Well it looks like he went from 13/17 to 22/36 on the last images he posted so that's about 10MB/s gains.... I just don't know what it should be for the C300
-
yeah maybe if we had a database for each ssd It could help to understand what external factor improve the 4k results as well like the one in the first page of the DIY Vidock thread
-
Again, what is the C300's specified maximum 4K random speeds? That is what you should be looking at to know if your close to full performance, not other peoples benchmarks... -
agree with you JJB because of the so many factors which could have some effect over the results when comparing different laptop but the full performance on a single laptop not totally as there is no proof that the safe mode result can't be surpassed by using any tweak (at least for what I know)
An example that I see would be an hardware upgrade or a driver upgrade which could increase the 4k result in normal mode which could eventually surpass the old safe mode 4k result of CDM benchmark
It's maybe not exactly the right example but I hope you see what I mean by that
To understand better the aspect that I'm bringing here is that we could see your thread in an other way as there are several possibilities to see it:
How To Improve SSD performance on Intel Series 5
is it JJB tweak only?
or JJB tweak + any other tweak (except the stamatisx one)?
Anyway, I wanted to thank you many many times as without your efforts I wouldn't have a perfectly working laptop right now -
-
-
is it this one?
download.micron.com/pdf/datasheets/realssd/realssd_c300_2_5.pdf
random 4k read : 50,000 IOPS
random 4k write (128GB) : 30,000 IOPS
Typical I/O performance numbers as measured using Iometer with a queue depth of
32 and write cache enabled.
4K transfers used for READ/WRITE latency
values.
Those results have been obtained in specific conditions so they could be different depending on each specific laptop and on the tools used to get those results but they should give us at least a good idea of what to expect
here is how to convert the results :
IOPS * TransferSizeInBytes = BytesPerSec (with the answer typically converted to MegabytesPerSec)
which should be:
IOPS * TransferSizeInBytes / (1024*1024) = MB/s
or here we have 4k = 4 * 1024 bytes (they wrote K and not k in the datasheet)
so
random 4k read (max value) = 50 000 * 4 /1024 = 195.3125 MB/s
random 4k write (max value) = 30 000 * 4 / 1024 = 117.1875 MB/s
(pending results waiting for confirmation)
looks quite high some mistake maybe
Any PHD in mathematics here? -
I think you did that math correctly as far as I can tell
I am getting 158MB/s Read and 97MB/s write with the tweaks on my intel drives. Remeber you need to look at the 4K QD32 CDM results not the 4K (QD1) numbers.... And IIRC the C300 is supposed to have higher 4K speeds.
Edit: Just checked Sean473's results for 4K QD32 = 180 / 130 so he is close to specified max speeds with the tweaks assuming the math is correct. -
I don't see the 4k QD32 in the datasheet
-
Haha I had that in my previous previous post and I haven't seen it
Maybe Intel has decided to put the average value rather than the max value in the datasheet
For the C300 they have taken the max value
Ok it doesn't work for me neither
I get 158.2/118.2 QD32
but the Intel values are : 36.72/12.9 (35K/3.3K IOPS) -
Well intel is typically quite conservative with their specs so that kind of makes sense, but it is quite a large difference, more so than the other seq numbers which test out very close to the specs...
I bet a lot of it is as you said above, variations in testing, platforms and measuring methods... -
I think I have the answer
IOPS - Wikipedia, the free encyclopedia
Intel's data sheet claims 3,300 IOPS and 35,000 IOPS for writes and reads, respectively. 5,000 IOPS are measured for a mix.
I don't know what mix means
EDIT: here is what is written above
In a test done by Xssist, using IOmeter, 4KB RANDOM 70/30 RW, queue depth 4, the IOPS delivered by the Intel X25-E 64GB G1 started around 10000 IOPs, and dropped sharply after 8 minutes to 4000 IOPS, and continued to decrease gradually for the next 42 minutes. IOPS vary between 3000 to 4000 from around the 50th minutes onwards for the rest of the 8+ hours test run. [4] Even with the drop in random IOPS after the 50th minute
It looks like it's due to the 50MB size we have chosen in CDM -
Look back at my copy from intels data sheet, the read speeds match (35,000) but the write spec is 8,600 (for the 160GB) and 6,600 (for the 80GB)
Don't know where the 3,300 came from unless maybe the older G1 (50Nm) drives....
Edit: never mind, the 3,300 was from the older x25E not the x25M -
look at my edit above
but still the C300 doesn't get the same gap
I already heard about this sustainability problem but it was common to all consumer ssd so it doesn't explain why it's just the Intel ssd which the formule doesn't work.
