Are My Writes Slow?

[idata]

In comparing my benchmarks with other I have seen on this forum, I find that my read speeds seem similar to others but my write speeds are lower. This has been the case since the first benchmark run, right after setting up the new PC, so I'm guessing it's nothing to do with TRIM. Win 7 64-bit and my driver is msahci.sys, no RAID. Any thoughts as to why the writes are slower than others have reported?

Thanks,

Sally

[idata]

You asked - Is Superfetch disabled?

The service is running. But it was my understanding that it would actually not be active on my SSD drive. I know that Windows has recognised my SSD as it is not available in the Defrag Scheduler. So I have been assuming that it has also stopped Superfetch on the SSD, in spite of its status as running.

Microsoft have said here:

http://blogs.msdn.com/e7/archive/2009/05/05/support-and-q-a-for-solid-state-drives-and.aspx http://blogs.msdn.com/e7/archive/2009/05/05/support-and-q-a-for-solid-state-drives-and.aspx

"If the system disk is an SSD, and the SSD performs adequately on random reads and doesn't have glaring performance issues with random writes or flushes, then Superfetch, boot prefetching, application launch prefetching, ReadyBoost and ReadDrive will all be disabled. Initially, we had configured all of these features to be off on all SSDs, but we encountered sizable performance regressions on some systems. In root causing those regressions, we found that some first generation SSDs had severe enough random write and flush problems that ultimately lead to disk reads being blocked for long periods of time. With Superfetch and other prefetching re-enabled, performance on key scenarios was markedly improved."

[idata]

If it's running, it is active. You can disable it in local services if you want to do that. According to that "blog" you quoted, Windows 7 should disable Superfetch on SSDs that are fast enough, and I think these Intel SSDs are fast enough.

[idata]

Oh, just remembered you have non-SSD drives in your computer, so I guess you shouldn't disable it. Don't be so sure that Windows 7 is not running Superfetch for your SSD. Windows 7 has some flaws... If you want to test it, disconnect your non-SSD drives and see if Superfetch is running.

[idata]

Having the same numbers of your original post. My hardware is:

MB: MSI 790FX GD70

PROC: Phenom II x4 955BE

RAM: 4x2GB OCZ DDR3 PC3-12800/ 1600MHz / AMD Black Edition

SSD: Intel 510 Series 120GB

HD: WD Black Caviar 1TB + 64MB cache

VGA: Radeon HD 5870

Windows 7 Enterprise 64 bits SP1

The reading speeds are great, almost saturating the capacity of the SATA II port for a SATA III device. I used the stock cable SATA cable that came with the SSD, should I use a SATA II cable, as one of those that came with the mobo??

My writing speeds wont go over 60MB/s using crystaldiskmark. I did all the adjustments suggested in this website:

http://www.mydellmini.com/forum/windows-7/2441-windows-7-ultimate-solid-state-drive-speed-tweaks.htm... http://www.mydellmini.com/forum/windows-7/2441-windows-7-ultimate-solid-state-drive-speed-tweaks.htm...

That in short are:

Disable indexing

Disable defragmentation

Disable Write Caching

Configure Superfetch

Firefox - Use memory cache instead of disk cache

Disable the Page File

I dont have right now, the screens from the CrystalDiskMark, but as I said, it wont go higher than 60 MB/s

Any suggestions?

[idata]

Guest wrote:

The reading speeds are great, almost saturating the capacity of the SATA II port for a SATA III device. I used the stock cable SATA cable that came with the SSD, should I use a SATA II cable, as one of those that came with the mobo??

There is no such thing as a "SATA II" cable, or a "SATA III" cable. Anyone claiming otherwise or selling you such things is lying and wants to make a buck off you. The SATA (revision 1; SATA150), SATA (revision 2; SATA300), and SATA (revision 3; SATA600) standards all use the exact same cable and connector types.

If someone tells you "well it matters because a better quality cable with better shielding will guarantee faster speeds", that is also a lie. There are SMART attributes that track retransmissions (specifically CRC errors) between a SATA controller and a hard disk (specifically SMART attribute 199 / 0xC7). An excessive (and ever-incrementing) number indicates CRC errors, which is indicative of a bad/faulty SATA cable, dust/debris in the SATA connector port (on the disk or the mainboard), or a faulty bus between the SATA controller and the port (or the SATA controller on the disk and its port). However, for whatever reason, Intel did not implement this attribute in their SSDs, so to my knowledge there's no way to track CRC errors with their SSDs. (Intel engineers, are you listening? Does your controller PHY track these stats? If so, please implement a tie-in to SMART for this!)

TL;DR -- your write performance is at 60MByte/sec not because of "bad or cheap" SATA cables. 60MByte/sec on a consumer-grade SSD is quite good, especially on the X25 series. SSD write performance is sub-par and not stellar; consumer-grade SSDs perform extremely well with reads, not with writes. Write performance will suffer when there isn't enough free space on the drive (for wear levelling; excessive GC and NAND flash block erasing has to occur in this situation, which greatly destroys write performance). Keep in mind that the OP's SSD is an X25, while yours is supposedly a 510.

Finally, the benchmarking application you use matters severely. Remember: benchmarking applications just show you numbers, and unless you know what the benchmark is testing and what the numbers mean in respective context, then you're wasting your time. For example, AS SSD Benchmark's documentation is only available in German, so it's amusing to see people concerned about benchmarks talking about numbers when they very likely don't even know what all the tests mean/represent.

Anyway, please try using something like AS SSD Benchmark and ATTO (test block sizes between 8KByte and 512KByte; no need to test anything smaller or larger -- and change the queue method from "Overlapped I/O" to "Neither") and provide screen-shots of your results. AS SSD will also be kind enough to show if your partition is properly aligned or not.