'Laptops w. Intel Series 5 chipset can not take full advantage of fast SSDs'

*** Update: there are two tweaks to handle this issue:
http://forum.notebookreview.com/har...rformance-intel-series-5-stamatisx-tweak.html
http://forum.notebookreview.com/har...ssd-performance-intel-series-5-jjb-tweak.html

The issue is summarized here:
http://www.storagereview.com/how_improve_low_ssd_performance_intel_series_5_chipset_environments ***

According to benchmarks run by several members it seems that laptops with Intel HM55 and PM55 can not take full advantage of fast SSDs. These chipsets are very common in modern notebooks. The performance hit is especially visible in 4K random read and write performance.

The problems seem to be caused by an agressive implementation of power saving features, as pointed out by tilleroftheearth in this post.

If you have this chipset in your notebook and a fast SSD, please post CrystalDiskMark 3.0 results here. Update for comparison, please post only 50MB test size, 3 runs, 4K results.

Here's a CrystalDiskMark result of Corsair Force 120 on HM55 chipset by member KolosoK:


Here's what the Force 120 performs on a different system:

( source)

Here's a AS-SSD result by KolosoK of his Corsair Force.


Here's what a Corsair Force is capable of:


Here's one by tilleroftheearth: Patriot Inferno (sandforce)


And a normal performing Sandforce:


Here are more threads with the same topic:
http://forum.notebookreview.com/asu...-g2-160gb-g73jh-why-sluggish-performance.html
http://forum.notebookreview.com/har...2379-my-patriot-inferno-ssd-performs-bad.html

 

This might be specific to certain SSDs, in my case for example it isn't true. Laptop in sig.

 

Yup. If I read up on this before purchasing my SSD, I may have just gone with a Momentus XT instead. However, the SSD was technically a gift from my family, and it did improve my loading times (both Windows and gaming) tremendously, so I'm happy with my Corsair Force 120gb. I just wish they didn't spend so much on something that will not perform up to full spec on my machine. Not the drive's fault.

EDIT: Phil, it seems that the drive is not the OS drive in the benchmarks you posted - this might skew test results a little bit. The other benchmark probably used the drive while it was empty (confirmed: look at the screenshot - 0/112GB used).

 

Your performance does look a bit better but still your the 4K random performance is 20% lower than mine. Your 4K random write is about 50% lower.

This is my C300 64GB on a Intel GS45 laptop:

Yeah may be true but nonetheless, Sandforce SSDs should get around 20MB/sec 4K random read, while the capped ones get about 12-14 MB/sec.

 

Wrong drive, Phil The "normal" one is the OCZ Vertex 2

EDIT: Ah, never mind, you just said it was a "normal" performing one, not the exact drive :s

 

Thanks Phil for dedicating a new thread for this issue.

---Nvidia chipset-------Intel PM55 chipset--


This is the same SSD and you can see its performance on two different chipsets
The problem is obvious on the PM55 chipset
Laptop: Alienware M17x-R2

 

I'd like to see some comparisons with intel drives.

You'd hope intel's ssds would be compatible with their own chipset

 

Stamatisx' results right above your post are from an Intel X25-E.

Seems like Intel chipsets HM55 PM55 can't take full advantage of the Intel SSD.

 

Noooooo!!!! *goes on a long rant and cursing streak*

Angry geek is angry!

Mr. Mysterious

 

Here's a Vertex 2 result by rankinging.



Laptop: Toshiba X505-Q885.
Chipset: PM55

4K random speeds are limited like expected. The sequential speeds are limited too. Something else going on.

 

well this is quite unfortunate *looks over at his envy 14 and it's hm55 chipset*


need to decide soon on a hdd too. wd black 500gb or momentus xt 500gb or "screw it, get a decent 320gb to replace the dying one you have and get a nice intel ssd sometime after christmas"

this makes me wanna knock the ssd option off the table til a new computer some 3 years down the line.

 

Phil, check this thread as well with the Asus
http://forum.notebookreview.com/asu...-g2-160gb-g73jh-why-sluggish-performance.html

 

Will a PM45 take advantage of the SSD's full potential?

