Review Of A Few Of The Latest Nvidia Drivers (incl* Xtreme-G) - Overclock.net - An Overclocking Community

Forum Jump: 

Review Of A Few Of The Latest Nvidia Drivers (incl* Xtreme-G)

Reply
 
Thread Tools
post #1 of 380 (permalink) Old 04-08-2011, 07:14 PM - Thread Starter
Retired Staff
 
munaim1's Avatar
 
Join Date: Dec 2007
Location: .:.
Posts: 10,805
Rep: 921 (Unique: 603)
Review of Nvidia Driver's *inc Modded drivers (Updated Regularly)


I figured why not create a thread to test how Nvidia drivers are performing. Since coming from ATI (lets not go there regarding driver issues rolleyes.gif)I have not had the chance to really see what Nvidia is capable of. I remember back in the day's when I use to go through several drivers before actually being satisfied and even then I used to go round asking, 'what are the best drivers'

Anyways I will try my very best to update this regularly and present my findings as clearly as possible. I have tried my very best to maintain the conditions and the settings when running benchmarks, however, anomalies may appear here and there. I have also tried keeping 'things' consistent when comparing these drivers.

Comments and suggestions on how to make this thread better is always nice.

Just a note these drivers have been tested using my sig rig, cpu at all times at 5.1ghz, however core clock varies between some benchmarks.

Hope you enjoy and find my results helpful. Now lets get started


SINGLE CARD TESTING

1. Xtreme-G 266.58

2. Xtreme-G 267.24

3. Xtreme-G 270.51 (Must be BETA aswell)

4. Nvidia 266.58 (WHQL)

5. Nvidia 270.51 (BETA) - COMPARISON 1-5 HERE

6. Nvidia 267.60 (WHQL) - COMPARISON HERE

7. Nvidia 268.03 (WHQL with modded inf) - COMPARISON HERE

8. Nvidia 270.61 (WHQL) - COMPARISON HERE

9. Nvidia 270.71 (WHQL with modded inf) - COMPARISON HERE

10. Nvidia 270.73 (WHQL with modded inf) - COMPARISON HERE


17/05/2011

Since I got my second card I decided to introduce the SLI testing in this thread. I will be combining both Single and SLI testing in one, however, i will be using the above data for future testing. From this point forward I will not be doing single card testing separately so please refer to 'SLI & Single Testing'

I will continue making recommendation's for both single and SLI user's thumb.gif


SINGLE & SLI TESTING


1. Nvidia 266.58 (WHQL)

2. Nvidia 270.51 (BETA) - COMPARISON BETWEEN 1 AND 2 HERE

3. Nvidia 275.27 (BETA) - COMPARISON HERE

4. Nvidia 275.33 (WHQL) COMPARISON HERE

5. ~~***NEW***~~ Nvidia 280.19 (BETA) COMPARISON HERE






Some drivers are causing folders a few problems, please refer to the folding section for more info

Quote:
Originally Posted by juano;15358901 
FYI Munaim1 the problem CSM mentioned with the drivers getting stuck at 405Mhz is problem that all of the 270, 275, and 280 drivers have had during folding. When you pause and then resume the client they will get stuck at 405Mhz almost every time requiring a reboot and will occasionally do it randomly without even pausing too. They really suck for folding, luckily the 285s don't have that problem but they aren't the most reliable as of yet.





**~**NEWS about upcoming driver's**~**

Quote:
Originally Posted by Anth0789;15373263 
Quote:
FYI. New drivers will be released the day before Battlefield 3 launches. Make sure to grab them especially if you will be purchasing Battlefield 3 as there are numerous bug fixes and better performance as well. For those who have modded the latest Quadro driver (http://www.nvidia.com/object/quadro-tesla-win7-winvista-64bit-285.58-whql-driver.html) it is very close to the final driver but is missing a few Battlefield 3 fixes so if you do download and mod them, make sure to update to the newest Geforce drivers next week. The Quadro driver does contain the fix we've added to address the TDR issue.

http://forums.nvidia.com/index.php?showtopic=211130&view=findpost&p=1310428





**~**Driver Sweeper**~**
Quote:
Originally Posted by Anth0789;15306742 
Driver Sweeper gets updated look here:
http://phyxion.net/Downloads/





**~**How To Unlock Femi Voltage**~**




Firstly you want to download a couple of things and make sure you have a usb stick at hand:

Nibitor 6.02- this one worked for me HERE
Nvflash - HERE
GPU-Z - google it lol
Hp utility with win98 - HERE (you need this in order to make the usb bootable)


Now you have everything we can begin.

