Intel X3100 users rejoice! It's finally here! New Pre-Beta Drivers

I took that into account, so one of my benchmarks was turning everything off as I could which in theory it will help me to compare it with win2008, but after seeing the results , I just decided win7 + x3100 not useful with me. However, it was a nice experience.

 

I am wondering if anyone have this file GFX_Vista64_15.13.3.1787_PV_Intel.exe (Intel driver - v15.13.3.1787 - for vista64)
if someone have it can he/she upload it for me?

 

whats the difference with that driver?

 

the newest win7 drivers causes pixelations during startup for me after the swilring windows flag at the top of the screen? vista drivers dont do this

 

In general it's available here:
http://www.opendrivers.com/driver/2....64.1787-windows-vista-x64-free-download.html

But in this case it links to the non existent intel download page. It's not yet mirrored at opendrivers.

GFX_Vista64_15.13.4.64.1829_PV_Intel.exe is available at intel. But I guess that's not the one you are looking for, is it?

 

also note the last vista driver for 64bit does not hold the color settings after a restart on windows 7, i have to go back in and click on one of the sliders and they correct themselves. othrwise they seem okay

 

It was the best driver in performance that worked with me (under win2008-32bit)

Unfortunately I already have that driver and the one I need isn't anymore on intel server, also tried other websites still couldn't find it.
Anyway Thanks for your help



Edit:
For those who want the driver please take it from this link:
ftp://ftp.zdnet.de/pcpro/treiber/grafik/intel/GFX_Vista64_15.13.3.64.1787_PV_Intel.exe

 

the vista drivers though running under windows 7 will not remember color settings for me for some reason i have reinstalled 7 a few times now thinking i messed it up by installing and removing so many drivers.. the windows 7 driver works fine with that but i get the artifacts but only at that one startup screen after that there are none??

 

zfactor you might want to try this modified driver (v.1787) http://forum.notebookreview.com/showpost.php?p=5020310&postcount=1752 (32-bit) unless you have 64-bit running you will need to get igdlh.inf from the given link then extract the 64-bit (check the post before yours for the link) then rename igdlh.inf to igdlh64.inf and replace it with that, that's all.
Good luck

 

Very cool, Server 2008 is fun to play around with. How large was the difference in benchmarking scores?

 

Windows Server 2008 can be used as normally OS? I mean for gaming, using photoshop ... ?

 

Well, not much but in 3dmark03 (battle of proxycon), I can get from the begging of the scene (15 secs of start) an average of 5~7 FPS while on modified driver (1787) on win2008 it easily averaged 11~12 FPS with exception that it dropped once or twice to 9 FPS ,which still higher than the win7 drivers. if you will compare the results between the modified 32bit driver (1787) in win2008 and win7 (64bit) you will notice an average of 9 FPS in the same scene.

on Performance-Test (this from what I can recall and NOT accurate but you can take my words on it):
(First 3D test)
Win2008-32bit-No Aura - Driver 1787(modified) *installed without rebooting*: 170+/- 1%
Win7-64bit- Original Driver + Aura: 144 (Without Aura add 10)
Win7-64bit- 1825 Driver + Aura: same as above +4~5
Win7-64bit- 1787 Modified Driver+ Aura: 144.1 *Fresh Test!*
Win7-64bit- 1787 Modified Driver+ Without Aura: 161.0 *Fresh Test!*
Win7-64bit- 1787 Modified Driver *installed without rebooting* + Aura: 159.7 *Fresh Test!*
Win7-64bit- 1787 Modified Driver *installed without rebooting*+ Without Aura: 171.5 (Strange? without aura here the difference is 11.8) *Fresh Test!*




Ok something interested I just noticed it! with the same modified driver on win2008 (32bit) and win7(64bit) they both averaged the same score! I even tried it on 3dmark03 and I can tell you where I had before in a part of the 2nd test where I was getting 12 FPS now I can get 14 FPS with drop to 11! (I did the test with firefox + other stuff running so without it I can assume it will be even better!)
for those who might wonder how I can compare 32bit with 64bit ,I already mentioned in one my posts that the difference between benchmarks of those two were almost close to each other which you can neglect

Edit:

You will have to install other parts and do a little tweaks to make it more like vista (search google for windows 2008 workstation) about programs you might have to run some under compatibility or modifying the installer to get it working fine under win2008. For gaming I wont recommend it ,for real work I say go for it (Remember Linux is something to think about)

 

7oby, notice here how the combination of new drivers and better developer support allows the GMA 4500 to be much more competitive: http://www.anandtech.com/mobile/showdoc.aspx?i=3618&p=8

ET: QW used to be 7 fps(albeit with 1024x768) on a desktop G45, but now gets 12.7 fps on a laptop. What used to be average of 4x away from the best competing IGP solutions, now its only slightly over 2x away.

