M17xR2 Bios A08 Discussion Thread

Does anyone have a copy of A07 or A06? Where can I grab these. If I can grab them, I will post results with a 920XM, and 940XM ES. Tommorrow I have some benching planned with a 720M, If you can point me to the FTP site, and I can download an old bios, I will try and throw this in the mix. But it will also be helpful if others post as well I will post Vantage, and 3DMark06 benchmarks, as well as anything else requested. Most likely some CPU benchmarks like WPrime, Super Pi, and PiFast. I will be installing my 40GB Benchmarking SSD as well, as it has only benchmarking programs installed on it.

 

Older BIOS revs can be found @ support.dell.

see this link for the list (A07, A06, A05, A02) -

Dell - Support

 

A07 = R276847
A06 = R273034

 

Thanks Brian!

 

Make sure to do more than just synthetic tests if you're going to do a gamut of benchmarks. I know from experience that Dell engineering doesn't look favorably on synthetic tests, they are more interested in real world gaming results. CPU heavy games would be ideal, like SC2, GTA IV, Bad Company 2 etc. Also, applications that utilize CPU power that demonstrate a loss in rendering time would also be applicable here. I'd help out with the testing but I'm quite busy for the next few days.

 

^^ Lol, Murphy's Law. I have plenty of time but don't have the xtreme processor. If Dell could ship one to me - I'd do every possible bench

 

well really need responsible guys at Dell like you.

i uploaded all M17X R2 bios versions here

http://rapidshare.com/files/418183157/All_M17X_R2_BIOS.zip

 

No sympathy for the man with an 8740W AND a M17xr2.

Send me your 8740W and I'll send you my 940XM.

 

all..
it's compatible with R1?or this only compatible with R2..

 

All R2 chiefs, nothing to do with the R1. So your all good.
Would be nice if they could fix the R1 touch pad toos.

Also just noticed the base price for the R2 just went up $200 on the AUS site.

Cheers.

 

I need suggestions.
If anyone else on here is using Autodesk software suite (or any rendering software for that matter) how do we put the performance hit into numbers for the engineers? I judge by amount of time the same scene takes to render in either BIOS but that may not be good enough for engineers. Not sure how to represent that so as to give them solid proof of a decrease in performance close to 30%.

Feel free to PM me and let me know what software/monitoring or otherwise I should run in tandem to give the engineers some numbers.
Most likely I will be running A07 first since I have a ton of work to do this weekend. But once I get through that I can flash to A08 and give it another go.
Software designed to bench may not be their 'cup of tea' but it makes life easy when reporting numbers. Oh well all suggestions are welcome.

 

Here ya go infernia, try this pdf from Intel using 3ds Max 09. Number 10 is what you are shooting for and the rest are steps using an example scene to perform the test with. Hope this works for ya, I just don't have the time or I'd do my own tests. But so long as someone else is doing some Autodesk comparisons.

 

Guys,

As part of my career I frequently do Audio/Video work at home for my employer using Adobe Premier and After Effects. I just completed an HD project for them. To render and transcode the media, it took exactly 1 hr 23 minutes under A07.

I know that exactly because I almost missed my FedEx pickup the first time!

Trying to think of a real, real-world test I came up with the idea of flashing to A08 and then having Premier re-code the files.... exactly the same job.

EXTREMELY disappointing results.... My project now takes 1 hr 46 minutes.

This appears to be very consistent with the predictions from UncleWebb and Joker... It's almost a 28% reduction in performance!

** EDIT **

This particular project was composed of ten individual segments ranging in size from 300MB up to 13GB each for the main pieces with After Effects overlaying throughout for chromakey, titling, etc. Entire finish product was transcoded to h.264 for DVD authoring.

 

Thank you so much. I only need to make sure it will work with 2011. (just upgraded from 9 since I got the student discount. Hence I am re-doing a lot of work this weekend making sure all is compatible.) Anyway I appreciate the doc. I just read the first few paragraphs and it would seem this is exactly the kind of test engineers can't argue with.