Maybe as you said Intel is more conservative than others and has decided to put the right values -
The X25E G1 is an antiquated (50Nm) model from intel with the original controller, they don't even have the data sheet availble any more. A lot of those issues went away with the newer 34Nm devices and newer controller, firmware , more onboard cache and TRIM support....
Wow I can't believe I just called a 2 year old state of the art product 'antiquated', how times have changed..... -
If I'm reading it correctly, 117MB/s seems quite low for a "max" C300 4k QD32 write speed.
From earlier today:
-
I don't think it went away if I remember the article it was an unsolvable problem only the professional ssd wasn't concerned by this
They had tested something like 10 different ssd if I remind it correctly
I'm trying to find this article -
Anandtec thouroughly praised (and tested) the newer 34Nm G2's and IIRC said that issue was no longer a problem on the new drives. I will go with his findings as he is who the engineers from Crucial and intel call for advise...
-
@5482741
Look back at page 5 where sean473 posted his results for a 128GB C300, the score are much lower than yours. Does the smaller size SSD have lower performance numbers than the larger capacity drives (like intel)? Or are their didderent grades of C300's ? -
you're talking about the trim technology?
the sustainability in the ssd life time or during a single transfert
which is not the same thing. The last problem require a cell redundancy which lack in consumer ssds. It would require to increase the price to solve the problem
Is it this problem you're saying has been solved since? -
They are in line with Phil's pre-tweak 4K speeds on his 64GB C300 as well:
With my previous post I was actually referring to the calculated maximum 4K QD32 speeds from Post #57 as being low. -
ok i have found the source regarding the sustainability problem
It hasn't been solved since the article is dated of april 2010
the article has been written by Logan Harbaugh
He named that problem The 'write cliff' effect
here is the article
Storage smackdown: Hard drives vs. SSDs
but there are some improvements since june 2010
http://www.computerworld.com/s/arti...tart_up_claims_huge_boost_to_SSD_reliability_ -
@5482741
Thanks for the clarification.
@erig007
I was just listing all the upgrades in the intel drives when they went to 34Nm, and in total they greatly improved overall sustained speeds as well as wear leveling.
I'll read that article soon, thanks for digging it up along with the info on the isreali drive info, looks like our drives will be antiques in the near future -
oh yeah even sooner than you think
I read somewhere that some faster ssd from ocz and others would come this year before christmas -
On PM45 (no tweaks):
@ JJB, I've changed the other thread's title to include more chipsets. let me know if you want that to be done for this thread too. -
@Phil
I just noticed the last line on your post above. Feel free to change the title as you see fit, it makes sense to me to add the other chipsets that this may work with.
Thanks! -
Has this been tested to see if there would be any performance impact on a Momentus XT drive?
-
One person got better results with an XT. I believe he used Stamatisx tweak.
-
.. thanks a lot JJB! Just one more thing.. How do u apply these tweaks in lets say a power saving profile when u need a lot of battery but proper SSD performance?
Thanks
Sean473 -
@Sean473
I created a custom power plan:
Click battery icon > more power options > creat power plan (left side column) > Select 'Power Saver' and name it > next > create.... Then select the new power plan name you created > change plan settings > Change Advanced power settings...
Then just adjust the tweaks for processor power 'on Battery' to the same values as before and you can adjust any of the other options (except processor) any way you want for max battery life.
I actually went through and adjusted the 'plugged in' settings in my custom 'power saver' plan with the tweaks but changed the wifi and pci etc to max performance. This seems to still allow the SSD improvements and my temps dropped a couple of degreesI have been using ths for 2 days and don't notice any difference other than slightly lower temps... (as compared to the high performance P.P. tweaks)
Also if you use Battery Bar Pro you can set your computer to boot with whatever power plan you want and also select which plan it switches to when you unplug. I like this so I don't have to remember to change power plans when I reboot or unplug, I have max SSD performance all the time with no adjustments needed. -
i think i'll skip it.. i just have a high performance plan.. i don't need full SSD speed all the time.. Anyways , really thank a lot for your tweak. Really has helped with performance
-
NotebookGrail Notebook Evangelist
@JJB and Phil, Wouldn't it be a good idea to include a link or write up that explains the fundamental tweaks that need to be done for a SSD (disable prefetch, superfetch, enable TRIM, etc). The reason is over time people coming to this thread from 'Google search' will just blindly do the things mentioned in the thread and forget the fundamentals.
