SSD Performance affected by benchmarking?

GullyFoyle

Diamond Member
Dec 13, 2000
4,362
11
81
Do benchmark programs such as AS SSD benchmark degrade perfomance of the drive they are testing by the act of executing their write tests?

I was reading another thread in this forum and it mentioned the AS SSD benchmark program, and how it checked drive alignment.

So I installed it. I ran the benckmark and got a final result of 148. It also said my drive was not aligned on a 4K boundry (actually, none of my drives/partitions were).

I went out and got this Paragon Partition Alignment Tool and ran it. Re-run the benchmark. The offset is now good. But now my benchmark result is 108. Down by a third. :confused: (I can't provide any benchmark screen results until tonight at the earliest)

So, what wrecked my performance? Running the benchmark tool twice, or running the Paragon Partition Alignment Tool?

I was thinking maybe making a backup, re-initialize the drive with HDDErase, then reload from the backup? Using Macrium Reflect (free)? Would this preserve the correct disk alignment? Would it "straighten out" the disk similar to a defrag or restore it sector by sector, restoring any disarray existing on the disk?

My SSD is a Kingston 64GB SSDNow V Series. It is less than half full.
My motherboard, an abit ip35e, does not support AHCI.
I'm running Win 7 x64. I checked early on and it said TRIM was enabled.

Thanks for any feed back you might have.
 
Last edited:

wpcoe

Senior member
Nov 13, 2007
586
2
81
While I can't answer your questions, your post piqued my curiosity, since I have the same Kingston 64GB SSD. I'm not familiar with the "AS SSD Benchmark," as I just now downloaded it and ran it after reading your post.

When you say "a final result of 148," which result are you reporting? Is it a combination of the three unmarked blue boxes in the "Score" area at the bottom of the window? If so, my 64GB Kingston SSD (after about a month of regular use) gave a total of 137.

What exactly do those numbers in the three blue boxes mean, anyway?
 

GullyFoyle

Diamond Member
Dec 13, 2000
4,362
11
81
Here is a sample image (not mine):
as-ssd-bench-english.png


When I said final result, I meant the number in the bottom box.
I'm sure the number at the bottom is the numbers from the top, put through some mathmatical fromula.

It's really only meaningful as a comparison to other scores from other systems.
Mine had high numbers (200/300) in the Seq test, then teens in all the 4K tests, like 0.3xx/0.2xx in the access time on the final test yesterday.
 

wpcoe

Senior member
Nov 13, 2007
586
2
81
Crikey, considering we're using the same Kingston 64GB SSD, my test results (77) are a bit startling compared to your test results (108-148)!

as_ssd_kingston.png


Compare with a recent CrystalDiskMark:

CrystalDiskMark_SSD.004.png


How do those numbers compare with yours?
 

=Wendy=

Senior member
Nov 7, 2009
263
1
76
www.myce.com
There is no faster way of degrading the performance of an SSD than benchmarking it, especially if you use AS SSD and CDM. CDM with the default settings will write nearly 1TB of data to the drive in 12 runs.

Only Indilinx based drives with their aggressive GC will recover the performance in a few hours, the rest, including Intel, Micron, and Sandforce, may take days to recover, if they recover at all.
Sometimes the only way to get performance back after doing a lot of benchmarking is to secure erase the drive.
 

sub.mesa

Senior member
Feb 16, 2010
611
0
0
If you have TRIM capability then the file that AS SSD/CrystalDiskMark wrote to will be truncated; without TRIM like on RAID0 arrays on windows yes benchmarking would slow down the drive; but normal usage does that too.

@wpcoe:
you have a JMicron SSD; the brand may be Kingston, but the controller chip is labeled JMicron. Some models actually laser-etched JMicron off and put Toshiba on it, to obscure the fact that it is powered by a JMicron controller. The JMicron controller had caused 'stuttering' SSDs in its first generation (JMB-602) and therefore got its bad reputation. More recent chips from JMicron had this problem under control, but it remains the 'budget'/'value' option amongst SSDs and concertainly not as advanced as Intel controller.

For example, JMicron does not support NCQ. NCQ is needed for SSDs to receive multiple I/O commands at once. With NCQ disabled, the 4K random read and 4K-32/64 scores as seen in CDM/AS SSD would be about the same just like yours is. On systems with AHCI/NCQ enabled and with an SSD capable of NCQ (Intel, SandForce, Micron, Indilinx) you would see much higher scores for 4K QD32/64 reads (not writes; that works differently).