Mr. Mysterious

 

Honestly I don't know, I haven't seen any complains concerning this chipset so I guess there is no problem with the PM45 and SSDs performance.

 

My GS45 performs fine with SSDs. I would guess PM45 is comparable.

 

Comparable? What does that mean? Do you mean that a PM45 will be able to take full advantage of intel SSDs?

Mr. Mysterious

 

I expect it will. If there were problems with PM45 we would have heard about it before.

 

Yes, there is a good chance that the performance of the SSD will be as expected. Your best chances are to find people with that chipset and a fast SSD and ask them to perform a couple of benchmarks. This way you will be sure that there is no problem.

 

If at all posible, the data should be collected in a spreadsheet format so that we have a shot as identifying the testing variables.

Such a spreadsheet could also be used by other users to test their own particular laptop/os/chipset/ssd combo and contribute additional data to the analysis.

As presented and discussed, there really isn't enough information to come to any kind of conclusion other than that some users are seeing some kind of problems with some chipsets under some conditions with some benchmarking tools.

 

whoo! There's hope for my machine then!

Mr. Mysterious

 

Actually every HM55 and PM55 based laptop shows more than 20% lower 4K random read and writes. I haven't seen any exceptions.

Here are more threads on the same topic:
http://forum.notebookreview.com/asu...-g2-160gb-g73jh-why-sluggish-performance.html
http://forum.notebookreview.com/har...2379-my-patriot-inferno-ssd-performs-bad.html

 

Is this a driver problem or an hardware/firmware one?

 

I don't know. Maybe stamatisx has an idea, he's been looking at this issue for longer.

 

There is no way for us to exclude the BIOS or the design of the chipset unless one of the manufacturers manages to isolate the problem (we don't have access to the BIOS or the chipset drivers). In safe mode the speeds are capped at 18MB/s 48MB/s respectively and that indicates a power saving feature behind the problem that could be solved with a BIOS update but still, if you see my post with the numbers of my SSD they are far from what they should even in safe mode

 

With the same OS? With what patches/updates? With the same driver levels? With the same CPU? What BIOS level in the machines? With what benchmarking/measurement tools? On battery or AC? Have the power plans been 'hacked'? How much free space on the drives? How much page space configured? Has defrag and search/index been turned on/off? Is TurboBoost, ReadyBoost, or other cpu/memory 'assists' enabled or disabled? Are there other SATA devices in the machine? Was testing done in safe mode or not? And so on.

Too many undefined/unknown variables here.

 

The results sofar are consistent, over all the unknown variables.

 

The same results with an i7 820QM and i7 940XM (with and without overclock)
With the most commonly used benchmarks the results are similarly low.
Either on battery or AC it made no difference
I have tried every possible power plan combination hacked or not
With 80% free space
Every possible tweak to increase the speed (index off, 8.3 check off, defrag off, readyboot off, restore off, etc...)
Benchmarked with and without other disks (I have tried even with the RAID controller)
For the safe mode see a couple of post prior to that.

 

Is this issue only related to notebook chipsets or could the problem exist with desktop motherboards?

Harvey

 

There are references concerning desktop chipsets with the exact same results but they have the option to disable powersaving features like C1E and EIST through BIOS that eliminates the problem (something that our BIOS doesn't have as an option)

 

Here is the same drive, but CDM ran in safe mode:



So it seems that the 4k read/write is what's taking the hit.

 

128gb C300 on my Vaio Z11. replaced ODD with ultrabay adapter

safe mode results are quite a bit higher though

 

no there should be no problems.. nver have been... this is purely a chipset issue.. once CPU is put under full load , the SSD's perform properly as was found in the Asus G73 SSD problem... we need a better BIOS or intel needs to update its chipset drivers.. someone contact them.

 

i am pretty clueless technologically and just stumbled onto this thread while researching what type of hard drive to get so I may be way off here but does it have anything to do with this?

 

Will someone email intel about this so they can release some new chipset software?

 

It is not the drivers !

