Nic, it would be useful to know if the performance increases are supposed to be all round, or whether we're talking about fixes and optimizations for specific applications/games. If there is supposed to be an all round "significant performance improvement" then unfortunately I haven't seen it. As this is just a point release (originally for an OEM) I'd assumed it was bug fixes rather than any real performance boost.
Thanks for the feedback Mark.
Release notes are here; http://downloadmirror.intel.com/22479/eng/ReleaseNotes_GFX_1089_32bit.pdf
They look like bug fixes mostly, but a lot of people were looking for those. I'll make an amendment to the original post, stating that we also want feedback on any pre-existing issues that were not resolved with the driver update.
Using driver version 18.104.22.1689, I am experiencing graphic artifacts glitching on certain Flash-based games, specifically on GlobalPoker from Facebook. The glitch will appear at random times, but would disappear when the game redraws the graphic on the affected area. If memory serves me right, I have not observed such behavior with driver version 22.214.171.1245.
Also, setting the Power Plans from the Intel Graphics and Media Control Panel doesn't persist upon reboot. Graphic Power Plan Settings for both Plugged in and On battery default to Balance, but if I set On battery plan to Maximum Battery Life, then restart the computer, the setting reverts to default. Is this an intended behavior?
They look like bug fixes mostly, but a lot of people were looking for those.
A lot of people were also looking for some DX10.1 support also, which is what the GPU is technically capable of
But anyway, it's the best driver released yet, shame it's merely papering over a few cracks in the wall, rather than fixing the roof...
So...why is a 3 month old driver just now being released?
Anyway, this driver is a big improvement, but not big enough. Here's what I've noticed:
*Significantly improved DirectX performance in many apps, but not all. Tribes Vengeance, Doom 3, Warcraft 3, etc. are all noticeably faster, but the latter 2 are still unsatisfactory.
*Less OpenGL glitching. TORCS is now playable without all the flashing junk polygons all over the screen. Glest no longer crashes when clicking on a unit. Minecraft is now glitch-free (albeit at 15 fps).
*Unreal Engine 1 games can now use the original DirectX renderer without junk polygons, but at very low framerates. A GMA 950 netbook can manage a consistent 60 fps, and we're stuck with 12? Why?
What's worse since 1065:
*Midtown Madness 2 framerates are oh-so-awful. Running without hardware T&L makes it run faster, but introduces many glitches (morphing polygons/terrain, textureless trees).
*What's still broken:
*Games without hardware T&L support suffer from low framerates and morphing polygons
*AssaultCube and AssaultCube Reloaded still refuse to start
*Battlecraft 1942, the map editor for Battlefield 1942, crashes on start. The cause? Intel's OpenGL dll.
*2D acceleration. I shouldn't have to use Microsoft's default driver to get from 2 fps to ~14 fps in Cave Story.
*All Sims and anything they interact with are invisible in The Sims Complete Collection. Microsoft's default driver can run it just fine.
*TAGAP 1 & 2's character images are very distorted and don't look at all like penguins.
*Overall poor performance
I'm noticing an overall improvement in performance, especially scrolling and opening things quicker too. Not sure if it's to do with B-3 stepping support in the drivers but all of the bugfixes and performance issues are welcome. Thank You Intel. Hopefully you will continue to improve the Drivers, but there are still some issues that I how you will fix. Thank You!
I have been testing this driver for a month now and here is what I've found with the 1089 driver (Intel GMA 3600, Windows 7 Home Premium 32-bits) :
- sluggish aero experience under Windows 7, even when no other process/program is using the IGP (or putting a noticeable burden on it)
- sluggish video playback (seen with Media Player Classic) with Aero enabled
- in the Intel GMA control panel, the Power Plan setting won't persist after reboot, especially when the performance setting is chosen on power plug
- sluggish performance with games using Unreal Engine 1 (e.g. Adventure Pinball)
- SWAT 4 crashes during launch
- Quake 3 crashes during launch
That is all I can remember for now, I will post again should I notice other bugs.
In the meantime, can anyone tell if a new driver release is to be expected in the next days/months ?
Thanks for your help, best regards,
Sorry for my English. I use a translator.
Intel, I was hoping to buy a netbook with graphics from Intel, but you deceived all tucked into it some graphics PowerVR SG545. This fake is not capable of anything:
slow interface Windows 7,
slow scrolling in browsers,
inhibit the ancient Games, published 10 years ago,
not declared support DirectX10.1,
no support for Windows XP, where performance interface would be better,
no support for Linux, where the performance would be very fast
Your previous graphics card GMA 3150 just mega performance compared with that subquality.
Now I do not trust Intel.
I've experienced much of the same. Fortunately, I've been able to find a few fixes. For Unreal Engine 1 games, alternate renderers, especially OpenGL ones, have boosted my framerates from 15 to 75 in Unreal Tournament(yes, I'm exaggerating a bit). As for Quake 3, it'll crash if you used 1024x768 or 1152x864 (with the Display1_DownScalingSupported hack) before you updated. The registry value resets with every driver update, so you'll need to re-enable it. If that's not the cause...I don't know what to tell you. My Q3 performance is largely unaffected, but it was satisfactory in the first place.
Thanks Kory ! I've applied your fix and now Quake 3 runs fluently without any noticeable problem. Still, an under the hood fix for running such an old game feels somehow odd...
Scrolling in browsers can feel slow sometimes especially with heavy websites (typically when they use flash, so a lot of them actually).
Sluggish Linux drivers is also an established fact, but I would rather have a functional windows driver than a linux one if I had to choose.