Games like Demigod, Empire: Total War, Ascaron's Sacred 2: Fallen Angel has been optimized with new tools like the GPA and are much more competitive.

2x more performance and they'll be at 780G levels. 3-4x and they'll take the crown.

 

Did drivers really improve? The ET:QW you reference ( click) use 1024 x 768, 8x AF, most likely not Minimum Settings and a QuadCore CPU. I'm having a hard time to tell whether drivers really improved or whether the completely different setup and settings (I'm in particular thinking of the memory bandwith hungry AF setting) account for the different observations.

And even if it did: The OpenGL stack is definitely not one, which received performance improvements. ET:QW still runs 4x slower on intel hardware in the mobile chipset comparison on anandtech.

The open source linux drivers did get support for the upcoming Clarkdale & Arrandale integrated graphics engine
http://www.phoronix.com/scan.php?page=news_item&px=NzQ1Ng

That's the graphics core in the upcoming 32nm Core i3/i5/i7 CPUs with integrated graphics:
http://en.wikipedia.org/wiki/Intel_Nehalem_(microarchitecture)#32_nm_.22Westmere.22

I did browse through the source code a little bit. I can't tell how many execution engines the new graphics engine will have, but it looks like a very small improvement compared to the existing Gen4 intel graphics (e.g. G45, GM45). While G45/GM45 is designed in 65nm, Clarkdale/Arrandale GPU die will be designed in 45nm (remember the cpu core will be 32nm). I expect some improvements in GPU render clock and maybe an upgrade to the number of ECUs. Due to the 45nm process I'd expect a 800MHz -> 1GHz jump in renderclock (for desktop chipsets). Since it is integrated in the heat spreader design of the CPU and connected to a powerful CPU cooler, I'd expect an additional render clock jump - maybe up to 1.2 GHz.

I've observered linux graphics stacks for quiete some time now. In particular the open source vs. proprietary AMD/nVidia ones. From this I can tell: The software stack is a similar asset as the hardware design. I'm convinced that intel's proprietary Windows drivers are many man years development time behind the quality and performance of proprietary AMD/nVidia drivers. Even if intel is able to manufacture the perfect GPU design in hardware (that's one of intel's skills) in the very near future, they won't catch up on the softwareside and loose a direct comparison regarding performance (no matter how smart their hardware is).

Still: I like intel hardware.

 

There's already an idea what the clock speeds will be. Xbitlabs showed 3 versions with 533MHz/733MHz/900MHz clock speeds and PCWatch had the mobile version as 166MHz/266MHz/500MHz. The 3.06GHz version(733MHz IGP) achieved 1944 on 3DMark06 with 1024x768 resolution, which can improve with final hardware and drivers.

Yes they did.

Older benchmark(difference between Q8200 and Q9550 is ~0): http://www.tomshardware.com/reviews/g45-geforce-9400,2263-7.html

Here's the comparison between two benchmarks from same site using the same E7200 CPU:

http://www.tomshardware.com/reviews/G45-780G-Efficiency,1999-9.html

WiC 1024x768 Low: 22 fps

http://www.tomshardware.com/reviews/amd-785g-chipset,2381-12.html

WiC 1024x768 Low: 28 fps

It even shows that the GPU is now comparable to the Geforce 8200 IGP.

Crysis used to run at single digit performance even at low:
http://www.pcstats.com/articleview.cfm?articleid=2365&page=13
http://hothardware.com/Articles/AMD-790GX-Chipset-Platform-Launch/?page=7

Tomshardware puts it as 16 fps and the game is now on Intel's page which for driver changes only is quite a large difference!

Look what the conclusion was by AT: "It wasn't long ago that Intel IGPs failed to run most games - even many casual games. These days the GMA 4500 is only about half the speed of ATI's HD 3200 solution. Their drivers have come a long way in the compatibility department, and another revision or two to their hardware could see Intel close the gap."

