M17xR2 Bios A08 Discussion Thread

Nah it's ok, the 3Dmark06 scores of the Shader Models 2 and 3 speak for themself what influence the CPU has in those GPU tests and you can compare it pretty well to a game that is using a healthy amount of cpu power aswell.

 

Keep the information coming - this is a great. I should have something from engineering by the first part of next week on this...

 

Hope its good news for all the exteme cpu owners....your doing great work Brian.....thanks (even tho i dont have an xm!)

 

Thanks Brian... I am still working on rendering as I type on the M11X. With 3DS max 2011 being a new install I had a time trying to find a taxing scene to render. The best, most intense thus far has been a 13min rendered scene. But I went with it instead of searching all day for something else. When I am done with A08, I'll flash to A07 and double check my numbers. Then I'll post the screen shots. Some questions:
Is there a need to flash to A06 and test?
Should I do a similar test with wprime just so that we have more data or is DR's test and images enough?
Finally if anyone has a link to a very intensive rendering scene please PM me. I have no intentions of using it for anything but this test. Still I know the copyright issues, etc. are touchy. Not suggesting a commercial one, but someone may know a link to a free scene, resource cd scene, etc.

 

We appreciate the help but how about making a bios so that we do not have to depend on Dells video drivers? I'm getting irritated not being able to play BC2 because of either the crash or the fricken lag

 

As promised:
-best I could give is screen shots, it's late and I'm tired.

920XM OEM

3DS Max 2011 Rendered Scene:
BIOS A07 = 13:42


BIOS A08 = 14:44


Wprime:
BIOS A07 = Thread 3 93% 461.508

(sorry must have bumped a tab when I moved CPU -Z showing the BIOS vers.)
BIOS A08 = Thread 3 93% 547.153


3D Mark 06:
BIOS A07 = 18699 cpu = 4449


BIOS A08 = 16324 cpu = 3946


I never could find a really intense scene to render. However I did render the scene at least 4 times in each bios to get a rough idea of the average. All and all the A08 felt slower for everything while working in 3DS Max. If I forgot something let me know I may give it another go.

 

I think the nail in the coffin for A08 will be a CPU intensive game test. Maybe GTA IV. I also want a statement from engineering describing where exactly they saw throttling occur.

 

+1 I want to see that too, really curious about those mysterious ways of testing

 

It looks like engineering saw throttling problems in the M15x and decided to castrate the 920XM and 940XM extreme CPUs in both the M15x and M17x. If the M17x was really throttling then why did performance decrease across the board with bios A08? In the results posted, there seems to be no justification for slowing down the M17x.

I recommended wPrime because that benchmark is 100% CPU dependent and is a quick and easy way to show when CPU performance has taken a hit. In DR650SE's test, RealTemp is showing a multiplier of 20.3 for bios A07 compared to 17.0 for A08.

This ratio ( 20.3 / 17.0 ) accurately predicts the change in pure CPU performance ( 434.692 seconds / 362.256 seconds ).

With bios A07, the reason that the 920XM is being limited to an average multiplier of 20.3 is because it has reached the Intel TDP/TDC limit for this CPU.

This is easy to prove. If you manually increase TDP/TDC while 4 threads of wPrime are running, when using bios A07, the 920XM will continue to run faster until it is using the full 24.0 multiplier. If throttling was being caused because you were reaching the limits of the power adapter, increasing the TDP/TDC limits would not increase performance. The CPU would continue to throttle the same amount.

The adjusted turbo multipliers in bios A08 was a fix for a problem that doesn't exist in the M17x. The best solution would be adding a bios option so a user can choose to run the same turbo multiplier whether 1, 2, 3 or 4 cores are in the active state just like all of the previous bios versions used to do by default. That has always been the Alienware advantage and should continue going forward with bios A09.

 

The real reason that this was applied to the xm's in both m15x and m17x is simple.....laziness and incompetence on behalf of Dell engineering. It stands to sense that they thought they could fix or remedy thhrottling issues in the m15x and at the time, some bright spark must have thought "Hey, Ive got a good idea, why dont we just splosh this fix around to EVERYTHING we can....call it a preventative measure and it will make our jobs ALOT easier because we will have fixed problems that havent even happended yet!!"

I can just see them patting each other on the back and saying "Well done guys, weve saved some more money on this one and no one will notice the real world drop in performance!"

OR WILL THEY!! - they forgot something quite relevant which is that Alienware owners arent your typical "run of the mill" computer owners....of course there are gonna be people out there who notice.....which brings us to where we are now......

BTW, Excellent work from Infernia and DR with those tests.....