**EDIT**
Denveresullivan - is there any kind of window in the software that tells you exactly how long the render has taken, if so maybe a screenshot of said window along with the version of BIOS. Sort of like they do in the Benchmarking threads with their screenshots. I am not clear on what these engineers want as proof but a screenshot surely can't hurt.

 

Infernia,

Not really, you just get a progress bar and a loud 'ding' when it's done in CS5. I always jot down the start time/end time of the jobs since my employer agreed to pay me for my time as well as the time my machine is tied up.

I usually start it, jot down the time and then watch TV or work on other paperwork from the office while I wait for them to finish.

Work in Premier is about as real world as I can get.

I'm going to see what this does to my NOD32 scan times tonight. As it is, that takes almost 3 hours due to all my crap. I'm almost afraid to look.

 

Interesting results DenverESullivan and that's about as real world as you can get.

Bios A08 : 1:46 ~ 106 minutes
Bios A07 : 1:23 ~ 83 minutes

106/83 = 1.277

which means with bios A07 your computer used to be 27.7% faster than what it is now with bios A08. Here's a quote from Dell engineering.

I guess as long as you don't want to use your computer for anything productive, you should be OK.

 

I've also noticed StarCraft 2 and Supreme Commander don't perform as smoothly as they did either. With Supreme Commander being the worst.

What's really sad is that it's a 'Plays best on Alienware' title

 

Dell can't possibly justify 23 min, and 27.7% reduction in speed as "negligible" And peoples work depends on real time rendering. I really hope dell takes some serious consideration of thier targeted audience, when they review this issue.

EDIT: I should have a Slew of benchs posted clearly showing the CPU Activity in an hour or so.

 

To be fair , they may genuinely of thought this would not have had a noticeable impact ... however this thread is clearly proving this is not the case....
I don't think they'd have a hope in hell justifying their statement now .

Hopefully Brian will be back to this thread soon to update us , I am sure everyone want's to know when this "mistake" will be corrected...

 

in fact i dont think they even spent a minute benchmarking. what they say is based on theory (i wonder if this can be even based on theory), or they just saying this just to get rid of us (normal tactic of Dell)

 

@Denver -

You might want to quantify the case you used - size of files, etc. Might as well provide as much info as possible to back up the numbers

Nice job on the data.

 

Batboy,

Good suggestion... I've gone back and tacked that data onto my original post.

 

Ok, Here is what I have been Able to drum up. I have done 3DMark06 And Vantage runs using the 920XM OEM and 940XM ES. I have done each run using Bios A06, A07, and A08. I have monitored CPU core activity using Argus Monitor, and it clearly shows the CPU is limiting the activity an several cores. In turn this is effecting scores. Though this is synthetic benchmarks, it shows that the results will translate to real world CPU usage. CPU Intensive applications are going to have reduced performance. This also goes for games with various degrees of CPU dependency.

I have also included a run of wPrime 1024 showing CPU activity for a 4 thread load. Obviously the CPU is capable of running 8 threads, but often times we are not going to use 8 threads, but 4 threads is more likely in everyday use. Given a CPU capable of running 8 threads, 4 should be fairly easy. And in this test we see that the A08 Bios clearly reduces performance at this level of activity. In this test, to get the CPU thread 96% complete, we see that bios A07 took 362.256 sec. Using bios A08 the same thread reached 96% completion at 434.692 sec. This is a differance of 72.436 sec.

