Samsung 830 256GB question

Toggle sidebar Toggle sidebar
jurge92

jurge92

Distinguished
May 6, 2012
48
0
18,530
Hi.

I just installed a 256GB Samusng 830 in my rig, but to my suprise, the CrystalDiskMark is way lower than advised. I've been doing all the sane SSD optimization tips, but with no increase in speed. I also get 7.8 in windows experience index, dunno if it should be 7.9.

Here is the speed:

Read:

Seq: 330.8
512K: 252.1
4K: 19.80
4K QD32: 93.69

Write:

Seq: 229.2
512K: 223.7
4K: 43.35
4K QD32: 93.17

Does anyone know why this is?
 

What mode is the SATA Controller in? They can run in either IDE, AHCI, or RAID. You can check this in the BIOS. If you are running in IDE mode, switch to AHCI to get a performance boost. HOWEVER you will need to make a registry edit prior to switching the controller's operating mode, otherwise Windows will BSOD during boot. Linux does not have this issue.
Another performance boost would be to check for firmware updates.
Though this usually fixes itself from what I understand, but make sure TRIM is enabled.
Open Command Prompt with Administrative privileges (Run as administrator) and enter the follow command.

fsutil behavior query DisableDeleteNotify

If the result is '0' TRIM is enabled.
Click to expand...
http://answers.microsoft.com/en-us/windows/forum/windows_7-hardware/how-can-i-know-if-trim-command-is-working-on/80e600c0-d2fd-41b4-801b-650e05dac350
 

In that case, the only thing left is to check and update the firmware. Those speeds do seem pretty low. Something I would do is slap Ubuntu on a USB flash drive, boot it up, and use Disk Utility to get another benchmark. This help determine if the problem is with Windows, or the SSD itself.
 

Does it feel slow? Because the only thing I have left is run other benchmarks. Re-run CrystalDiskMark. There is also ATTO and AS SSD. And Disk Utility on Linux.

The only poosibility I can think of is that you have it in an SATA II port, but your read speeds seem to surpass SATA II.
 
Did you stop your security software (anti virus, firewall) for the test? For further investigation some more information of your system would be helpful (MB, CPU). Beside that the sequential read/write rate seems to be ok.
 
I didn't stop Avast! for the test(it's the only anti-virus, etc. I have). It doesn't feel slow no, but I think it could've been faster. I have used CrystalDiskMark two times today allready, one before the optimizations and one after, with exact results.

My system is:

CPU: Intel Core i5 2500K 3.30GHZ 6MB
GPU: ASUS Radeon HD6870 DirectCU 1GB
MOBO: ASUS P8Z68-V PRO Z68 S-1155 ATX
RAM: Corsair 8GB Vengeance 1600MHZ CL9
SSD: Samsung 830 256GB
HDD: Caviar Green 1TB
PSU: Cooler Master GX ATX12V 2.3 750W
Case: Antec Miditower Three Hundred ATX Black
OS: Windows 7 Ultimate 64-bit

I'm pretty sure it's in Sata III, it also says it's connected to 6GB/s in boot.
 


It's probably Avast then; it uses a real-time scanner to scan incoming and outgoing files from the disk. Stop Avast, and run the bench again.
 
Ah I see. Well I dunno if I should run anymore tests today, since I've read it might damage your disk. It's not something urgent, I could probably wait until tomorrow.
 
1) I think you have two different sata III (6) ports. One is Intel, the other is a marvel based controller. You should be on the Intel sata III port. The marvel port will know performance down. Look in your manual (one should be blu and the other gray).

2) Down load and run AS SSD (DO NOT NEED to run benchmark), just open the program and look at upper left. should SEE:
.. Firmware version (already stated it was the latest).
.. Driver - For Your system, as stated should be on the INTEL sata III port. If so you should have down loaded and installed the latest Intel RST driver and AS SSD should show iaSTor. msahci as driver is slightly lower than iaSTor.
.. Partition algin - should show "OK"

A) Installation, you did do a clean install (recommended) ??? Only time I recommend migration is problem that prevents clean install. Do not recommend cloning HDD -> SSD. And yes there is a diff between cloning and migration.