 

wait so download the intel 64bit 7 driver or vista 64 last driver?

then extract the intel driver and take that file from it rename it then put it into the modded one?

or take the file from the modded 32bit one and add it to the newest intel driver? im thinking this is what you meant but i want to be sure.

and im running win7 64bit.

thanks for the info.. how bad does it degrade after a reboot?

 

Just download it from here (This is a little bit different than others as compressed a lot *sigh* low upload speed , that's why).

Intel Modified Drivers: (64-bit only) - Vista / Windows 7 - For those who want 32-bit go to page 176
Version:15.13.3.64.1787
Size: 10.7MB /It will requires a little bit more ram to unpack it (~100MB)

Howto: extract the rar and run "RunMe.bat" , that's all go to the new folder and start the setup and when windows ask you if you want to install this driver click on install it.



http://www.rapidspread.com/file.jsp?id=1thtku1b2z
http://www.fileducky.com/bHAWCDWA/

 

downloading now is this modded already or do i still need to mod it

thank you

 

Let's go donate GMABooster author
http://gmabooster.com/download.htm

 

Its already modded so nothing else you need to do.
Your welcome

 

This is going to get very interesting.

As I've previously shown (and others as well) intels graphics is very sensitive to dram bandwidth. Although triple channel memory doesn't have any impact on general purpose desktop applications ( click) this might change with the new IGPs. Considering the Core i7 goes up all the way up to DDR3-2000 ( click) this might indeed turn out to be very challenging for what AMD/nVidia will offer on the Phenom IGP side.

--

Vista 32-Bit v15.13.5.1861 dated 08/06/2009
Vista 64-Bit v15.13.5.64.1861 dated 08/06/2009

As it has been with the v15.13.4 drivers as well, those are useless for X3100 (GM965) users except for modding. Only useful for

Code:

*	Intel(R) B43 Express Chipset
*	Intel(R) G41 Express Chipset
*	Intel(R) G43 Express Chipset
*	Intel(R) G45 Express Chipset
*	Intel(R) Q43 Express Chipset
*	Intel(R) Q45 Express Chipset
*	Mobile Intel(R) GL40 Express Chipset
*	Mobile Intel(R) GM45 Express Chipset
*	Mobile Intel(R) GS45 Express Chipset

No changelog due to drivers from desktop mainboard section.

 

And I have a very precise idea of how many ECUs it will feature: 12

Code:

      /* WM maximum threads is number of EUs times number of threads per EU. */
      if (BRW_IS_IGDNG(brw))
         key->max_threads = 12 * 6;
      else if (BRW_IS_G4X(brw))
	 key->max_threads = 10 * 5;
      else
	 key->max_threads = 8 * 4;

http://cgit.freedesktop.org/mesa/me...c?id=3e2b6a204966b962c9881e90fe3f0b74cf84d8c4

 

Maybe the current iteration of the architecture isn't limited by EUs at all(akin to how GMA 950 wasn't fill rate limited). GMA 4500 got quite a bit of extra performance that's not just clock speeds and extra EUs. They are looking to get 3x performance out of the IGP in Clarkdale with only 20% increase in EUs.

 

I think this was you:

http://forums.anandtech.com/messageview.aspx?FTVAR_FORUMVIEWTMP=Linear&catid=31&threadid=2276310

I can image also a performance gain of up to 3x if you pair it with DDR3-2000.

I'm also convinced that intel did the right decision regarding for the number of ECUs. GPA on Gen4 allows to display the workload of the GPU. The hardware has this monitoring facility and intel is smart enough to use it to their advantage. Certainly ECUs limit to some extend. Going HW TL -> SW TL and freeing ECU resources this way in some apps proved to be faster already.

Arrandale/Clarkdale does have some more registers here and there, but it's architecture is only a minor evolution of the Gen4 core. Certainly it uses the same drivers (on Linux). Together with the high bandwidth Nehalem memory interface it will rock to some degree. My Linux boxes will get Arrandale/Clarkdale

 

You realize that the memory interface is dual-channel? BTW, I think the score was achieved by using DDR3-1333. The only way that Nehalem architecture would help is if integrating the Northbridge somehow gives a big bandwidth advantage over not doing it. I mean, even in the Northbridge solutions have the IGP right next to the memory controller so how would it help bandwidth in this case? The IGP is supposed to use DMA for faster-than-FSB access. Unless DMA is neutered to save costs and most real transactions go through FSB. The CPU benefits from being closer, but with simple inspection the GPU is not.

