Only Getting SATA I Speed

[jimlau]

i'm having trouble getting SATA III speeds with my SSD SATA III (Patriot Memory Pyro, 60GB). i have all the drivers for the chipset and motherboard. i have a SATA III cable. the SSD is filled about 88%.

my motherboard is rated for SATA III (Asus M5A88-M). i used HD TubneUp Pro and CrystalDiskMark for testing. they show the read speed is about 140MB/s.

i have windows 7, 64-bit, SP1.

i formatted the SSD with an allocation size of 64k. my main drive, that holds the OS, is a HDD. the SSD only holds data for reading, no writing.

any idea what is going wrong?

thanks.

 

[RetiredChief]

Download and run AS SSD (Do not need to run the benchmark) Uper left - what driver is being reportd.

Verify the drive is NOT using PIO for transfer, should be UDMA.

May not like the 64 K cluster size.
For a SMALL drive and unless your avg file size is large, very little advantage of 64 K.

Read up on page and block size which are used in SSDs.

 

[foscooter]

Are your SATA ports (in the BIOS) set to ACHI?

Did you do the "SSD Tweaks" which are in here somewhere, but lately I can't find them.

 

[daysyang]

^-- what he said... if you dont set it in ACHI, and its running on IDE it will run like a regular hhd.

 

[jimlau]

BIOS was set to IDE. was tricky, but i switched to ACHI.

i have 64k because the file sizes aremostly large.

i can't see on the AS SSD screen where driver type is listed

the benchmark results for 2 testers is about 200MB/s read rate.

i did see on Tune HD that there is another tab that tests FILE benchmark rate. THAT gets me to 500 MB/s. how is file different than benchmark?

ok, i also get that speed using OTTO, which is the tool Patriot told me to test with. on that tester, it shows speed increasing to SATA III rates from about 64k size and bigger.

so 2 of 3 testers say read rate is slow. is OTTO legit?

thanks.

 

[jimlau]

also, when i simply copy and pasted a file from the SSD to my HDD, that rate is just 90MB/s, which i can get with USB 3. why so slow?

in have a quad core 3.4 GHz AMD with 8GB RAM.

 

[foscooter]

The transfer rate is low because your are copying TO the HDD. You'd be lucky to get up tp 120MB writes on an HDD. Run ATTO on your HDD as see.

Your SSD speeds are as good as they'll get.

 

[jimlau]

thanks.

still unclear as to why 2 out of 3 testers give the slow rate of 200MB/s. any idea why? that's unrelated to the main drive being an HDD, right?

 

[Dereck47]

still unclear as to why 2 out of 3 testers give the slow rate of 200MB/s. any idea why?

CrystalDiskMark & AS-SSD use highly incompressible data to test Read/Write speeds.
ATTO uses highly compressible data to test Read/Write speeds.

Highly compressible data is "best case" scenario for SSDs; highly incompressible data is "worst case".
With SandForce based SSDs, the more a block of data can be compressed the faster it can be Read/Written to the drive.

 

[jimlau]

so which is more representative of real world scenarios? my SSD will be used to store thousands of .wav files, and some large compressed (i think) nkx audio sample files. i guess some of both type?

is there an option to compress files safely so transfers are faster? or would they have to be then decompressed before they are useable, making that point moot?

thanks.

 

[Soda-88]

ssd should never be filled up over 85% because the drive usually has to perform garbage collection during writes which severely hammers the performance

free up some space, leave it idling for couple of hours for GC to do its job and rebench again

edit: this is especially obvious in your case since you didn't have ahci mode turned on, hence no trim commands which improves garbage collection

 

[jimlau]

the brand i bought was Patriot Pyro, due to cost. any sense if my speed results are in any way related to this brand?

 

[Dereck47]

so which is more representative of real world scenarios? my SSD will be used to store thousands of .wav files, and some large compressed (i think) nkx audio sample files. i guess some of both type?

Music files and video files are highly incompressible, so your drive performance will be lower working with those types of files.

is there an option to compress files safely so transfers are faster? or would they have to be then decompressed before they are useable, making that point moot?

No, there's nothing the end user can do. The SSD firmware attempts to compress the data before it is written to the NAND chips on the drive.

 

[Dereck47]

the brand i bought was Patriot Pyro, due to cost. any sense if my speed results are in any way related to this brand?

Most manufacturers advertise their Read/Write speeds based upon ATTO because it is best-case scenario.

 

[smorizio]

also check the fireware on the drive...a lot of time with new ssd the vendors or drive controler will send out tweeks to make there drive perform better in real world.

 

[RetiredChief]

In your case, AS SSD is more representative of what you will get in real life - also more representative for a OS + Program drive.
And yes that is Pi@* poor for a sata III drive. When compared to an intel chipset using iaSTor (msahci provides slightly lower performance), Both my 128 gig M4 and Samsung 830 overall score over 700. Even my lowly Agillity 3 was over 400. Yours should be close so something is amiss.

I have heard that using the reg edit and reboot does not always do the trick (Even though the driver shows msahci)- at least one poster who did that also got poor scores so he re-installed window and his scores shot up to normal.

Not sure how much your chipset knock off on performance, when compared to an Intel chipset - Notice that the vast majority of Benchmarks for SSDs are done on an Intel chipset. Probably only a small difference.

As to cluster size: On my HDDs for My video partition, Yes I increase the cluster size but that is because they are from 1 gig to 40 gigs for a single file. Not saying that cluster size is the reason - without trying I don't know. But if it is affecting that much then the performace degragtion outways the increased cluster size factor big time.

Still faster than a speeding car, just not as fast as a speeding bullet and was faster than a HDD.

 

[jimlau]

thanks for all the input. will dig some more.

btw, does anyone know if the Pyro SE is much better than the Pyro?