 

keep in mind, you only did one scene.

Real world use of 3D Studio often is used to make videos, i.e., multiple scenes. Using the results you posted, a, say, 30 minute video, would take considerably longer to render under A08. This is not a minor issue; a company unaware of the changes could miss important deadlines, possibly losing a client.

 

How can i revert back to A07? After I flashed to A08 i get BSOD when loggin in to windows. So I cant get to the flasher. I can get on in safe mode but the flasher does not work in safe mode. So how do i revert back? This is extremely irritating..

 

Make a bootable CD/DVD with the A07 BIOS and flash from it

 

Wow, Bro, awesome tests, and I think your results speak for themselves, for sure. I really do hope this gets someone's attention in the Engineering group. I know Brian is all over it, like a cheap suit! Glad I have stuck with A07. I was one asking for the touchpad to be fixed, but not at the cost of 27% of my CPU performance.

 

Guys,

Flashed back to A07 and ran my weekly virus scan with NOD32. As usual, the process took 2hr 52min... I have a lot of crap!



Reflashed to A08 and re-ran the scan and went to bed.



Under A08, the EXACT same files took 3hr 39min! This is very depressing.

Virus scans, video encoding, 3D rendering, etc. are normal day-to-day user activities that the M17xR2 should (and did) handle flawlessly. Even my favorite game, Supreme Commander, is sluggish and it's a 'Plays best on Alienware' title!

A07 --- 2HR 52M = 172M
A08 --- 3HR 39M = 219M

Assuming my math is correct, this amounts to a 27.3% slowdown.

My VMware Workstation performance is also now sluggish but don't know of anyway to benchmark it since none of the normal tools will work correctly with a virtual machine.

 

And how the heck would I do that? I think i know how to make a bootable CD from googling and searching, but how would I flash the bios????

 

Now I can understand measuring the real performance loss of a batch job or any task taking time to process, but this "sluggishness" I dont really get. Forgive my nativity, but shouldn't sluggishness happen more with going from a SSD to a mechanical Hard drive, or less RAM, and not the throttling of extreme processors? For day to day tasks of starting up a computer, running word or web, or simple system administration CPUs have been fast enough to where I don't see a faster processor helping those tasks feel less sluggish. Maybe someone can explain that too me. I dont doubt what you guys are experiencing, but out of curiosity's sake maybe someone can clarify where (for example VMware) you are feeling sluggishness and why the CPU would be the culprit. Thanks

 

Some really good work from Infernia and DenverESullivan, and my friend DR as well. The one common denominator I have seen thus far, from every single one of these tests, by different people, is a loss in performance in every given example, from benchmarks to real world application. I have not seen one single instance where any sort of performance gains were exhibited. Rather disappointing to say the least.

 

Your problem isn't the bios, it's your drivers. While you're in safe mode, go to the device manager and delete both video cards. Reboot and reinstall your drivers.

 

Brian - erawneila,
would you mind to take a look at this:
How To: Improve Low SSD Performance in Intel Series 5 Chipset Environments | StorageReview.com

and now could you please take a look at these as well?
(please check the dates, beginning 08-04-2010, 04:05 PM)

http://forum.notebookreview.com/6556730-post118.html

http://forum.notebookreview.com/6577711-post25.html

http://forum.notebookreview.com/6583585-post49.html

http://forum.notebookreview.com/6583611-post6135.html

http://forum.notebookreview.com/6586935-post454.html

http://forum.notebookreview.com/6608952-post471.html

Now, I would also like you to see this:

Response of Intel on an email of mine concerning this matter:

I have lost count of how many times me and other members of this forum have posted this link:

Alienware M17x - R2 Killing the SSD performance - Laptop General Hardware Forum - Laptop - Dell Community

As you can see it is 44 days old and counting...
Is there any specific reason that Dell has done nothing so far?

Brian no offense but considering how many times you said you would take a look into this or contact the engineers, etc... repeating it once again this time will mean nothing (at least to me), I would really like to see a reply of yours or from Dell with a real answer and not another fake promise just to gain time or bury this post under other BIOS related posts.

 

i totally agree with stamatisx regarding the fact that this is a very acute issue and it should get the attention of the engineering team.
Also, Brian, i'd like to take this opportunity to stress the fact that we do appreciate what you are doing in here, and those complains are not directed at you but at the engineering team.
We know that you do a lot to help us out here.

We are all, and when i say we i mean all SSD owners with M17x-R2 are eagerly waiting on a response from engineering team.

 

You should put all that information into a new post stamatisx, and then it should be stickied! That is really good work at collecting the problem.

