Just be aware that Boot Timer does not include the time that it takes for POST. Comparison to people who time it from power button to desktop load won't be valid. Still it does have value in testing windows loading.
Then again you all probably already were aware...
-
Yea, Boot timer is just the time it takes to boot windows, not post BIOS.
On another note, I was in another thread and was enlightened that I needed to install Intel Rapid Storage Tech program, and enable "write-back" cache. Well that seems to have done the trick. Check out the before and after numbers.And I thought I had my SSDs tweaked out!
Before:
After Installed IRST:
Installing IRST also dropped 3 sec off my already pretty good windows boot time of 16+ sec, Now it's down to approximately 13.5 sec (13.447 in sreenshot) with just this one program installed.
-
cookinwitdiesel Retired Bencher
Seems TOO fast as I mentioned in the other thread, you exceeded the 2x300MB/s limit of SATA II haha.
What tweaks are you using? The intel drives may benefit from tweaks that the sandforce ones will not but still interesting. -
Finally bought and installed a 60Gb Vertex 2 for my M15x.
Ran the ATTO benchmark and got these results
What do you think? -
cookinwitdiesel Retired Bencher
Looks about right to me
-
I was thinking the same thing on the 2x300MB limits.
as far as tweaks, the system was tweaked using the tips on page one of this thread, as well as Les' SSD optimization guide. Then I used JBBs tweak How To Improve SSD performance on Intel Series 4, 5, 965 Chipsets (JJB Tweak) . I saw improvements there as well. Then installing Intel® Rapid Storage Technology program is what really helped. I downloaded the top one on the list. Then opened it up after it installed, and clicked enable:
Also, scook, next time you run CDM, try running wprime 1024, 1 thread. After starting wprime, run CDM so they are running simultaneously. I have found through the JBB thread, that a load on the CPU helps. I saw improvements with just running iTunes playing a song, but to me it looked like a 10%-15% load was optimal. I have yet to apply stamatisx' tweak.rawhide85 likes this. -
cookinwitdiesel Retired Bencher
I will try out the wprime idea. Already have IRST installed and enabled that this morning. Made all the writes go up considerably but knocked down the sequential reads a bit.
I guess for my desktop I will also look into the Intel chipset tweaks since it is an x58 system. -
Looking forward to the results. Hopefully you should get a decent boost. Just running wPrime seemed to give a pretty good jump as far as the numbers go.
-
cookinwitdiesel Retired Bencher
is the 16kb stripe size working well for you?
That is what I am currently at as well, it seemed like the best balance for large and small files -
Yea, so far it's working really well. Though I han't really tried out any other stipe sizes. From what Les had tested it looked as though he concluded 16k was best so I've stuck with that. Most of my large files, data, music, movies ect goes on my 750GB drive, which will be replaced with a 1TB in the next week as the order was submitted yesterday. I've probably re-imaged these drives 6-7 times tweaking the raid setup, benching, installing apps, ect, so that I can have a complete and tweaked ready to go image. Not sure if it makes a differance, but after messing with the system benching and tweaking, I like to secure erase (takes me 5 min now) and start with a fresh setup since TRIM doesn't work. Not sure if it really makes a differance or not, but thats my train of though, start with all new unfilled blocks.
-
Just reran CDM for giggles. Came out even better. I wonder if these makes would be even better without the chipset issues.
Even the Seq reads were better this time around. I think maybe I need to cut back on my CDM. I don't want to have to RMA the drives too soon. Especially since I just RMAed a 40GB intel drive
-
-
cookinwitdiesel Retired Bencher
Not sure why my sequential speeds in CDM went down the drain but the rest looks good
Can you run ATTO please DR650se?
-
Any thoughts on why speeds in ATTO benchmark might fall at 256 up in the testing?
EDIT: See image below in thread. -
cookinwitdiesel Retired Bencher
Need to redo your image
-
With something to compare to now my write speeds drop off starting at 256 badly compared to what FirewaveZ is seeing for essentially the same drive.
