc4 - 7/20/13 at 6:19am
post #5 of 8
Quote:
Originally Posted by Luc4 View Post

...
I'm not sure I should feel like scammed by Asus about it, honestly when some time ago (months) I've been reading some reviews this card didn't seem to have any unhappy surprise, I should have checked again before ordering it. Price was even appealing 29€ compared to the usual 39~40 I've seen previously. I wonder how it really works with those models on the compatibility list, on their forum there was also somebody that was using it with a P35.

My comments about the false claims made about the performance of Marvell SATA chips like the 9123, are caused by the fact that this has been done for years, and is still being done today.

It's also based on the contradictory information that ASUS (and others) has on the product description page itself: http://www.asus.com/Motherboards/U3S6/

You should also read some reviews of your Force 3 SSD, that is also a factor in the test results you get: http://www.legitreviews.com/article/1723/6/
Quote:
Originally Posted by Luc4 View Post

Do you think/know there's some other valid better controller to replace this? I could do with the readings speeds but in this situation what bothers me most is the very low writing speed.sadsmiley.gif

SSD owners with SATA II mother boards have been looking for add on SATA III cards to use with their SATA III SSDs for years. SSD enthusiasts know that the cheap cards like yours and others never provide the performance of the Intel SATA III chipset. As rui-no-onna said, only expensive RAID cards can provide the speed you would like to see, but you can get a new SATA III mother board and CPU for the same price.

Frankly, as can be seen in the review in the link I added above, even on a true SATA III mother board, the write performance of a 120GB Force 3 is basically what you are getting now, under some circumstances.
post #6 of 8
Quote:
Originally Posted by parsec View Post

You should also read some reviews of your Force 3 SSD, that is also a factor in the test results you get: http://www.legitreviews.com/article/1723/6/

He's testing with ATTO which uses highly compressible data so he should actually be getting advertised speeds.

@Luc4
Just a thought, do you by any chance have compression enabled on your drive? You're getting really high read speeds so I'm thinking it might not be the interface that's the problem.
Garnet
(11 items)
 
Lucifiel
(13 items)
 
Metatron
(13 items)
 
CPUMotherboardRAMHard Drive
Intel Core i5-3450S Intel DQ77KB Corsair 16GB DDR3 1600 SO-DIMM Samsung 830 256GB 
Hard DriveHard DriveCoolingOS
Crucial m4 256 mSATA Samsung 840 500GB Intel BXHTS1155LP Windows 7 Ultimate x64 
KeyboardCaseMouse
Logitech K800 Lian Li PC-Q05B Logitech M570 
CPUMotherboardGraphicsRAM
Core i7-860 Gigabyte GA-H55N-USB3 EVGA GTX 460 1GB GDDR5 Kingston 2x4GB DDR3 1333 
Hard DriveOSPowerCase
G.SKILL Phoenix Pro 120GB Windows 7 Ultimate x64 Silverstone ST45SF 450W Silverstone Sugo SG05B 
CPUMotherboardGraphicsRAM
Core i7-860 Biostar TH55B HD MSI GT 240 1GB GDDR5 Kingston 2x2GB DDR3 1333 
Hard DriveOSMonitorPower
Intel X25-M 120GB Windows 7 Ultimate x86 Samsung 22" Antec EA-380D 
Case
Rosewill R101-P-BK 
  hide details  
Reply
Garnet
(11 items)
 
Lucifiel
(13 items)
 
Metatron
(13 items)
 
CPUMotherboardRAMHard Drive
Intel Core i5-3450S Intel DQ77KB Corsair 16GB DDR3 1600 SO-DIMM Samsung 830 256GB 
Hard DriveHard DriveCoolingOS
Crucial m4 256 mSATA Samsung 840 500GB Intel BXHTS1155LP Windows 7 Ultimate x64 
KeyboardCaseMouse
Logitech K800 Lian Li PC-Q05B Logitech M570 
CPUMotherboardGraphicsRAM
Core i7-860 Gigabyte GA-H55N-USB3 EVGA GTX 460 1GB GDDR5 Kingston 2x4GB DDR3 1333 
Hard DriveOSPowerCase
G.SKILL Phoenix Pro 120GB Windows 7 Ultimate x64 Silverstone ST45SF 450W Silverstone Sugo SG05B 
CPUMotherboardGraphicsRAM
Core i7-860 Biostar TH55B HD MSI GT 240 1GB GDDR5 Kingston 2x2GB DDR3 1333 
Hard DriveOSMonitorPower
Intel X25-M 120GB Windows 7 Ultimate x86 Samsung 22" Antec EA-380D 
Case
Rosewill R101-P-BK 
  hide details  
Reply
post #7 of 8
Thread Starter 
Thank you again for your time guys,

@rui-no-onna: no I don't have compression enabled but - although I'm not sure this could be falsifying much the tests - on that unit I've 3 directory junction points: one is to a ram drive, the two others to two different directories on two different hds; moreover, there are just 54GB free of 111, Windows is running on a different hd but the user directory is on the ssd and so in use.

@parsec: I'm not sure I can spot the contradictory information on that page, perhaps do they mean that the full bandwidth it's split among the 4 ports in equal parts? That could be weird but plausible...
I've checked the prices of those cards and indeed a new cpu and mb would cost less! (depends biggrin.gif )

By the way, thru crystal disk info I've just noticed that the ssd's firmware is the 1.3.3 while the last seems to be 5.05a.
I'm not having any system crash, unusual reboots, bsod or anything else. The only few freezes I had in the past were related to the faulty power connector of the hdd with C:, I really can't recall anything in particular about the ssd, not even a check disk at boot.
So let's say it's stable and apparently working well with his firmware hence I would not be very keen on an update but I'm curious to know what all those changes are so later will try to have a read about, you never know.
post #8 of 8
Quote:
Originally Posted by rui-no-onna View Post

He's testing with ATTO which uses highly compressible data so he should actually be getting advertised speeds...

Seriously, do you think I don't know that? wink.gif

Yes, he should be getting better write speeds in ATTO, even with a Marvell SATA chipset, but he doesn't. Why?

That review also has an AS SSD test result, but given the SSD has a SandForce controller that does not perform as well with 100% non-compressible data (as you know), we still aren't seeing the same result that we do with other SandForce SSDs, that can provide ~500MB/s sequential read speeds. Again you know that is due to the asynchronous NAND used in a Force 3.
Quote:
Originally Posted by Luc4 View Post

Thank you again for your time guys...

@parsec: I'm not sure I can spot the contradictory information on that page, perhaps do they mean that the full bandwidth it's split among the 4 ports in equal parts? That could be weird but plausible...
I've checked the prices of those cards and indeed a new cpu and mb would cost less! (depends biggrin.gif )

It's in several places, blatantly ridiculous IMO.

Right under "Future Transfer Technology" we see "True SATA 6Gb/s Support"

Then we have this picture:



Which then shows "500MB/s PCIe Gen2 bandwidth", twice, going to a SATA 6Gb/s controller, and drive. But wait, isn't PCIe Gen2 bandwidth 5Gb/s? Yes it is, but why is that not in the text?

Is the text in the orange box that states, "Sufficient bandwidth for USB 3.0 and SATA 6Gb/s", with an arrow pointing to the PCIe 2.0 lane, a correct statement?

Anyway, I'm done with this, I wish you luck with that card and your SSD. The SandForce 5 series firmware debacle is yet another trail that will lead nowhere, sorry to say.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: SSD
Overclock.net › Forums › Components › Hard Drives & Storage › SSD › Corsair Force 3 + Asus U3S6, faster readings but slower writings!?