Modified vbios for GTX580M at widezu69's request

Also i have tested the new matched vbioses provided by Saltius and here're my test results for 800/800v 0.92v bios:

the GPU score in Vantage was as low as 25.7k and in 3dmark11 for GPU - 7.8k, it's the same as if i used a 750/800 clocks based on my prvious results.

The reason for that is very simple, during all the tests i can see some micro stuttering every few seconds and the fps drops down 10-20% and goes back on. I assume it swithces from P0 state to P1 and back again causing those fps dips. So for me personally it didn't solve the throttling issue and it gets really annoying wathcing the screen freezing every 3-5 sec... i remeber 0x29A experienced the same issue a while ago, post #50!

don't really understand how Xenize and bar-code received such a good scores in vantage (28k for GPU) and i'm really happy for u, guys... but as for today i don't really have any solution regarding my freeze issue!

EDIT: here's my 800/900 run with 0.92v bios, it's really too low for the GPU score...

 

Hopefully it is NVIDIA's implemented throttling only for those benchmark/stability tests. Otherwise we need other solutions and further testing.

 

thanks i will try that when i get off of work! do you think that 5.95 or whatever i'm currently using, doesn't have the code for identifying the 580m's (since they are relatively new)?

 

ummmm, what good score? his best scores are with stock vbios 28.8k with lower clocks. hell...my scores are better when i teamviewer in and bench his laptop.

 

Yeah my best scores were not with this Modded Bios... in fact we were reviewing this last night with John and as it turns out I have a 28900 GPU (Vantage) score with the original VBios. That said I did get a 7700 (3DM11) @ 800....

 

As said, this is a temporary fix. The 'stuttering' will not go away.

 

more like a delay of game....

 

I don't see the stuttering in the original mod Vbios (620/1500/0.870) even when I OC to 744/1750...

I'm wondering if at all possible we could get a mod with the same clock speeds but a 0.890v or 0.900v...

Just wondering if the extra voltage is an issue.

 

well sir....get ta testing...

 

test what? lol

 

testing that micro stuttering throttle...what else...

 

it's there withy 0.92 regardless of the OC... I'm getting that Kilowatt meter today, we'll see if power has anything to do.

 

I think this isn't about power at all. This is just throttle kicking in -> the stuttering is caused by switching between P0 and P1. That's milliseconds, yet it is clearly visible on load graphs.

 

I would agree except it doesn't happen AT ALL with the original Mod. I OC up to 750/1800 @0.87 and I do see the clocks revert back with no issues.

Also stuttering is every 5 seconds.... the switching of clocks is not.

Another thing to note is that the load is never ever 100% even at stock clocks, I probably made 20 runs and never hit 100% load, so I guess there goes the 100% load theory. If there is a load embedded in the formula is not 100%, it's less

 

Good! A Kilowatt meter can tell many things.

 

Is the 'original' mod stock clocks P0=620Mhz/P1=620Mhz ? If so, it does happen. I experience micro stuttering in STALKER CS. Use a very accurate tool so you can observe changes of clocks and load in milliseconds. Note that most tool refresh at a second or less. 100% load in the equation throttle has already been confirmed by Dell, so I guess there's no discussion there needed. 5 seconds figure is correct. That's pretty much what I observed too.

So in the end maybe there is yet another condition to be met there? You running 0.92v? Maybe at that high clocks overvolting protection comes into play? To be perfectly honest I am so tired of the whole throttle bull$hit. There is always something there. If not one thing than another. Surely, running any temperature test with the stuttering issue present is pointless.

I think I will just wait for that talk with Dell people about solving the throttle issue. I am done for now with the whole 'throttle research'. All's been said. We just keep running in a loop there. Like mice

 

I hear ya, I'm worn out as well, but I can assure you Dell confirming 100% is annacurate. MY throttling will kick in and I never see 100% load... figure that one out.

Yes I experience the Micro St. ONLY with 0.92v. All I'm trying to say is that the stuttering seems to happen every 5-6 seconds while the clock changes from P1/P0 only happen once maybe tweice a run.... see where I'm going? It makes no sense.

 

Same I am so sick of this throttle BS, the only one that should exist is the thermal protection one. I have a dark feeling that Dell put in this whole power/load throttle because the 580M's needed a larger power supply than 330W. Anyways we will find out today when Xeneize hopefully is able to Kill A Watt his m18x.

 

I've updated the test vbios in post#53

Specially added a 725-900 vbios at stock 0.87v.
Changed 850-850 0.92v to 825-950 0.92v
http://forum.notebookreview.com/ali...-gtx580m-widezu69s-request-6.html#post7978530

 

Thanks!, I'll give these a run.

Is it possible to have a 0.089v? or something in between 0.87 and 0.92?

 

Mavke thinks no while I am not sure.

NVIDIA GeForce GTX 580 570 560 460 ATI Radeon HD 6970 6950 6870 5970 :: MVKTech :: On the Edge of Technology - Re:tweak GTX580M BIOS voltage - MVKTech Support Forums

 

he is right...if there is no physical table to edit it's regions...then just changing the label wont work...

you have that on your 9800m gtx as a matter of fact.