Note: the first run was before the tweaks here, the 2nd after... they improved a bit with the changes but the drop is huge...
Now, I cloned the drive, would alignment cause that kind of drop THAT badly? I recall selecting 4k for alignment but since the drive was cloned does that even have any effect? -
Well done.
-
Here are two differant runs of Atto. Run 1 was just Atto, Run 5 was Atto+wPirme 1 thread.
RUN 1:_________________________________________________________________________ RUN 2:
I ran the bench several times, and at differant spots, the mark would just drop. It happened at differant sizes, so it seemed kind of hit or miss but gave a good rough estimate. One of the reasons I posted the two runs to show the differances. Scook, did you run Atto or CDM with wPrime? How did it impact your results? It would be nice if these intel drives didn't have such a disparity between read and write like those sandforce drives. Judging from the benches, they seem really well balanced. -
cookinwitdiesel Retired Bencher
I did it both with and without, the difference was minimal. The one I posted here is without wPrime running. I think ATTO is a little more telling of the overall picture than CDM which is why I wanted use it to compare to 2 sandforce drives against 2 intel drives.
I like the outcome
I am finally seeing why the whole internet is crazy about the sandforce drives... compare the IOPS for your setup and mine(columns on the right)
-
You mean columns on the left?
Yea, yours is more then twice my speeds. Which is why I think I might like Sandforce drives too lol. I think aybe I will sell these for some G2 Sandforce in the future if I get bored.
-
cookinwitdiesel Retired Bencher
I was actually happy with the speed of a single drive, I only did raid 0 because the price per GB was GREAT ($1.50/GB) and I needed the space.
The MicroCenter 64GB SandForce drives are proving to be a good value so far and do carry a 3 year warranty I believe.
And unless I am crazy.....they are on the right......(always had trouble with L and R as a kid haha). The L one is write IOPS and the R one is read IOPS -
Oh ok, I was looking at the it differantly I suppose. But the write speed is what I was focused on. Thats where the Sandforce makes its money. The writes almost match the speed of the reads where as intel has write speeds <1/2 of the read. You think the Intels would perform differantly on a differant chipset? Granted I doubted they'd be close to the Sandforce. Still a nice comparison to actually see the two side by side.
-
steviejones133 Notebook Nobel Laureate
Been reading a fair bit lately on SSD's as im thinking of buying one....
I have been toying with a x25-m g2 160gb...maybe two of them.....obviously read of G3 too but doesnt seem anything special in terms of performance over G2 ...only difference i can tell of is that the 25nm size should make the same size drive more affordable....although Intel prices for G3 are yet to be seen.
I can pick up a G2 for around £200 gbp atm ($310usd) so its quite tempting....
BUT, having read and read and read a bit more....seems like sandforce SSD's would be the better option and I just found this:
OCZ announces Vertex 3 SSD series | bit-tech.net
Im not in a hurry to buy, just when i DO buy, I would prefer to get the best SSD I can afford.
What do you guys reckon? - I would be pretty tempted by the vertex 3 if the price was not too disimilar to the G3 Intel... -
cookinwitdiesel Retired Bencher
The Intel is controller limited on the writes. Sadly nothing you can do to bring it up - the main criticism of them.
-
cookinwitdiesel Retired Bencher
Do not bother with the Vertex 3 unless you are putting it into a Sata III (6Gb/s) system.
-
steviejones133 Notebook Nobel Laureate
-
cookinwitdiesel Retired Bencher
If it is a sandy bridge part than the PCH will have native SATA III in it
-
steviejones133 Notebook Nobel Laureate
-
cookinwitdiesel Retired Bencher
It should work in a SATA II Slot just limited to around 285-290 MB/s
-
steviejones133 Notebook Nobel Laureate
Thanks Scook, anyone know if R3 has SATA III support yet or is finding out like spittin' in the wind....
-
cookinwitdiesel Retired Bencher
It IS Sandy Bridge so will have it.