It also has more fixed function units to "significantly improve vertex throughput"(which would help against hardware vs. software VS), and improved extended math. I think they were talking about how extended math being the weak point with one of their presentations.

Why do you call it the ECU? They call it the EU.

I agree it will "rock". They might actually make a good impression with IGPs for the first time ever. But the question is how they really do it. I'm waiting for Fall IDF.

Update: Regarding this comment by me(yes it was me)

"I lost most of the hope for Intel IGPs but the least I don't think the gain won't be as pathetic as G35 to G45 was."

I kinda want to retract the comment. I was honestly impressed:

http://www.youtube.com/watch?v=wKiVxtsW09w&feature=related
http://www.youtube.com/watch?v=5F2hmXs_R9o&feature=related
http://www.youtube.com/watch?v=YmUzRRsjplE&feature=channel_page

Intel IGPs has come a long way from being incompatible with some existing games to being able to at least run the new games. The 4500 is substantially faster! Kinda shows better future for Clarkdale.

 

You are right, there's only the dual-channel option and maybe even only bandwidth of up to DDR3-1333.

Currently the IGP is integrated into the MCH, but it doesn't have direct memory access. Instead the MCH operates as a switch with a high bandwidth to mediate between all interfaces. If the IGP moves to the CPU at least the CPU <-> GPU interface will double in speed from 6,4 GB/s (mobile) / 10,7 GB/s (desktop) -> QPI 25,6 GB/s. Intel doesn't specify the CPU <-> GPU speed, but I assume at least QPI speed. That's helpful for those transfers that don't use DMA (e.g. register setups).

Since the Westmere platform has to live some time I assume also the Memory <-> GPU interface has been improved. When intel designed the nehalem plattform it knew it would attach 4 and even 6 cores to the memory controller as well as a future IGP. I expect the new IGP to have a higher bandwith connection to the DRAM.

EU is the correct term instead of ECU.

When comparing Gen-4 vs. Gen-5 performance the results of anand with G35 vs G45 use DDR2-800. While when comparing GM965 vs. GM45 you have at least DDR2-667 vs. DDR2-800. That explains part of why G35/G45 are rather close in tests of anand whilst Gen-4 vs. Gen-5 shows greater difference in other reviews.

 

the new Wolfenstein somewhat works on the X3100.. i can play Singleplayer fine till i walk out the train, then i get the blue screen of death.. pretty solid framrate tho.. id say around 25 FPS... has a Config file to.. just like Quake Wars, so this is deff possible to get it to run good on the X3100, only problem is the blue screen.

but me personally.. i prefer Multiplayer.. but when you load Multiplayer it crashes once it loads "tools"

Hope someone finds a fix for this..

 

You also have to take into account that while memory is faster on the laptop platform, the IGP clock differences are minimal, 533MHz vs 500MHz. On the desktop the G45 operates at 800MHz compared to G35 which is at 667MHz.

See on the AT review it manages an average of 20% advantage: http://www.anandtech.com/showdoc.aspx?i=3417&p=4

With the problems people are having with badly optimized software solutions in general, I do believe the recent differences with drivers are rather a big part in getting the 4 series further ahead from the GMA 3000.

Are you sure about the not having DMA part?? It was a widely advertised feature of Intel IGPs. It kinda seems backwards not to feature it with recent IGPs.

 

I recognize this is very misleading: Certainly the IGP has DMA in the sense that it can load e.g. textures from main memory to the IGP EUs withouth requiring the CPU.

But I doubt the IGP can access the memory e.g. while some PCIe device is doing DMA or (and that's the better example) while the CPU is setting some registers of the IGP and the question is how long these other tasks take. The GMCH (memory controller hub) mediates between all functions and has a similar hub architecture as the ICHx. A hub or switch is always limited by it's internal bandwidth (attach 10x 1GB/s devices to a network switch and test whether it is able to serve 5x FullDuplex connections). The Nehalem architecture will allow to attach up to 6 cores to this memory controller hub inside the CPU. It has to have a high bandwidth. I assume the IGP to profit from this as well if it has ever been an issue or would become an issue with increasing EUs and clock speed.

This was just to point out that if there's a change in IGP performance by putting it next to the Nehalem/Westmere core, I expect the IGP to profit from this. Even though the memory controller is on the 32nm silicon and the IGP on the 45nm silicon. Before that both were on the same silicon.