Intel controller has 10 parallel flash channels. In benchmarks this could be seen as the 4K QD32/64 scores would approach the theoretical benefits of 10 parallel channels, and thus is up to 10 times the score of the normal 4K test with only a single queue.

I will add, though, that while it may be one of the slowest SSDs, it is still a giant leap faster than any HDD regarding latencies. Booting and starting apps on any SSD is way much faster than a fast HDD. Even cheap shitty compactflash could beat a Velociraptor or even 15k rpm SAS disks in some latency-sensitive workloads; even though they are limited to something like 20MB/s throughput. The latency aspect here is key.
 

GullyFoyle

Diamond Member
Dec 13, 2000
4,362
11
81
Crikey, considering we're using the same Kingston 64GB SSD, my test results (77) are a bit startling compared to your test results (108-148)!

Compare with a recent CrystalDiskMark:

How do those numbers compare with yours?

Well, I warned you that running the benchmark might degrade your performance, and you went ahead and did it anyway...Thanks for validating my results. Your first result was 137, the second it went down to 77. Are you up for thirds?

I guess I'll have to switch this off my boot drive to secure erase it and re-image it. What a pain in the arse.

I'm gonna check the stickied post in the forum and make sure a warning is included about benchmarking!
 

skid00skid00

Member
Oct 12, 2009
66
0
0
I used AS SSD to bench both of my gen 1 80 GB Intel X25's weekly, for a YEAR, and only saw the typical, expected degradation of a non-GC, non-trim drive.
 

wpcoe

Senior member
Nov 13, 2007
586
2
81
@sub.mesa: Thanks for the recap of the capabilities of the Kingston post-gen1 drives. (Mine is a -425 model, the -125s are the Kingstons to avoid, I understand.)

While it is true that the Kingston is not the top-of-the-performance-chart SSD, it does offer great value for its price. Where I live (Thailand), and when I bought the drive, these were the only options available:

ssd.gif


When I bought the drive, I thought that the Intel would be too small, but in retrospect, I probably should have gone for it. My boot C: drive with Win7 Home Pro and all my apps, with everything loaded, and including a 3 GB hiberfil.sys and a .5 GB pagefile.sys, uses only 22.2GB. Oh, well.

@=Wendy=: Here's a nice little bonus from the JMicron controller on these Kingston SSDs, quoting from Review of -325 (2nd gen) 128GB Kingston:

"I don't have and pretty charts or graphs to explain this next part, but I will share an observation I made during my fragmentation testing. When running my fragmentation tool, I observe IOPS drop as the drive becomes more and more overloaded with the task of tracking the random writes taking place. Here the JMicron controller behaved like all other drives, but where it differed is what happened after the test was stopped. While most other drives will stick at the lower IOPS value until either sequentially written, TRIMmed, or Secure Erased, the JMicron controller would take the soonest available idle time to quickly and aggressively perform internal garbage collection. I could stop my tool, give the drive a minute or so to catch its breath. Upon restarting the tool, this drive would start right back up at it's pre-fragmented IOPS value.

Because of this super-fast IOPS restoring action, and along with the negligible drop in sequential transfer speeds from a 'clean' to 'dirty' drive, it was impossible to evaluate if this drive properly implemented ATA TRIM. Don't take this as a bad thing, as any drive that can bring itself back to full speed without TRIM is fine by me, even if that 'full speed performance' is not the greatest.

This type of self-healing (i.e. without needing TRIM) is great for those wanting to run a few SSD's behind a RAID, since no RAID implementation is currently capable of passing TRIM from the OS to the arrayed SSD's. Better yet, considering this drive is tailored to the budget crowd who may very well still be running XP or Vista, it's good to have a few choices that don't require TRIM to maintain decent levels or performance.

I can't help but think this background task might have been so aggressive it was getting in the way of normal operation of the drive, and might even be some of the cause for the poor IOPS performance seen throughout our testing. What should have been idle activity might have been taking while the drive was actually under read/write load from the OS."

So, while relatively poor in some areas (like writing), at least in garbage collection it excels. I don't regret getting the drive at all, even if the OP gets far superior ratings on his drive from the AS tool. :D

PS: Hmmm. As I was reviewing my post and debating whether to actually post it, because I thought it was going too far off-topic, I glanced at the thread title, and actually the excerpt from the review answers the question "SSD performance affected by benchmarking?" by saying that basically for these Kingston SSDs, the answer is "No, these drives are not affected by benchmarking," as far as rating performance.

(However, benchmarks might be affecting the life span of the drive by writing tons of data on cells which have a limited number of read/write actions in their lifetime....)