First and foremost we want to save the original bios of the card, people make the mistake of using gpu-z which can cause issues, something to do with the checksum size (thanks brad) so what we have to do is use nvflash to save it.

You can go ahead and download the HP utility, it will have the win98 files in it to make it bootable, for instructions on how to do that click HERE

After you have done that, go ahead and copy the nvflash files (around 5 files I think) on to the USB.

Restart the pc and change the boot priority to the usb, once it's loaded you will be presented with a commant prompt.

Type this in to save the original bios of your card to the usb - nvflash -b backup.rom

This should save your original bios. Exit and restart windows and in the usb you should find the backup.rom file.

Now this is where you need to pay attention, open up Nibitor and open the .rom from it

You should get something like this:
Capture117095.jpg


To get the voltage editor click tools then Femi voltage. The only thing to change is tab right at the top, scroll down and select the maximum voltage you require.

Hit apply and then save the bios as modded.rom or something to your desktop. Then what you want to do is copy the modded one on to the usb, at that point you can make a backup of your backup.rom somewhere on your pc.

Now we want to do the same as before, restart pc and boot with usb. this time round there is a few commands that you must key in to flash the bios with your modded one.

nvflash -r

nvflash --eraseeeprom

nvflash -4 -5 -6 *.ROM (* denotes the filename for the BIOS, in this case will be modded.rom)

that's it biggrin.gif

If you have any problems refer to this thread Here (I always recommend saving and flashing in dos, however, I encounted errors when trying to flash in dos, so it left me no choice but to flash within windows. People will say it is risky flashing in windows, however, I just want to say that I have done it coutless times and had no problems. This should only be done as a last resort. As said beofre please refer to the thread I linked if you are having similar problems. thumb.gif

This was a quick write up, so if I have made any mistakes anywhere please let me know asap but I will go throught it again.






Hope you guys enjoy the 'reviews' and if you have any questions fire away. thumb.gif
munaim1 is offline  
Sponsored Links
Advertisement
 
post #2 of 380 (permalink) Old 04-08-2011, 07:53 PM
New to Overclock.net
 
XxRZxX's Avatar
 
Join Date: Jan 2011
Location: St. Louis
Posts: 502
Rep: 19 (Unique: 18)
Thanks for the information!
XxRZxX is offline  
post #3 of 380 (permalink) Old 04-08-2011, 08:01 PM
New to Overclock.net
 
TickleMeElmo's Avatar
 
Join Date: Dec 2010
Posts: 1,595
Rep: 118 (Unique: 104)
I don't think any of the differences between the drivers are statistically relevant. Please resize the graphs you have so that the differences don't appear larger than they are.

TickleMeElmo is offline  
Sponsored Links
Advertisement
 
post #4 of 380 (permalink) Old 04-08-2011, 08:08 PM
 
Join Date: Jan 2011
Location: SoCal
Posts: 1,229
Quote:
Originally Posted by TickleMeElmo;13056082 
I don't think any of the differences between the drivers are statistically relevant. Please resize the graphs you have so that the differences don't appear larger than they are.

