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

I wish I could JJB unfortunately it's not my post, it's LOUSYGREATWALLGM's post

 

Check CPU core clocks (idle) if stable or equal to default and a few more CDM tests for this finding "the ideal setting seemed to stabilize the variance in CDM runs and kept the idle wattage from jumpimg around...."

I would like to suggest to reboot and wait for the startup apps to finish loading before test.

PS: Just want to make sure Single gives poorer performance (your settings)

I'll try to retest Single vs Ideal too (my settings)

 

OK sorry about the confusion on my part, LOL... your avatars look similar (kind of)....

I'm rechecking this setting, I think i changed it because it restored full turbo boost in super pi, this may take a while since I have a meeting coming up shortly, will get back in a few hours..

 

Good work guys!

i bet this future post JJB will make including everyone's work will help a lot of SSD users in the future!

+1

 

OK, I'm back and have verified that the 'ideal' setting seems to be better overall than the 'single' setting (at least for me. 3 reasons; 1) More stable clocks at idle 2) Slightly lower wattage with less variance at idle 3) More stable 4K & 4K 32 readings during CDM (less variance).

Below are my results running 3 different CDM runs of (50MB x 5) for comparison. The first was at full idle nothing else running. The second was the same but with iTunes music playing (~2% CPU load). The third was CDM while running Super Pi ( ~26% CPU load).

The runs were made after reboot and waiting for all things to load, then I ran intel toolbox TRIM manually before starting. This was only done once for each setting and then I ran all 3 variations in the same order both times.

I also confirmed that Turbo boost is working in both settings, ran Super Pi at the 24M setting at full idle with just the intel Turbo gadget and HWmon for temps. Stopped both runs at the 8th cycle and results were within 1 second of each other. Also was hitting 3.06Ghz intermitently in each setting. Temps were about the same ~60/50C average.

So for the reasons above I am recommending the 'ideal' setting. Also note that in the image below the highlighted 4K run was an anomaly there was a huge variance in that run (did it twice) it's just showing the highest of the 5 runs and the lowest was ~38 on the write, did not see this big of swings on the 'ideal' run at 4K.

Settings for each are at the bottom of each CDM screen:



Note: It is easy to gain full Seq and 512K R/W speeds by adjusting the two 95% settings upward, but this will raise idle temps and wattage.

 

Okay. Ideal it is. One more question though, was the 4K read/write still showed big swings (Single-Idle) on your recent tests?

Agreed. I got the same results (1sec difference)

 

Is it normal for the core temps to increase by +15 to 20C when running Hyper Pi.?
In a QM57 chipset, with a i7-840QM processor, the idle temps are between 53-58C.

 

With HyperPI yes, with SuperPI no

 

Thanks.
Here is my results with 3 modes- all @ Ideal. By the way, the Max turbo i get is 2.93GHz. Is there any tool to get to the Max turbo? Mine is advertised to go to 3.2GHz

 

Yes, the 'ideal' 4K runs were much more stable than the 'single' runs.

I also ran Super pi in the balance power plan with no tweaks and both the ideal and single settings, rebooted before each run (16M setting):

Balanced no tweaks = 6m 23.8sec
High Perf w/ 'ideal' = 6m 26.0sec = + .58%
High per w/ 'single' = 6m 28.4sec = + 1.2%

I am writing this off as 'margin of error, or if if not I am not worried about a .6% drop in max CPU speeds...

I also redid the above posted CDM post to include my 'Balance' no tweaks CDM speeds for comparison. Oddly my 'Balanced' write speeds are much higher than they were before I started all the tweaking, I have a feeling I may have done some inadvertant tweaks in the 'balanced' plan when I thought I was in te High perf plan.... Oh well. The 4K write used to be around 22MB/s max and now I see 27 - 29MB/s....

Here's the revised results:

 

Try Super pi, it runs on less threads and should allow full Turbo Boost speeds to kick in. Also make sure you run it with the system at idle (no apps running). My 540M kicks into Max 3.06G after the second pass (32M) and then kicks in and out during the run, it's in full turbo maybe 20% of the time...

Looks like your SSD speeds improved also, not as dramtically as mine but the 4K speeds went in the right direction. How close are those numbers to what they spec your drive at?

 

Does anyone know of any way to verify that the intel IGP is truly disabled?