 

Thanks Saltius.

 

Seems like one of my adapters doesn't like high memory clocks, artifacts appear all over the place if i set to 1900Mhz, so for now 1850Mhz is fine, but i may try even lower value to check how far can i go with the core clocks. Here's my last run 0.92v 800/1850:



The temps saty amazingly low, with both the system and the laptop pad fans on full blast the temps with the OC are 66/60C for the GPU's)

EDIT: looks like i got to the limit of my GPU's, the MAX OC is 830/1850 at 0.92v which is pretty neat. I went for Unigine 2.5 bench, cause i didn't notice any stuttering there like i had in 3dmarks. Also the temps peaked at 67/61C for the cards during the run

 

Update the vbios in the top post.

Two new zips that contains all the AW version 580M vbios I've modified.

 

As always outstanding work Saltius the 0.85v mod one looks interesting less volt less heat and more 700mhz tnx

Hey Dell are your paying attention here!!!

 

Thanks Saltius!!

011 Results:

750/1750 @0.880

KiloWatt Meter Results
Vantage GFX test #1: Max: 292 -Avg: 280
Vantage GFX test #2: Max: 295 -Abg: 265

Temperatures were fine: 76 both GPU max.... (sorry the screenie sucks)

Very low results/performance overall, stock BIOS performs at ~28k GPU at 740/1700/0.870 as a reference.
The low performance is mind boggling....my CPU clocks are 45x, we should be looking at ~29k GPU scores.... Any ideas what's up here?

I Did not unistall the drivers, is that even a factor?

 

If possible.

Try three vbios in the top post.

1.AW_GTX580M_06_725_900_087.ROM
2.AW_GTX580M_06_750_850_092.ROM
3.AW_GTX580M_06_800_850_092.ROM

Report your vantage and the KiloWatt Meter Results with the same CPU set.

 

you got it

 

725_900_087


GREAT SUCCESS!!!

My Best GPU score in Vantage so far and not in optimal thermal conditions (Room temperature close to 26c ugh)

Vantage Run

KILL A Watt Results:

GPU Test #1:
Max: 304 (peak, but for a sec then never went above 295)
Avg: 285

GPU Test #2
Max: 288
Avg: 275





3DMark 2011 Run

KILL A Watt Results: (Max/Avg)
Test #1: 276/272
Test #2: 261/245
Test #3: 250/245
Test #4: 257/248
Test# 5: 228/200
Test #6: 287/280

 

Xeneize what do you think is the explanation for getting a higher score on a lower core clock and voltage at 0.87v vs 0.88v

 

Ask Saltius or John.... I have no idea... My only thought could be a load throttling by Nvidia... hardcoded... who knows. No idea.

 

Alright, thanks will look into it.

 

Ok went for the 800_900_092

And took my Laptop to an AC room

Awesome results!! smashed my record by 1300 Vantage points!

Kill A Watt showed a max of 328 and averaged 315, no hickups whatsoever, I'm thinking we got room for 850_900_092 with the PSU.

The 580M are simple amazing! The fact that you can OC 30-35% is incredible!, thx for this BIOS, I'll keep pushing once I get my CPU temps under control, but the GPU will definitely break 31k as Johnksss showed provided the right CPU raw power.

 

yep...once you move your cpu up a bit in power...you should start maxing out around 350w to 360w.. which is what i was hitting with my other killowatt meter. but those will be spikes not just sitting there.

great job there sir!

 

Thx I guess you have really shown the potential of the 580M SLI/XM CPU, I think our cap will be around 850/900 @0.92... which is insane lol, but Dell didn't do that bad of a job with the 330w after all...

 

Good job both of you with that wattage tests. \o/

 

as in the words of scook9...to da freezer!..er ah i mean... to the mirc....

so far 845/L/1000 on single card.
825/L/2000 on dual card...but i was tired and called it a night.

thanks x29a

 

Johnkss or xeneize have you guys noticed a difference in overclocking via vbios vs using a tool like nvinspector. Maybe that is causing the disparity in results?

 

that's what i call a bad run:




no way i can get higher than this...

 

what are your temps?

 

with the system and pad fans on full around 66/61C for the GPU's, so it's not an issue

 

those scores don't sound right unless the CPU is indees the bottleneck.... wha'ts your 3DMark2011 score with 800_900?

3DMark 2011 scores are far less CPU dependant.... you should be around 7500.

 

that's my best 3dmark11 run so far



but downclocks are all over the place (every 3sec), due to switching between P0 and P1 states

 

Ahh I bet you're not using the new modded files on posy #1 in this thread, dod you try that one? I do not see the drops with the new mods.

 

how about this....are you sure your setting all four places when over clocking?
you have to make sure it takes in p3/p2 gpu 1/0

 

sure, i set both P0/P1 levels for GPU's 0/1. also i used the custom bioses (725_900_087 and 800_900_092v) where all the clocks are matched, so there's no chance i messed up with states/clocks, i left it at default

 

i may have to match yours and see what the problem is.

 

null null null

 

You got issues that no fan speed increase will resolve.... I've ran the cards at 850/1800/0.92 and during peak temps it never ever went over 85 and that's with room temps @ 80f.....