920XM OEM 3DMark06
Bios A06

Score: 18992
SM 2.0: 7591
SM 3.0: 9476
CPU: 4684


Bios A07

Score: 18969
SM 2.0: 7590
SM 3.0: 9485
CPU: 4654


Bios A08

Score: 15464
SM 2.0: 5845
SM 3.0: 7869
CPU: 3978


920XM OEM Vantage

Bios A06

Score: P14982
GPU: 14650
CPU: 16075


Bios A07

Score: P15036
GPU: 14762
CPU: 15923

Bios A08

Score: P14495
GPU: 14650
CPU: 14046


--------------------------------------------
940XM ES 3DMark06

Bios A06

Score: 19501
SM 2.0: 7930
SM 3.0: 9754
CPU: 4678


Bios A07

Score: 19449
SM 2.0: 7870
SM 3.0: 9723
CPU: 4700


Bios A08

Score: 17748
SM 2.0: 6878
SM 3.0: 9247
CPU: 4234


940XM ES Vantage

Bios A06

Score: P14981
GPU: 14694
CPU: 15914


Bios A07

Score: P14966
GPU: 14665
CPU: 15947


Bios A08

Score: P14627
GPU: 14571
CPU: 14796

--------------------------------------------
920XM OEM, wPrime 1024, 4 Threads

Bios A07


Thread 4 is 96% complete @ 362.256 Seconds

Bios A08

Thread 4 is 96% complete @ 434.692 Seconds


These are the results of the testing I was able to do today. Took a little while . If you closely look at the CPU Activity on the Argus monitor CPU Core Graph, you can see the system is holding back several cores. On Bios A06, and A07 it is not. During the wPrime runs, the CPU avtivity was up and down, where as on Bios A08, CPU activity remained at a constant frequency, but a much lower frequency accounting for a differance of 72.436 sec.

 

Nice job DR!

 

Thanks, I think the most useful part of the screen shots is the Argus monitor, as it clearly shows the CPU Activity. The A07 Bios from the Dell Support site is reported as X30 in CPU-Z, which I found odd, but I'm sure Dell is aware of that.

 

Wow, awesome job DR650SE. I don't even want to imagine people that are doing rendering now on A08 over night and are wondering why it is not finished as usual in the next morning.

+rep on that one.

 

This is great information - thanks for spending all the time to do this. It looks like in 3D Mark (which from what I understand is a better representation than the other measures previously outlined in this thread) that the real world performance drop is closer to 10% than 30%. This is great information and I will take this back to engineering for review. Some of you are reporting > 10% drops in specific applications, so please continue to provide me with that data...

Thanks again!

 

It would be great if somone with an OEM 940XM can run the same test. My 940XM ES doesn't seem to perform as strong as the 920XM OEM in various cases. In 3DMark06, My 940XM ES chip showed an approximate 10% drop in performance (A08), where as my 920XM OEM chip showed an approximate drop in performance (A08) of almost 20%.

 

I think the 30% were the difference of the multipliers only 25->18 940XM and 24->17 920XM.

That's an easy math

But the percentage for wprime is around 17% performance loss / longer from my over the thumb calculation.

 

^^ yes, and also,

@Erawneila,
3DMark06 is comprised of 2 graphics and 1 CPU tests. On the CPU part there is a ~15% performance drop with the A08 BIOS.
Let's focus on the CPU performance please.

 

I got 18,888 with A07 in 3DMark06 and 17,131 with A08. Don't have all the screenshots and all, sorry.

With ThrottleStop setting multiipliers back to 24 and A08, I get 18,672.

On a positive note, the touch pad *is* night and day better. I'll keep A08 and use ThrottleStop (thanks unclewebb) until Dell issues a permanent fix for this.

N

 

3dmark06 has two cpu tests if i am not wrong.

 

Yea, 3DMark06 does have two CPU tests, as does Vantage. In my screen shots I labeled all Argus monitor charts as to what test was going on at any given moment. You can see the CPU activity when loading, under stress, ect. It charts each thread. Maybe the most useful part of my posts if anyone is inclined to analyze them. I was looking at them and it's clear whats going on. The CPU isn't running nearly as fast, and it's holding back several cores in differant situations.

 

I meant the scores, 2 parts for graphics and 1 for the CPU.
That's why looking at the total score isn't too accurate when talking about CPU performance drops.

 

I respectfully disagree. The drop in CPU speed causes all three scores to drop, not *just* the CPU, so the total score is useful.

N

 

Look at the charts in my benchmarks, you'll see why the score in all three catagories is dropping.