Why? It's clearly meant to make the minute differences more distinguishable. Nvidia/AMD do this all the time


Shredicus is offline  
post #5 of 380 (permalink) Old 04-08-2011, 09:17 PM - Thread Starter
Retired Staff
 
munaim1's Avatar
 
Join Date: Dec 2007
Location: .:.
Posts: 10,805
Rep: 921 (Unique: 603)
Here are my findings for Single card drivers 1-5:

Metro Benchmark tool - DX11 - 1920x1080 - Tesselation (enabled) - Very High - AAA - AF16x - No Physx or DOF

GPU at 960c/1920s/ 2050m

Xtreme-G 266.58
01AgU.png


Xtreme-G 267.24
XLotk.png


Xtreme-G 270.51 (must be beta aswell)
TgrG1.png


Nvidia 266.58 (whql)
6UcQz.png


Nvidia 270.51 (beta)
5WGAx.png


nRPa2.png

Seems as though the nvidia 266.58 and xtreme-g 270.51 perform the same.confused.gif



Now lets see how these drivers compare to the epic games of crytek, crysis and crysis warhead biggrin.gif. GPU @ 870/1740/2050

Xtreme-G 266.58
6FVch.png
QQmXp.png


Xtreme-G 267.24
HaMyL.png
arFO2.png


Xtreme-G 270.51 (must be beta aswell)
rpqDm.png
M6Qbn.png


Nvidia 266.58 (whql)
lN0nw.png
P05rO.png


Nvidia 270.51 (beta)
UnhAA.png
3MmLu.png


nOBvl.png

All the data into one. Still confusing I know rolleyes.gif There is only a marginal difference between the drivers in those two games.


Taking ss while gaming proved to be a little difficult, as you can see. I'm only going to show a few ss out of the 20 I took but all the data well be presented in a graph. While in gaming the GPU clock speed are 870/1740/2050 on all the rest.



I used crysis 2, metro (dx10) and Homefront. These are the in game settings for all three.

xb7L8.png

MASTy.jpg

MMoqt.jpg



Again here are a few in game ss.

Mg4XG.jpg

uSJSK.jpg

Dh3rn.jpg

0eERJ.jpg



yEyVB.png


All the data in the graph. For some reason crysis 2 just stayed at 47 on all 5 drivers, maybe I did something wrong confused.gif but it does seem like the nvidia 270.51 is doing better than all rest on average which is closely followed by xtreme-g 270.51.



Lets take a look at how they performed in Vantage, again GPU at 870/1740/2050.


Xtreme-G 266.58
qfEyM.png


Xtreme-G 267.24
KhKwr.png


Xtreme-G 270.51 (must be beta aswell)
xwrmi.png


Nvidia 266.58 (whql)
AN7pg.png


Nvidia 270.51 (beta)
HtoDd.png



4ZTr5.png

Looks like Nvidia's 270.51 just marginally won, however the difference between these driver are minimal when it comes benching, however the same could be said about gaming. I'm sure that each of these driver shine differently in different games but I still have yet to discover that. This doesn't really help pick a driver, it's just what I found. Maybe it'll help maybe it wont, but I can see that there has been improvements going from both the Nvidia's 266.58 and XtremeG 266.58 to eventually the 270.51 (from both Nvidia and Xtreme-G)

Thanks for reading biggrin.gif
munaim1 is offline  
post #6 of 380 (permalink) Old 04-08-2011, 09:56 PM
New to Overclock.net
 
TickleMeElmo's Avatar
 
Join Date: Dec 2010
Posts: 1,595
Rep: 118 (Unique: 104)
Quote:
Originally Posted by Shredicus;13056156 
Why? It's clearly meant to make the minute differences more distinguishable. Nvidia/AMD do this all the time

Their marketing departments do this because it helps to bring attention to small differences that may or may not be statistically relevant. In the pursuit of truth (and the relevance of any differences), one should not take the same approach as the marketing departments which have a vested interest in highlighting minute differences.


@OP: I'm saying that if you only ran each benchmark once even a 5-10% difference cannot be said to relevant with any kind of statistical certainty. Because of the nature of the benchmarks and how background processes can influence the results, you need a larger sample size for your results to be indicative of any pattern.

