M17x-R2 BIOS settings to partially restore SSD performance

For a month now, I am trying to figure out the reason for the bad SSD performance on my R2.

I was also trying yesterday morning the new VBIOS and beta drivers from Dell and I was still having issues with the sleep mode.
So after two formats and clean windows installations I managed to figure it out and led me to this thread:

Helpful links:

Partial Solution to the SSD performance
You need to have the A07 BIOS installed
then proceed
Change the BIOS settings to the ones you see on the screenshots below:

(I can confirm that with 3% Overclock the results remained the same, so you can safely change that)

You need to have the chipset drivers and the RST drivers from Dell's website installed.


Tips to further increase SSD's performance
http://forum.notebookreview.com/har...ce-tips-underperforming-ssds.html#post6632298
http://forum.notebookreview.com/alienware/474521-installing-ssd-tips-tricks-benchmarks.html

Benchmarks confirming the improved performance

Normal Mode / 1% CPU Load / Write Cache On
Prior to BIOS settings -> After BIOS settings


Normal Mode / 13% CPU Load (while running SuperPi) / Write Cache On

(As you can see with the CPU under load the performance gets even better)

How the 4K random reads and writes should be:

(the screenshot is from an Alienware M17x - R1 with NVidia Chipset and the Intel X25-E 64GB SSD)


Conclusion
It's been a month that I have posted this issue on both Dell's and Intel's official support forum but so far I haven't received any reply from either of them.
Brian - erawneila said that he was going to take a look at this issue but still no news despite the promises.

All the evidence so far indicate a problem to the power saving features and the BIOS that prevent the chipset to perform as it should and utilize the SSD at 100%
The following post confirms the fact that C states when disabled from inside the BIOS change the behavior and the performance of the SSD (the chipset is the HM55, very similar to the PM55 that the R2 has)
http://forum.notebookreview.com/6644770-post64.html

I also have to mention that the new VBIOS may have helped the situation but I cannot confirm it since I haven't run any benchmarks before and after in order to make comparisons (I just speculate here)
This thread will only provide you will a temporary solution to this problem since you don't get the proper performance out of the SSD. The real solution can be provided only by Dell (or even Intel as well)

 

it looks like you had all the settings I had mine set to when I posed my horrible results on mine, except maybe usb powershare, but that's just for the ESATA port being able to deliver power when the computer is off. Which is use to charge the mamba. Are you using the 8.6 RST Dell drivers?

 

I use the 9.6.0.1014 (you can see it on the 4th screenshot)

 

interesting, whenever I tired to use their drivers it gave me an error that said it wasn't supported or some nonsense. The ones from Intel worked just fine, the Dell one's didn't want to install for some reason.

 

thanks for that stamatisx!
can you explain why is A07 needed as those settings seem to be available in A05 as well.

 

There is one nasty post lock up upon waking up the system from sleep if performance options in Bios were changed, especially DDR3 Voltage 1.7 for example causes this with the new HD5870 VBios, someone else also said to have issues when ethernet was disabled, so there might be some few options, if not set to default causing issues. Don't know how releated that is to a SSD setup and it's performance. But it concerns your issues about power management i guess. It was completly driver or vbios unreleated. It is also known from someone mentioned that to me, that the A07 Bios is the culprit and doesn't like all the changes made to it.

 

The reason is that we will have the same base and this way we can eliminate problems due to different BIOS versions (the same applies for the chipset and storage drivers)

 

The reason I mentioned VBIOS is that it solved the sleep/wake up issue. Considering the fact that BIOS affects everything including the GPUs and the chipset I cannot exclude an indirect dependency between all those variables.

 

I can't say anything at all comparing it to the SSD perfomance. What i can say tho is that there can be further sleep issues upon changing Bios values from default factory settings, to something else. So Disabling Bluetooth, Firewire, and USB Emulation or anything from default to custom might cause instability aswell, it has influnece on power management, so it might have perfomance influence with SSD aswell. What is the state on it when BIOS is set to default, and you just change your HDD settings, so you can boot up? How is the performance then?

 

Well, I leave that to the rest of you to try and find out since the settings provided on the OP work just fine for me without issues

 

Oh i see, goodie, but to make it proper, maybe we just have to wait for a newer Bios A08 or even A09, that will fix it.

 

i'll try your BIOS settings on A05 and if that doesn't help i'll upgrade to A07. RST version you menitoned is already installed.

BRB with results.

 

Well, all settings as stamatisx suggested (RST + Bios changes) but with A05 BIOS don't improve anything..


Installing A07...

EDIT:
So, i was really hoping it'll work but even with all mods applied running the correct RST version i still get low results while idle on A07:

 

The only explanation I can give is the new VBIOS (I also had a clean windows installation but I don't think this would change anything in your case mfractal)

We need someone with a fast SSD that has applied the new VBIOS in order to confirm the improvement

 

So us 4870 folks are going to be left in the cold if the VBIOS is the reason as it's only for the 5870

 

Unfortunately it didn't work out for me either. However, this has gotten me to think that perhaps a program that continually polls the CPU to be at least 2-3% active on all cores should restore SSD speeds fully on PM55/HM55 chipsets. I think Unclewebb could probably pull something like that off with a special TS feature, I'll have to ask him.

 

ok, thanks guys for trying it out. It's a mystery then how my speeds were improved (not fully but significantly) without applying an extra load.
They still are far from what they should but it's better than nothing.
I still hope this will work out for someone...

 

There's no easy way to do this. CPUs usually run at full load or no load. You can continuously load a CPU 100% for 3 milliseconds and then go back to zero load for 97 milliseconds and create an overall average load of 3% but you're not really creating a steady 3% load by doing this.

The RealTemp folder should have a program in there called LoadTester.

RealTempBeta.zip - [email protected]

It is designed to fully load a single thread of a CPU and is then adjustable from 0% to 100% in steps of 10%. You might want to try some benches while playing with this. You can also use Task Manager to lock the LoadTester load to a single thread to see if that makes any difference.

I've sent a PM message to saturnotaku who might be able to show me what bits in what register in his CPU is being changed when he toggles the C State option in his bios. A simple fix like this would be great. This is not publicly documented by Intel so it will take some trial and error and some luck to find a solution. Does the present C State option in ThrottleStop make any difference to SSD benchmark performance?

 

Unclewebb, if throttlestop can manipulate the C states in a future version, that would be great. (I have tried the current one but it didn't help)

 

Do we have the option to disable the C states on our Alienware? Can't find the option from my BIOS

 

Unfortunatelly not but if Dell could give us the option that would be great

 

Whats the deal with the Samsung SSD? This is the second drive since may they degrade insanely fast. I never had these problems with my SXPS 16 on the PM45 chipset do you think this could be related to the poor performance on the PM55 chipset?
this is my CDM score



Uploaded with ImageShack.us

 

HM55/PM55 chipset = cap problem.

Is it TRIM enabled? Can you check your alignment? Whats the firmware version?

PS: You need to create new thread for further assistance so we wont hijack this thread

 

I created a new thread as suggested
but here is what you wanted


Uploaded with ImageShack.us