This computer has no switchable graphics only the 5830 GPU. I am on battery now and have HW32 open to monitor CPU power usage and I jsut noticed that their is an IGP temp reading. What baffles me is that the IGP is always a degree or two higher than the CPU cores. At first I thought this was just thermal heat transfer, but now that it's running warmer I am thinking it must be powered up and wasting energy. Any thoughts or ideas would be much appreciated....

PS; I have checked and cannot find any drivers other than the ATI for graphics and nothing in device manager under CPU that remotely looks like anything to do with graphics.

 

It might be a wrong hwinfo reading..
Please enable debug mode in hwinfo preferences, that will create a dbg file in the installation directory. Reproduce the problematic readings and then post that dbg file along with a screenshot to the hwinfo thread in my sig.
The author of hwinfo is a nbr user and he gives awesome support for his program.

 

This is for an Asus G73JH

I only disabled the Intelppm driver service and the 4k reads/writes on my 80 GB Intel SSD jumped from 13.34/20.43 to 20.48/44.80.

Disabling the Processor driver service is not needed.

RESULT:
Disabling Intelppm also disables Powerboost. Utilizing 1 thread will not make the thread run at above 2 GHz anymore. For the 720qm this means it will either run at 1,6 GHz or 1,73 GHz.
The CPU temperature is only 4 degrees higher at idle but this is mostly because the fan runs a little faster. It basicly will run at the same idle speed as the GPU fan.

 

Given the opportunity, I would like to show you something

SuperPI 4M run

without tweak--> with the registry tweak



wprime 32M

without tweak--> with the registry tweak


As you can see in every case, with or without the tweak, my CPU is beyond the 3.33Ghz Turbo Boost. Moreover, this enhanced CPU speed is active in more than one cores, contrary to what Intel specifies for Turbo Boost (which should run on only one core).
Why is this happening? Because of the BIOS. I currently have the A07 BIOS, but, if I install the A08, I will loose this enhanced performance.
Which means that the BIOS of the system plays a very important role in this situation and that's why the tweak I mentioned (to disable the CPU drivers) by itself is NOT the solution to the slow SSD performance. It definitely improves the situation but it's only a temporary patch. Plus, the CPU because of the slightly increased heat, will perform a bit slower than without the tweak, but this will vary in every single laptop.

 

I now tried the hidden 6 high performance powerplan options as I didn't like the loss of the turbo boost option. The ones noted on page 51.

For the G73JH I didn't lose the low idle temperatures and powerboost. 4k reads/writes went to 17.23/44.07. This is better but not as good as disabling Intelppm driver service.

 

look at this thread for an answer
http://forum.notebookreview.com/har...e-integrated-graphics-core-i3-i5-i7-cpus.html

 

Interesting thread, really nice info here. I just did the Stamatisx on my T9600 with vista64 and original intel firmware (not updated), never trimmed (like 9 months of operation).

When i did the change the T9600 stayed locked on 2600mhz locked no throttling nor going to 2800, the temp did increase couple of degress (im guessing cause the multiplier stays locked up). I revereted the change cause i wanted the cpu to be able to go up n down depending on the load.

 

I applied both JJB and Stat's tweaks. My results are looking good but idle temps went from 45-47C to 48-52C. About a 4C increase, not too bad.

 

I would be interested to see what your temps are if you just use my settings only. IIRC I also noticed a 3-4C rise applying stamatisx' tweak but gained no real speed improvement over my settings.

Also I posted earlier that you can adjust the two '95%' settings upward to gain adittional Seq and 4K performance. Give 99% and 100% (= higher temps) a try.

 

also, stamatisx mentioned it but I can't find where.. When you roll back from his tweak it's not enough to change the values from 4 to 3 and reboot. You should uninstall the CPU under device manager and then reboot again or the CPU won't turbo boost.

 

I've been doing some real world testing of the tweak (Stamatisx one). Sofar the effects on real world SSD performance turn out far smaller than I thought, even when doing some basic multi tasking.

In theory they could help when their is some heavy usage going on but it's seems impossible the CPU would still be in that power saving mode.

I will be doing more benchmarking this week, I will update with results.

 

They are using the same heatsink. Heck, they're using the same heat spreader. It's entirely possible that the GPU is simply heating up because the CPU is. Just because it can read the thermal diode on the GPU doesn't mean the GPU is enabled. The thermal sensors are addressed separately from the hardware itself.