TickleMeElmo is offline  
post #7 of 380 (permalink) Old 04-08-2011, 10:20 PM - Thread Starter
Retired Staff
 
munaim1's Avatar
 
Join Date: Dec 2007
Location: .:.
Posts: 10,805
Rep: 921 (Unique: 603)
Quote:
Originally Posted by TickleMeElmo;13057157 
Their marketing departments do this because it helps to bring attention to small differences that may or may not be statistically relevant. In the pursuit of truth (and the relevance of any differences), one should not take the same approach as the marketing departments which have a vested interest in highlighting minute differences.


@OP: I'm saying that if you only ran each benchmark once even a 5-10% difference cannot be said to relevant with any kind of statistical certainty. Because of the nature of the benchmarks and how background processes can influence the results, you need a larger sample size for your results to be indicative of any pattern.


Cool thanks for the info, i'll bear that in mind next time.cool.gif
munaim1 is offline  
post #8 of 380 (permalink) Old 04-08-2011, 10:39 PM
 
Join Date: Jan 2011
Location: SoCal
Posts: 1,229
Quote:
Originally Posted by TickleMeElmo;13057157 
Their marketing departments do this because it helps to bring attention to small differences that may or may not be statistically relevant. In the pursuit of truth (and the relevance of any differences), one should not take the same approach as the marketing departments which have a vested interest in highlighting minute differences.


@OP: I'm saying that if you only ran each benchmark once even a 5-10% difference cannot be said to relevant with any kind of statistical certainty. Because of the nature of the benchmarks and how background processes can influence the results, you need a larger sample size for your results to be indicative of any pattern.

Yes, I am well aware of why their marketing departments do this. In fact the preceding sentence explains their motivations. On the same token you seem to have forgotten the credo of the forum you are currently posting on. the Pursuit of Performance. Why do you think overclockers talk about adjusting their clock speeds by 5mhz? Because it will net them a 20% FPS increase, or 400 points in a benchmark? No, they do it because they constantly pursue performance, tangible or not. Overclockers are people obsessed with minutia for the most part, so denigrating his test results due to them being insignificant is essentially a moot point on this type of forum.


Shredicus is offline  
post #9 of 380 (permalink) Old 04-09-2011, 12:39 AM
New to Overclock.net
 
TickleMeElmo's Avatar
 
Join Date: Dec 2010
Posts: 1,595
Rep: 118 (Unique: 104)
Quote:
Originally Posted by Shredicus;13057557 
Yes, I am well aware of why their marketing departments do this. In fact the preceding sentence explains their motivations. On the same token you seem to have forgotten the credo of the forum you are currently posting on. the Pursuit of Performance. Why do you think overclockers talk about adjusting their clock speeds by 5mhz? Because it will net them a 20% FPS increase, or 400 points in a benchmark? No, they do it because they constantly pursue performance, tangible or not. Overclockers are people obsessed with minutia for the most part, so denigrating his test results due to them being insignificant is essentially a moot point on this type of forum.

No the point I am contending is quite different from what you mention.

Increasing an overclock by even 1 MHz (assuming you have a clock generator that does 1 MHz increments) is a definitive increase in a performance metric. Unless your card is faulty, there is not going to be any change in the clock that you set it at.

However, in a normal operating system, there are a lot of background processes vying for the different system resources/eating up bandwidth. This introduces an element of variability that is impossible to control even if you have a clean OS with no additional programs running. This is why a 200 point difference on a 10 000 score may or may not be significant, as you sure as hell can't tell with a single run unless you know the variance beforehand.

TickleMeElmo is offline  
post #10 of 380 (permalink) Old 04-09-2011, 04:55 AM - Thread Starter
Retired Staff
 
munaim1's Avatar
 
Join Date: Dec 2007
Location: .:.
Posts: 10,805
Rep: 921 (Unique: 603)
Wow when I said discuss i didnt mean the way i conducted my tests but about the results 'if you want', you have a problem or opinion about it pm me, dont broadcast it because it sounds like *****ing, quite literally handbags coming it out and everything. This seems to be happening in ocn very frequently nowadays and its getting a little annoying. I didnt have to share what i found but i thought why not maybe it'll help maybe it wont.

If You have a problem pm me.
munaim1 is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off