But that's too far fetched. I'm looking forward for benchmarks. I just found something unrelated but funny in the intel mobile 965 chipset documentation. They even admit their IGP is bandwith limited p.79:

 

I also believe so too. Yea it makes sense.

Clarkdale 3.06GHz benchmarks(apparently the IGP runs at 700MHz): http://global.hkepc.com/3878

1600 in 3DMark06 is about the same as the 1944 score extrapolated to 1280x1024 resolution. The comparison with Nvidia is still disappointing, sigh. I can only hope the performance improves significantly with final hardware and drivers.

Preliminary G45 scores also used to be horrible: http://www.kbench.com/hardware/?cc=6&sc=3&no=56547&pg=4

Update: I want to see how 3DMark05 runs. Seriously, 3DMark06 and Vantage(even more!) are s****y benchmarks. They artificially inflate scores by including CPU scores(as if running the GPU portion isn't stressing the CPU at all...), which makes it a terrible GPU benchmark. I can't believe how retarded a group of well-paid developers can be

 

New Windows 7 drivers! Check whether black border bug still persists!

Win7 32-Bit v15.12.75.2.1867 dated 08/13/2009
Win7 64-Bit v15.12.75.2.64.1867 dated 08/13/2009

Code:

*	Intel (R) 945G Express Chipset
*	Intel (R) 945GZ Express Chipset
*	Intel (R) 946GZ Express Chipset
*	Intel (R) G965 Express Chipset
*	Intel (R) Q965 Express Chipset
*	Intel (R) Q963 Express Chipset
*	Intel (R) G33 Express Chipset
*	Intel (R) G35 Express Chipset
*	Intel (R) Q33 Express Chipset
*	Intel (R) Q35 Express Chipset
*	Intel (R) G31 Express Chipset
*	Intel (R) 945GC Express Chipset
*	Mobile Intel (R) 945GM Express Chipset
*	Mobile Intel (R) 945GME Express Chipset
*	Mobile Intel (R) 945GMS Express Chipset
*	Mobile Intel (R) 940GML Express Chipset
*	Mobile Intel (R) GM965 Express Chipset
*	Mobile Intel (R) GME965 Express Chipset
*	Mobile Intel (R) GLE960 Express Chipset
*	Mobile Intel (R) GL960 Express Chipset
*	Mobile Intel (R) 945GSE Express Chipset

No changelog due to drivers from desktop mainboard section.

I dislike 3DMark even more (any version). It is a tech demo for which nvidia and ATI optimize their drivers for. It is supposed to reflect how the game engine of tomorrow stresses your GPU and supposed to demonstrate how well your system is able to play those future games. If you take some older 3DMark it translates to how good your system plays todays games.

But this tech demo is simply very different from real game engines. At the point of time where it hits the market the tech demo has to run with the amount of GPU onboard memory that's common at that time, run the current technology stack (PhysX, AI, ...). Games can't use the 3D Mark tech demo, because it's performance would suck once you increase the mini scene to a world that's present in a real game. Future games will use different technology stacks and run on different hardware and different drivers. It tells how the performance of ATI/nVidia/intel compares running this tech demo. Real games might show different behavior and different performance ratios.

Intel also messes with 3dmark, since all three (3dmark03/05/06) run in SW TL, whilst HW TL is the default for most games.

To tell about the performance of intel IGP, I appreciate real games. E.g. two OpenGL ones, two DX9 and one DX10. And probably all of them have to have AF and AA turned off. The intel visual experience won't be as great as on a discrete GPU, but if the question you want to anser is "Can I play Anno 1404?" than that's the benchmark you need.

. I'd love to see gaming benchmarks over time how and whether at all performance evolves. I have only a german source for some ATI/nVidia driver comparisons.
. when arrandale/carkdale arrive you won't be able to play the most recent games and it probably won't compete against any discrete GPU. It might catch up to other vendors IGPs and enable to build very power efficient notebooks, which still allow the one or other game.

 

do win7 drivers work with vista?

 

The benchmark itself, at least 3DMark05 is a relevant benchmark with hardware T&L. I've tested and compared both hardware and software VS along with laptop users in NBR. Like Laptop users with Core 2 T5600 or w/e and X3100 would score close to me in 3DMark05 software mode(I use Core 2 E6600 with X3000. They'd get like 900 I get 1050) but on hardware VS, they drop to 700 or so while my system basically stayed the same.