-
Okay, had some feedback from the OCZ forums, I'm going to reinstall from scratch and check out the numbers.
One thing I am curious about, about how long do you guys think I should wait to reinstall the drive to give sufficient time for the drive to "recover" from benchmarking, etc?
Also, I read that 80% is about the absolute limit of utilizing SSD drive space... at what point are you guys planning on not loading the SSD up, ie 60% or 70% or so? -
I'm just going to keep going, maybe tone it back if I see a performance drop. You may also consider a security erase before reinstalling.
-
FWIW, in response to my asking about sizes, I was told that it was best to keep it at or below around 60%, with as much static data as possible... but man, that's a low number!
Interestingly, they also recommend putting the swap on a standard HDD if you can, supposedly performance is better in a dual SSD-HDD configuration... got me to thinking, now more RAM is even more a necessity to cut down on swap file use combined with SSDs.... -
Yea, my thing is, whats the point in having two SSDs if you can only fll up the amount of one SSD?
-
One question, besides secure erasing the drive, is there anything else from your reading and experience you would recommend to prep the drive for a fresh installation?
-
Not really. I just secure erase, and then re-image. Or you can do a fresh install, but with everything I have on, it just takes so long to tweak and customize, so thats why I image it. Secure erase will reset it to factory speeds.
-
-
Aristotelhs2060 Notebook Virtuoso
i installed intel rapid storage technology and my ssd speeds suck with or without it. i use ocz vertex 2. except the 4k reading the read and write reported is 206mb/sec and 86mb/sec accordingly! the highest speeds as per ocz is 280 read and 270 write. all those for single ssd. so whats wrong?
i am sad i paid for a new ssd once again. this machine is an ssd disaster. maybe it good only for intel ssds? -
Alex Intelligent Software - Downloads: AS SSD Benchmark
It shows a few interesting bits of info, it's safe, and the OCZ guys pointed me to it... but no need to run it yet, just let it get the info, and post a SS of the results.
Mine were like this:
Note the green number in the top left... that is insanely off for an SSD which pointed to the Acronis software having somehow accounted for the maintenance partition on cloning we believe. -
cookinwitdiesel Retired Bencher
-
-
OKay- I'm confused, I thought Write-back cache needed to be disabled for SSDs in general... is that not correct, or correct for the Intel SSD drives?
-
cookinwitdiesel Retired Bencher
no you want that enabled I think, it is different from the windows control I believe
-
-
Aristotelhs2060 Notebook Virtuoso
http://forum.notebookreview.com/ali...g-ssd-tips-tricks-benchmarks.html#post6127614 on post #4
Any clues guys? i am realy dissapointed! even my previous ocz vertex 1 120gb which is the third one which stopped working on this machine had much better write speeds and only 20mb/sec less read speed. And now i have OCZ Vertex 2 -
cookinwitdiesel Retired Bencher
With CMD the sandforce drives do MUCH better with 0fill or 1fill data vs random due to the compression scheme they employ. That may be that case as well with this AS program
-
Aristotelhs2060 Notebook Virtuoso
-
cookinwitdiesel Retired Bencher
Goto File>Test Data>1fill
then re-run the test
But I agree, even for random data on a sandforce drive those 4k numbers do look low
Here is a single 64GB SandForce drive of mine for comparison:
-
Mine are all currently on my M11x, so its not a good comparison.
You're offset looks good, and I am sure you have done the tweaks.
How much data is on the drive? From what I am hearing you want to be around 60%, 80% at most....
One issue with benchmarking is that quickly scores will fall with a lot of runs close together. You might let it idle (NOT SLEEP OR HIBERNATE) overnight, in fact you probably want to get rid of the hibernate file on a small drive and just not use that to allow more NAND space. I know when i dropped my space from around 80% to 60% my speeds above 256K improved somewhat.... I'm curious now to let the sytem idle for a couple days to let the drive fully recover.
Installing an SSD? tips/tricks/benchmarks
Discussion in 'Alienware' started by mfractal, Apr 9, 2010.