 

i also agree plus the point we dont all have intel ssds and this should not be limited on this. and the same occurs on all of us.

 

I finally had a bit of free time today and investigated why Dell engineering would have cut back the multipliers on the extreme CPU's like they did and I believe I know why they did: The M17x-R2 has an inadequate power supply! The 240W is simply not enough to furnish power to both 5870m GPUs, RGB LED, 2 HDDs, external USB ports, BT, Wifi + an extreme processor.

How to test for this and probably what the Dell engineers did after seeing my M15x throttling thread (I feel partially responsible for this change, sorry guys).

Test 1: Insufficient power confirmation

1. Turn on furmark and run the stability test. Be sure to tick displacement mapping, extreme burning mode, post fx.

2. Turn on GPU-Z to monitor GPU load %.

3. Turn on ThrottleStop and if you're on Bios A08 + 920/940xm, revert your multipliers back to 24x-25x across all 4 cores.

4. Activate WPrime 1.55 and start the 1024MB test with 8 threads active WHILE furmark is also running.

What you will see:

-GPU-Z will show that the GPU load% will constantly fluctuate between 0% and around 70-75% load. The fact that it falls to 0% means the system is throttling the GPU's due to lack of sufficient power.

Test 2: Why Dell made these changes

1. Turn off WPrime 1.55 and furmark.

2. Using throttlestop, set your multipliers back to what A08 had them (e.g. for 940xm it's 25/24/18/18).

3. Re-do the previous test steps with furmark/wprime/gpu-z.

What you will see in Test 2:

-GPU-Z will show GPU load% staying steady between 70-85% the entire time and it should never drop to 0% like before. This is STILL 20-30% short of full GPU performance!

-You can drop the CPU TDP/TDC + multipliers way down and it still does not compensate enough for the GPU load to ever hit 100%. W

-Only when WPrime is shut down does GPU load% jump back to 100%.


What does all this mean?

-Dell built the M17x-R2 + M15x with inadequate power supplies from day 1.
-The M17x-R2 with the 5870m Crossfire + RGB LED and other hardware could never sustain 100% GPU load in extreme conditions where a game may utilize 100% of the processor.

How realistic is this test?

-Not very realistic. The games that utilize 100% of the GPU like furmark does in conjunction with 100% CPU load are few and far between. The most extreme example I can think of off the top of my head is Bad Company 2 and even that doesn't stress the GPU 100% but it does push all 4 cores to their max capacity routinely.

What do we need?

-Simple, a bigger power supply and bios that will utilize it. Will Dell realistically ever offer something like this? I wouldn't count on it. Their solution was to pull the extreme multipliers down to default so that the GPU's would not throttle to 0% when the CPU is fully loaded.

-Dell can make a change in bios A09 to have a switch that toggles extreme mode and sets the multipliers back to their pre-A08 settings. It should be accompanied with a warning that system performance can be degraded by toggling this switch. It's very rare for a game or application to ever push the CPU/GPU in tandem to 100%. Dell was being cautious after the M15x throttling and numerous other throttling systems they had. Ultimately they should have designed these systems with better cooling and much larger power supplies but hindsight is 20/20.

 

Well we've known for quite awhile that the 240 power supply was weak, and is basically the only thing keeping these from performing full-bore like they could. Good way for them to cover their mistakes but was never asked for or complained about to cause this tweak.

The BIOS option would be nice, but what would be even nicer is if Dell would create a beefier power supply and send it out to M17x-R2 owners for free if they call in about it. But good work on the tests there Joker, hopefully one way or the other something positive will be changed ASAP.

 

put 'kill A watt' data also..this prove that m17x require more than 240watt current..

 

no laptop ever ship with a decent psu the m17 180 wats psu well let's just say it resulted in the quad problem

 

Joker do you think that the 240 watt PSU is enough for a i7 720 and crossfireX 5870's?

Or is this PSU really not good enough for any of the m17x's?

 

Hmm... Could the 300W PSU from Clevo x7200 work perhaps? Maybe we could ask Justin or BabyHemi to send us one for tests.
Regardless, Dell could try testing a stronger PSU and then issue a free replacement for those with XM CPU's and CF.
Just a thought...

 

Would just a PSU suffice? I think the BIOS should support that or am i wrong ?

 

The BIOS detects if you are using the correct PSU or not, then whines if you aren't. It can even glitch from time to time and say you're using a wrong one but you're using the one that came with your system. They would need a BIOS to go with a beefier PSU.

 

Nope won't work because the Clevo 300W PSU uses a different connector and it is 20v. Dell would have to make a 300W adapter.