Also, on software mode I noticed it doesn't care as much about memory bandwidth. On hardware mode where its relevant for games, the effect on memory bandwidth for performance is surprisingly large!

 

Yeah, without any problems. Windows 7 and Vista have same kernel, so drivers works "no problemo".

 

WOOOO! Newest drivers Raised my Win7 WEI score from 3.0 to 3.1. YESS! lol

 

I've been following this thread for awhile now and unfortunately there hasn't been much progress made with this gpu.

I commend all of you who are dedicated to get some decent performance with this chip.

I've tried xp, vista, and 7 with this gpu and haven't noticed an overwhelming difference between them. XP had the best performance followed by 7 and the bloated vista.

For those who are still getting the black bars using win7 drivers just go into the graphic properties then to display settings and change the aspect ratio from "maintain aspect ratio" to "full screen (no borders)"

Maybe someday Intel will work something out with Nvidia and finally be able to deliver more gpu performance with better drivers and hardware...

 

The new Wolfenstein runs perfectly OK in windows 7. No crashes at all, with around 30FPS. Check this : http://www.youtube.com/user/akash1988
Its my youtube channel for X3100. I'm gonna post the video of Wolfenstein by tomorrow.

 

I can play Wolfenstein on Windows 7 (X3100) only with XP drivers. When i have installed drivers for Vista or 7, game crashes.

 

how come when i use windows 7 when playing game it is nt in full screen? is it a bug?

 

Queue does not work in VMR9-YV12 on this chip. I don't know for some reason does not fully support DirectX 9 under XP. I have an older version of their driver that does just fine on the YV12 output, I think it was ver 8.12 something, the one that Sony recommends on their support site for Intel video chips. However this version being older lacks support for other things and has its own flaws.

 

Do you mean the resolution isn't stretched to the size of your display?
If that's the case, you need to install Intel's drivers, so you get the Graphics Tray which lets you set stretching.

 

ya... i installed the driver from intel website stil remains the same...

 

New round for Win7 X4500 drivers:

Win7 32-Bit v15.15.4.1872 dated 08/20/2009 ( Changelog)
Win7 64-Bit v15.15.4.64.1872 dated 08/20/2009 ( Changelog)

As it has been with the v15.15.[0-3] drivers as well, those are useless for X3100 (GM965) users except for modding purpose. Only useful for

Code:

*	Intel(R) G41 Express Chipset
*	Intel(R) G43 Express Chipset
*	Intel(R) G45 Express Chipset 
*	Intel(R) Q43 Express Chipset 
*	Intel(R) Q45 Express Chipset 
*	Mobile Intel(R) GL40 Express Chipset
*	Mobile Intel(R) GS40 Express Chipset
*	Mobile Intel(R) GM45 Express Chipset
*	Mobile Intel(R) GS45 Express Chipset

 

@7oby, can you give a step by step guide on how to mod the X4500 drivers to work with the X3100?

Would be much appreciated

 

Unfortunately not, since I'm not modding these drivers. There are two approaches: Either exchange particular .dll's of the X3100 drivers with those of the X4500 drivers or modify the .inf file to add X3100 support prior to installation as described here:
http://forum.notebookreview.com/showthread.php?p=5020848#post5020848

Although I can't help you, others might be able to help if you post your specific problem. Modding drivers might not be very stable and some features might be missing.

 

I tsted the new Win 7 drivers and got 444 poitnts on 3d mark 06, with the old XP drivers i only got 394 points. Little bit better now

 

You tested it on XP?

 

what do these GFX drivers do? is it merely updates from Intel or modded drivers? and why the GFX prefix?

 

Enable graphics features such as Aero, OpenGL, DirectX 10.x on intel integrated graphics.

Updates from intel.

Because they are taken from the desktop section such as this download. intel desktop mainboard section downloads are often more recent than chipset specific (e.g. GM965) downloads. They are bit to bit the same as the .zip version once unpacked and once the .zip version appears in the chipset specific download section. Desktop mainboard section contains only .exe drivers and often no changelog.

--

For the interested ones: Read page 3-6 to see how GPA enabled to boost performance of the Demigod game from 14fps -> 24fps on GM45 hardware:
http://isdlibrary.intel-dispatch.com/vc/2667/demigod_rev10.pdf

 

Is any way to use GPA with X3100? From 14 to 24 fps is very good, wow..