Must be something in the OS or in the Bios. I have tried many different Intel chipset drivers and Microsoft default drivers (msahci) and there is no problem there.

I have an XPS 1530 (T8300, 4GB ram, ICH8M) laptop with an Crucial C300 256GB.

Here are the results of C300 256GB from my XPS (windows7 x64):

and these are from the same SSD (I just remove it from the XPS and connected to my desktop which has Win7 X64 and Intel ICH9R) :


And these are the results from the XPS laptop under safe mode :


So the problem is not just with PM55 HM55 but also with ICH8M

(here is my benches and impressions from C300 256GB for a Greek forum and also my problem explained google translate)

 

Carlos, is your chipset a PM55? In safe mode your numbers seem better than the rest of us and I wonder why?

 

You should try ATTO 2.46, it´s the only benchmark that measures correctly the speed of the SF-1200 controller.

I get 250+ Mb/s, exactly what OCZ says on the box.

 

Yes, PM55, and i want my safe mode numbers in normal boot.

 

There is no issue with the PM45. This is what is in the P7805-u and with a Sandforce 2 I get full normal performance. I have seen many others post about this and determined a while ago it seems specific of the mobile variety of the chipset. The desktop chipset version does not seem to suffer the anomaly.

It seems Q32 is not too bad when you saturate the bus but Q1 is bad. Since safe mode does not run in lower power modes I would suspect that. I think even in safe mode the numbers are a bit crippled but not all too bad.

What surprises me is Intel wouldn’t have tested this out with their reference design. Their own G1’s and G2’s have been around for a while and will show the issue as well. In all it would be nice to hear from them on this issue…………..

 

Thanks Phil for pointing this out. I was desperate to find a way to improve SSD performance in the last few days. I think the problem is with the driver.

Motherboard Name Toshiba Qosmio X505
Motherboard Chipset: Intel Ibex Peak-M PM55, Intel Ironlake-M
BIOS Type Phoenix (06/11/10)
IDE Controller Intel(R) 5 Series 6 Port SATA AHCI Controller
Disk Drive OCZ-VERTEX2 60G


Here is the ATTO benchmark in SAFE MODE they look normal...

Here is the ATTO benchmark in NORMAL BOOT they look slow...

 

Yeah, most probably a BIOS update or new drivers would give a solution to those low results we all experience.

As you can see CarlosGFK's results with the C300 on a Clevo with the PM55 chipset, in safe mode he gets numbers much higher than those 18MB/s 48MB/s capped ones we get with the rest of the laptops. (That's why I suspect the BIOS)
Concerning the rest of the chipsets and their low performance, the PM55/HM55 seems to suffer the most

 

Exactly.

The fact that safe mode gives much better results could also indicate a driver problem right?

tilleroftheearth also showed an improvement in safe mode: http://forum.notebookreview.com/har...t-inferno-ssd-performs-bad-2.html#post6631553

It's a much smaller improvement though.

 

Possibly, but I have tried every possible driver available with no success. The results were the same. In safe mode there aren't the same power saving features so we get the performance we get, without having to put some load to the CPU like in normal mode.

 

We should try to figure out if Macbooks have the same problem. They use PM55 chipset afaik.

Sequential speeds aren't the problem. It's 4K random read and writes.

 

I will try to use iozone in linux and reproduce the problem. Maybe in linux we get better performance. I don't know when I will do that though, cause I have to format the disk and install Linux on it.

 

stamatisx, is their a Macbook Pro with PM55 chipset? I can't find it on Google.

 

Core i5 MacBook Pro teardown shows modified Intel chipset | Electronista

 

Yep. "Apple have also modified the standard Intel HM55 Express chipset".

I wonder if they will see the same performance hit.

 

Unfortunately Phil my knowledge when it comes to Apple and the issues concerning their products is limited. I guess input from those users would be helpful at this point

 

Yes I started threads on Macrumors and our Apple forum.

Notebooks with Intel HM55 chipset show capped SSD performance. MBP 2010 affected too? - Mac Forums

http://forum.notebookreview.com/app...s-intel-pm55-chipset-affected-same-issue.html