 

Makes sense except for the fact that core 1 and core 2 are even closer together with the same heat sink and the readings between the to vary greatly +/- 10C yet the IGP sensor seems independant of those readings, sometimes higher and sometimes lower.

 

Ok here are my results using that register tweak!.

AWESOME!. This keeps my processor at idle and the SSD is performing like a champ!. Seriously guys someone should compile this thread and edit the first post! this solved ALL my issues with my SSD. Nice job guys. I have the same temps and get the same battery life this is more than perfect! .

 

Do this please.

 

I've been doing more real world testing. Good news: the tweak does make a difference. Not for heavy multi tasking, not for file copying (in my setup) but it matters for installing programs. It matters a lot.

Installing CS4 without tweak 9:30 min. With tweak: 4:30 min.
Installing Office 2007 without tweak: 8:32 min. With tweak 5:18 min.

The tweak I used is this one*:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Processor
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Intelppm

Change on both the "Start" from 3 to 4

* This tweak can raise temperatures and decrease battery life. Consider only using it for the high performance profile.

Jayayess, have you tried that tweak?

 

For anyone who has benefited from the great work of guys like stamatixs, JBB and others: if you want to give them reputation points use the button underneath their avatar that resembles a scale.

@ stamatixs, great looking article stamatisx. I'll read it later.

You've been served:
How To: Improve Low SSD Performance in Intel Series 5 Chipset Environments | StorageReview.com

 

nice link full of data
nice work!

 

Been waiting to see how others fare with these tweaks first, but I am about to try them.

 

The tweak works well on my 1830T. Haven't tested battery life though.

 

Now we need to find out how Intel or any other OEM out there will proceed with a solution. I find it hard to believe that all those engineers haven't noticed anything all this time.... I mean what are the odds?

 

I would think they know about it, at least now they do.

stamatisx, did you know that only changing the intelppm to 4 seems to do the trick as well. any need to also change the processor value?

 

yes I know, but since I was using the same tweak for my R1 (I needed to apply both in order to overclock the CPU and not to have BSOD) I kept the tweak as is till today.
If it works without any issues with only one registry edit then even better, just give it some time to see if there are any side effects otherwise we can confirm it.

 

So, these tweaks don't work in Linux?

 

Kinda hard for them to work there since Linux has no registry
I still haven't tested to see if Linux is affected the same way like Windows. I don't have time right now

 

I knew that...

 

For sure they won't work
I was thinking to use IOzone to examine the situation with Linux

 

After applying the tweak my Device manager looks like this. Is this normal?



_______________________________

Here's a Sony Z benchmark before and after tweak result.

 

Hey Phil!
Yes, this is normal, stamatisx mentioned it.
Some users were reporting that turbo boost is affected with the tweak.
Could you check it out?

 

thanks mfractal.

It seems my Core i3 330um has no Turbo Boost.
Intel® Core? i3-330UM Processor (3M cache, 1.20 GHz) with SPEC Code(s) SLBUG

 

Yeah Phil what you see is normal when you apply the tweak. Just a thought, how much slower of faster is your CPU with and without the tweak? For instance if you run wprime or SuperPI do the times remain the same?

 

Ok so what's the simplest tweak? I don't want turbo boost etc being turned off.

 

This is the simplest
http://forum.notebookreview.com/har...-full-advantage-fast-ssds-20.html#post6693099

About the Turbo Boost, that depends on your BIOS, in my laptop it remains on, to others it turns it off... you have to try and find out

 

Just for informative reasons, that's my Vaio Z bench ^_^. But i used the ''no idle'' tweak, which requires some register editing.

 

Wasn't it you who said only Intelppm needs to be set at 4?

I think that's the simplest tweak.

 

Next to these changes I also disabled core parking. Makes no change for idle temps or turbo boost but I now get 19.28/40.24 for the 4k read and writes.

 

There is no difference in Passmark CPU X64. (Intelppm + processor tweak)

 

Thanks Phil

 

I have an Intel 80 GB SSD Postville in a G73JH laptop.

Default it performs like in the first thumbnail screenshot.


Disabling the Intelppm device driver service it performs like in the second thumbnail screenshot. I don't like this setting because it disables turbo boost for the G73JH.

The third screenshot is what I am going with now. I changed the 7 settings like in JJB's posts AND I also disabled core parking( 2 more settings). This seemed to be important for my results.
The upshot of this is no higher idle temperatures or fan speed and turbo boost is working.


I still think the 55 series chipset is flawed on a hardware level.