 

i have change the cpu from 620 to 920 (on 620 not problem)
this explains why I could not understand why if you increase the 5% of the cpu speed on 920 every now and then the notebook is not turned on .. now with the A08 bios always seems to go well...
this is a problem of power supply ..or internal notebook power switching?

 

I'm at least hoping for a BIOS option to re-enable the turbo multipliers. I liked the idea of Intel default multipliers when on battery and performance multipliers when plugged in. I can't imagine Dell ever offering now power adapters

For now I will stay with the A07 Bios, I don't use the touchpad, but what gets me is that in the future, let's say a new Bios comes out with support for new GPU's and I upgrade, I must now use TS to achieve the same level of performance as before. A07 will only last so long.

Thanks to Brian, Joker, and Uncle Webb (and others) for all the work you've done on this issue, it is definitely appreciated

 

Thanks Joker. At least we know what 'potential' problem they were trying to avoid.
I think that I will stick with A08 unless you guys are sure that using TS is a one time change and not a constant struggle with the BIOS. First I read that it was a constant battle between the two then later I read it wasn't.
Anyway, sticking with A07 will only be viable until they come up with a fix for the SSD's slow 4K read/write speeds.
I long for the day that laptops will be less proprietary.
Isn't this how the original Alienware company came to be? A bunch of enthusiats sick of the big corps controlling laptop configs and limiting the capabilities? I understand it's a balance of heat/power/performance but knowing a problem exists and being able to implement a somewhat simple and cost effective fix should be their priority. If it comes down to the fact that a new 300watt PSU is needed, it's a matter of manufacture and BIOS revision. Even making it an upgradable option for current owners would be nice.
We can only hope they give us (extreme cpu owners) the option to keep the multipliers as is or change them, in the next BIOS revision.
Cheers all.

 

It would be nice if Dell engineering could be honest for a change regarding what the issues are.

 

I thought I had read someone mention power issues over a week ago!

 

Do you have the 150W power adaptor? I never have issues. If your system didn't ship with crossfire gpus, or an extreem CPU, then you didn't get the 240w adaptor and are using a very inadequate PSU.

 

Yep - this is definitely true. If you upgraded to dual video cards or an extreme CPU, you need to purchase a 240 watt AC adapter.

 

my system start whit cpu 620 + 2 x 5870 and psu 240watt..
but after i change the Cpu to 920 xm.
then whit bios to A07 on normally and most in the 5% of Oc ,i have random not start ...

 

Are you on battery or AC? If on battery, make sure your battery charge is >70% with A07 BIOS or >50% with A08. Where did you get the processor and what kind of thermal paste did you use?

 

I just spoke to engineering, and while they didn't commit to anything yet, I should have some news for you guys later this week. I'll keep you posted.

In the meantime, thanks to everyone who provided the awesome benchmark and real world test case results. I can't even begin to tell you how helpful these were. At this time, I don't need any continued testing or results - we have what we need.

Thanks again!

 

Brian, just want to thank you for your efforts. If you need any other benching, don't hesitate to ask. I've got XM and non XM CPUs that can be tested as well as all the popular benching progams, and numerous bios versions. And if there is any testing needed on betat products (unlocked voltage! lol ) well we'll test that too.

 

thanks erawneila ...
the problem is on ac power, i'm not use battery only , the thermel paste is prolimatech PK1 , the temp of 920 is normally 39-t0 65 on max load
when not start , the led power is on and fan go to max rpm..(but on this time ,
after many reboot not have the problem..)

 

might want to re-check your CPU install since you upgraded your CPU.

 

Yea, are you sure you locked the CPU in place?

 

Brian, like always we thank you for your work. One question remains though. What is the status of the SSD performance issue that many have complained about and that I tested myself.

My M17X R1 has better 4K speeds on my Vertex 2 SSD compared to my R2? Why?


Concerning the PSU issue, I thought we had discussed this back in June and we had all agreed that the PSU is undersized especially with the power loss as it heats up. Dell needs to come up with a 300W PSU for all Crossfire, i7 CPU setups...

 

I proved a long time ago that the 240W power supply was grossly insufficient for the hardware in the R1. Tons of Kill-A-Watt tests showing exact throttling. With A07 and earlier I was able to get a few games (Crysis, Dragon Age) to throttle on the R2 if I really tried.

It's a shame. We're losing so much potential on these beasts due to the power supply.

 

Lets have a new power supply issued via Dell then eh???.....what do you think Brian???

 

Try to see if you can find answers on this link concerning the SSD, until Brian responds to our question.

http://forum.notebookreview.com/har...et-can-not-take-full-advantage-fast-ssds.html