I will post this message in the Stamatisx tweak thread as well. -
@NotebookGrail
Good idea. What do you think about adding a link to THIS post for gaining some space back?
Well I thought I had a link to another one also that walked you through all the basic setup stuff (enable write cache, disable surperfetch etc.), Their must be dozens of basic guides, any links that you recommend for this? -
Those tweaks only really need to be done if you are running anything other than Windows 7.
If you are running Windows 7, it automatically does all of that for you upon installation. For instance, I already hadsuperfetch and prefetching disabled, write caching was automatically enabled, disk defrag was scheduled only for my external mechanical drives (my ssd boot drive didn't even show up on the list so you couldn't even 'accidentally' put it on a schedule). If you format any ssd using Windows 7 disk management tools, all partitions will be automatically and correctly aligned. The OS was literally made for ssds.
I didn't have to do anything and 80% of the tweaks suggested by others on various forums were either already in effect or they made my PC worse in some way (for example, one tweak disabled intelppm.sys service loading which amongst other things stopped my cpu from turbo boosting). -
If your refering to my link above I think you missed items 1,2,3,5 and 6 (if you have a recovery partition). those will gain you back 5 to 10+ GB of valuable SSD space and none of them are done by default in Win 7.
-
Nah sorry, I was talking about the stuff like write caching, prefetching and superfetch. The second paragraph of what you wrote. In reference to the link you posted, I still think its mad to turn off things like System Restore and I'm glad he agrees. I've only had my laptop for a couple of days and I've already used it once.
I'm a compulsive fiddler and tend to break things trying to figure out how they work. System Restore has saved me alot of time in the past and I'd put it in the top 5 best Windows additions no doubt. -
LOUSYGREATWALLGM Notebook Deity
PS: please include your system specs
EDIT:
Nvm. Just saw your recent post.
FYI: Disabling System Restore is a must when you are on a SSD (this is just one of the important tweaks) -
got a problem.. turbo boost not working.. its stuck on 1.87GHz whatever i try... even tried super pi which stress 1 thread but it doesn't go up to 2.8GHz.. what to do? I'd done this tweak but nver noticed this...
-
LOUSYGREATWALLGM Notebook Deity
@sean473
Maybe the beta driver you used (intel) caused this? -
removed it.. still got the problem.. i'l recheck the tweaks and reinstall the driver.
Update: Its the bloody power 4 gear.. messed up the whole processor when i installed new version.. going to use the old one.. -
@sean473
Make sure when you run super pi that you have no other apps running. I noticed with just IE8 minimized I lost full Turbo boost speeds (with and without tweaks). If another thread is active I only see 2.8Ghz but with hit full 3.06Ghz with just super pi and the intel turbo boost monitor gadget running....
FYI, all the CPU benchmarks built into Everest give me equal or better results with the tweaks than without. Freeware version HERE
EDIT: Just saw your update. What is this P4G and why do you even need it? The normal intel drivers with your CPU should give you full performance, unless it's something specific with your machine.... -
the power4gear is an integrated Asus app for power settings... i just like having it
.. the newest version fixed the problem.
-
it seems the problem is with the tweak.. basically , those other settings... other than the 2 main ones seem to be causing the problems for me.. the settings i only adjusted were Proc. performance core parking overutilization Threshold and Proc. idle promote threshold to 100%... the other settings seem to make turbo boost stick at 1.87GHz and not up to 2.93GHz... With these 2 settings changed , i have no problems with turbo.. the rest seem to be causing the problems... my performance wasn't impaired.. was almost the same as with all the tweaks... in fact 4K writes improved... slightly...
-
Hi JJB I did your tweaks and these are the results. How do they look? I'm new to this benchmarking softwares and unfamiliar with the numbers
Attached Files:
-
-
LOUSYGREATWALLGM Notebook Deity
EDIT:
@ArchEnemy and JKleiss
You can set the CMD to 3 cycles and 50MB instead of 5 and 1000MB to reduce the wear on your SSD.
EDIT 2:
@JKleiss
Can you bench at 50MB (tweak applied + music)? -
Is AHCI enabled in the BIOS?
Did you install Intel RST 10.0.0.43?
Is write caching enabled?
PS. there is no need to play music if you applied the tweak.
Please use 50MB filesize to reduce wear. -
-
Strange, your 4K do look low. I don't know what it is.
You could try to see if SSD Tweaker makes a difference. I don't know if it will.
How To Improve SSD performance on Intel Series 4, 5, 965 Chipsets (JJB Tweak)
Discussion in 'Hardware Components and Aftermarket Upgrades' started by JJB, Sep 14, 2010.