 

I agree on the 3 single bench scores, but not that the overall score is "that" useful, as i don't know how it is calculated correctly (maybe its just a simple average).

It's more clear to me, if you just look at the 3 scores SM3, SM2 and CPU individually and compare them from A06->A07->A08, as you have a better idea on what is the unit for it. For the GPU tests it should be fps and for the CPU tests i think it's operations.

And since 3DMark06 is a "bit" more CPU intensive as Vantage, you also notice a "minor" decrease in the GPU tests there.

 

Of course they are all dropping!
Do you wanna look at the total score? Do you know the formula for SM 2.0/3.0 score calculations? Can you tell the exact CPU perf. decrease %?
Do you want Dell engineers to have the irrelevant data to base their decision on?
We are talking about CPU crippling here.
So the tests should be analyzed accordingly.
wPrime is a good test to rely on.
3DMark06 ( CPU only)
Video rendering/encoding would be a good test.

 

Here is a score calculator with the formulas for the Futurmark Benchmarking Programs.

The monitors in the charts I posted show exactly which cores are doing what, during which tests if anyone cares to actually look at them

And heres this

SM2.0 Score = 120 x 0.5 x (SM2 GT1 fps + SM2 GT2 fps)
HDR/SM3.0 Score = 100 x 0.5 x (SM3 GT1 fps + SM3 GT2 fps)
CPU Score = 2500 x sqrt (CPU1 fps x CPU2 fps)

 

<<<<<< Edit:

So what part/% of the SM2/SM3 score plays the CPU?

 

I know, it's just the a score calculator with the scoring formulas built into it.

I shold have choosen my words more wisely. the charts I was talking about were the Argus monitor charts in this post. Scores with CPU activity

 

Yes,...hopefully, they will look at the Argus monitor charts instead of saying "Mere 8% loss in 3Dmark06 isn't too bad. Case closed" That's what I was trying to avoid here.


Also, to understand the SM formula's we need to know how the GT1,2... are calculated. Those are mostly based on the GPU performance, AFAIK.

 

The influence of limiting the multipliers can be visibly seen on the gpu scores of 3dmark06 aswell, it's already integrated and it does effect rendering of images with Shader Model 2 and 3. So this is "real" comparison to a game that is cpu dependant aswell and not only doing every calculation on the gpus.

 

Here is what I found on the GS Scores.
GS for hardware capable of running all tests = 0.5 x (SM2S + HDRSM3S)
GS for hardware capable of running only SM2.0 graphic tests = 0.75 x SM2S

3dMark Score = 2.5 x 1.0/((1.7/GS + 0.3/CPU Score)/2)
i.e 2.5 weighted harmonic mean of GS and CPU Score

I'm hoping they take the time to actually look at them as well and not simply glance over the numbers.

 

Don't worry guys - the results DR posted are clear based on the CPU scores alone - which is what they should be looking at. The info he posted on WPrime alone speaks volumes.

Just my 2 cents on it.

 

I want to give to Unclewebb Credit for the wPrime runs. He PMed me with the idea as I hadn't thought of it. But it does show pretty clearly that the CPU is going to be affected across the bored.

 

Just looked at the numbers again, dunno where did Brian get 10% from. Even looking at the total 3DMark06 score before and after, - the difference is ~18.5% .

 

Just a thought, if you guys want to post % calcs - suggest you show your math Leave no room for argument. At the very least, if we are incorrect, it can be explained in detail.

I really like the real-world results being posted though. Having these results helps greatly to back up the performance loss. Keep it up.

 

There's no math (and I still managed to screw up, lol).
So, from the tests:
920XM OEM:
3DMark06:
A07: 18969
A08: 15464
Performance loss: 18.5%

940XM ES:
3Dmark06:
A07: 19449
A08: 17748
Performance loss: 8.8%

920XM OEM, wPrime 1024, 4 Threads
A07: 362.256 sec
A08: 434.692 sec
Performance loss: 20%

 

maybe selecting just the cpu test will be more informative but 3dmark06 as we all know is very cpu based unlike vantage