B) Repeated benchmark runs are NOT recommended, Not so much from a stand point of distroying the drive, BUT when run back-to-back benchmarks will progressively yield lower performance numbers. Reaso, Trim and Garbage collector require Time to work thie magic and restore the performance back to near factory fresh.

C) Also verify your MB Bios is the latest Rev - this has also been the problem for a few.

ADDED:
D) I do NOT use ATTO. However this is the Most widely published benchmark by manuf. ATTO was developed for HDD and as such it uses HIGHly compressable data - This is NOT real life. The overall compression rate of OS and program files loaded can only be compressed by a small amount. Also using Uncompressed data provides an advantage to SF based SSD that does NOT translate to real world. I prefer AS SSD as it uses compressed data for it's benchmarks, Still Not real world, But MUCH closer.

E) Also on the topic of benchmarks. Sequencial read and writes are the least important parameter for an OS + Program Drive, It is the 4 K random performance that is IMPORTANT.
 
Dude, you had some good points; especially the emphasis on random I/O, but there are a few things I'd like to add my 2 cents on.

If the benchmark degrades over repeat tries, then it's not a benchmark. Benchmarks need to be consistent in a controlled environment. Because of this, repeat benchmarks are actually recommended to ensure accuracy of data. The good benchmarking sites actually run their suites 3 times or more to ensure accuracy.
ADD: Also TRIM and garbage collection doesn't necessarily need time, they need idle. SSD's spend most of their time idle.

D) I do NOT use ATTO. However this is the Most widely published benchmark by manuf. ATTO was developed for HDD and as such it uses HIGHly compressable data - This is NOT real life. The overall compression rate of OS and program files loaded can only be compressed by a small amount. Also using Uncompressed data provides an advantage to SF based SSD that does NOT translate to real world. I prefer AS SSD as it uses compressed data for it's benchmarks, Still Not real world, But MUCH closer.
Click to expand...
I think you got it backwards there. SandForce controllers heavily relies on compressible data to achieve their speeds. AS SSD uses INCOMPRESSIBLE data. Regardless of compressible data or not, the Samsung 830 is NOT a Sandforce drive and DOES NOT rely on data compression for performance.
 
^
Quote
If the benchmark degrades over repeat tries, then it's not a benchmark. Benchmarks need to be consistent in a controlled environment. Because of this, repeat benchmarks are actually recommended to ensure accuracy of data. The good benchmarking sites actually run their suites 3 times or more to ensure accuracy
End Quote
Very true. But HDDs and SSDs are not the same. HDD you can just write over the same sector. SSDs use Blocks and can Not simply be writen over. As For Reviews, Yes they run multiple passes, However They Normally will do a Secure erease Between passes - Only way to get consistant results, Totally unneeded for HDD.. Stand By my statement. (For CG, I've seen post that recommend idle time of OVERNIGHT to allow for CG to clear a clobbered drive.

Quote:
I think you got it backwards there. SandForce controllers heavily relies on compressible data to achieve their speeds. AS SSD uses INCOMPRESSIBLE data. Regardless of compressible data or not, the Samsung 830 is NOT a Sandforce drive and DOES NOT rely on data compression for performance.
End quote.
I think you Misinterpreted what I said.
I stated "using Uncompressed data provides an advantage to SF based SSD" then said "AS SSD as it uses compressed data " Think that implies ATTO gives an advantage to SF while As SSD will show lower performance for a SF based SSD.

Yes the 830 does not use the SF controller, uses a Samsung controller.

Have 3 x M4s, 3 x Samsung 830, Plus 2 x Agility IIIs, and about 5 older generation 1 & II SSDs installed in 3 desktops and 3 laptops.
 
Thanks for excellent replies! I found the SSD was connected to marvell Sata III controller, so I changed it to Intel Sata III and now it show up in Intel RST. I have yet to do another benchmark, but will wait till tomorrow to prevent any damage whatsoever to the disk.
 
I ran the test today with Avast! and every other program off, and the Samsung 830 connected to the Intel controller. I used AS SSD Benchmark 1.6. The numbers were alot higher this time:

Read:

Seq: 507.60
4K: 21.96
4K-64Thrd: 282.46
Acc.time: 0.095ms
Score: 355

Write:

Seq: 393.34
4K: 71.40
4K-64Thrd: 88.15
Acc.time: 0.051ms
Score: 199

--Score: 737

Thanks for all the help!
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom