The well-known graphics driver development company, SciTech Software, are closing on releasing a beta of their Linux port of their SciTech SNAP Graphics for Linux. These drivers replace the 2D XFree86 modules for a large number of graphics chipsets. SciTech claims that their latest beta drivers have shown up to 12-15% faster 2D performance than their XFree86 counterparts (for some chipsets). In the initial final release, accelerated 3D via DRI won’t be supported though.
SciTech Upcoming 2D Linux Drivers Promise Speed Enhancements
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Xandros has already bundled crossover office, I can definatly see them including this as their driver base for X.
CrossOver is a needed addon for the OS’ functionality to get more users. SciTech drivers are not needed. They are just an alternative to X11’s 2D drivers. Especially when the Scitech drivers don’t support the DRI 3D drivers.
There’s no point in exchanging a working setup unless you get acceleration for RENDER which is basically unaccelerated on most chipsets today, except with the proprietary NVIDIA drivers & hardware. Have you noticed lately how slow scrolling under mozilla can be with subpixel antialiasing & Xft font rendering?
Actually believe it or not, performance is not the big issue and focus of this initial product release. We are matching and exceeding the performance of many of the existing XFree86 drivers, but we are not done with performance optimisations yet. Rather we have been initially focusing on compatibility across all versions of XFree86 and as many Linux distributions as we can. SciTech SNAP Graphics for Linux, unlike any other existing XFree86 driver solution, will work with any installed XFree86 4.x distribution from version 4.0.2 right up to the latest 4.3.0 release. We do not yet support XFree86 4.0.1, primarily because that version was not widely used and will require a lot more porting to support it. If there is demand for it though, we will consider porting our driver module to 4.0.1 as well.
Note also that since it includes support for all the latest graphics hardware such as the Intel Springdale chipsets, ATI Radeon 9700 etc, this is the best way for Linux users to get support for the latest hardware without needing to download and install a complete XFree86 distribution for their operating system.
As for the RENDER extension, this is naturally supported on XFree86 installations that have it enabled, but we are not yet accelerating that extension. Once we are done with the compatibility work we are doing, RENDER is one of the first things we will be accelerating on as many chipsets as possible.
On OS/2(eCS), it’s great! True plug and play. You won’t regret it.
I will site Mr. Benett’s comment as support to say that they are working to get it working properly THEN they can build in the cool features that we all wish wre in the drivers in XF86.
this is the best way for Linux users to get support for the latest hardware without needing to download and install a complete XFree86 distribution for their operating system.
A good point for certain, as many Linux users are not running the most current versions of Linux (for various reason) but still would like to have easy access to the latest HW. Speaking to the performance issue, rest assured we have been paying carefull attention to this as well, and to date like what we see;)
http://www.scitechsoft.com/snap_test_results/snap_test_home.html
What license will they be released under?
SciTech is a commercial company. They don’t work for nothing, they have families. So, don’t expect it to be open source, neither plug your little flamewar over here. OSNews is equally receptive on both OSS and closed source software. Each one has its place in this universe.
I suspect it will be made available under the famous USD (US Dollar) license you know the one that actually pays a company for supplying a worthy product or service
However you can never tell with these SciTech folks after all I see they allready support several GPL projects and the *Open* Watcom project.
Cheers and I look forward to trying SciTech SNAP Graphics for Linux!
From what I understand, SciTech have signed quite a number of NDA’s to get this product out the door, so that should tell you that this isn’t going to be source released.
I only await the day they get 3D sorted out and added to FreeBSD and OS/2.
Simply inquiring about the license of a product is not “starting a flamewar.” Just look at what he wrote. How could it possibly have been written in a less inflammatory way? Your response, however, certainly does look like a flame to me. In fact, if his question would have been answered with a simple statement like “It will be proprietary”, then nobody else (including me) would have felt it necessary to chime in with our own opinions. The poor guy just asked a simple question, and a rather important and relevant question IMHO. If it’s a product to be used in a mainly open-source environment, a lot of people will care about the license. It is not neccessary to flame him or to assume he is trolling.
Just my opinion, feel free to mod it down.
I’m a SNAP beta tester, unfortunately I am unable to test as my hardware is not supported. It also seems that OSNews can’t seem to figure out when the moderate button should not be used. I have a valid complaint, about a company that is refusing to provide necessary documentation to allow SciTech to build drivers.
Using capitals and an attitude, won’t make ATi release info to SciTech, Aitvo. Whatever you had to say, you should have said it better.
I might be interesting in buying this for my laptop..xf4.3 completely broke the savage driver on the vaio..i have hackishly compiled savage.c from xf4.2 with the xf4.3 tree to get my vaio to work with x.
If these guys support Xv and apm and stuff well..i’d be happy to pay for it..if they do 3d support..well, that’s be a nice bonus..even if its weakass 3d..What’s the upgrade policy gonna be? are the regonna be free upgrades?
Really, they sure did an about face when they checked IGP code into XFree’s CVS. You think it was because they wanted to? I don’t. I said it the way it had to be said. Remember when companies like Microsoft wouldn’t move on patches until they were forced to? It’s a similar situation.
SciTech has made drivers for X before, with some cards supporting accelerated OGL in X. i wouldnt be surprised if they put 3d in it as well. I dont see this ever being released as open source, though I dont see a reason for it. These guys put out quality products since the dos days. This is one of the few products that can be viable commercially in linux (people will pay for it). In my case though, I have a geforce3, so i doubt i’ll have a need for this.
keep up the good work scitech dudes
I wonder what the price and license will be too.
I doubt too many people will buy it if it is like Metro-X and costs double the price of the OS it is installed on.
Also, my main machine is running a GeForce 4 Ti 4400, so I’m out of luck there. I have good 2D/3D drivers for X anyway, but I trust SciTech products based on prior good work and I really like the concepts they are working on.
It’s a shame Nvidia won’t cooperate. It would help everyone in the long run.
I might upgrade to an ATI if this works out well and has good 3D. Nvidia, are you listening?
So basically, you think they should restrict the cost to $0.00?
what would be the benefit of using these drivers over the nvidia ones? especially now that the nvidia installation is very much simpler too.
Advantage? How bout never having to touch the video portion of XF86Config? Or being able to switch resolutions on the fly properly? SciTech SNAP drivers are really nicely done. I’m VERY happy that they’re making their drivers work in linux, and if they ever make them work for linux/alpha, i’ll surely support them.
Completely OT, but I have to throw in an appreciative comment to Eugenia, who always manages to smackdown the zealots (on either side) to ensure that debate is open, mature, and inclusive of all reasonable perspectives.
That is all.
/me returns to lurking.
“Or being able to switch resolutions on the fly properly?”
Gee, that’s funny. Seems I’m able to do that just fine with XFree 4.3.x w/ the latest NVidia drivers. What planet are you on?
I’m well aware of that, but its not possible to do it for every card out there that can run X. I’m also aware that there is a module that helps with all that. Not everyone has an nVidia card. I run an ATi Radeon 7000, and XF4.1, can i do this? Nope. I dont run XF4.3 because it has been HORRIBLY unstable for me, nor do i run binary only drivers because i dont run an x86 but a DEC Alpha.
Sounds good. I’m all for open source but a faster X sounds promising… Will there at least be a demo to try first to make sure it’s all it’s cracked up to be? How much will this cost? I’ll pay if it’s not too expensive.
What if I pay and get the first drivers that don’t accelerate render? Will I have to pay again to get newer drivers?
Increasingly, I am more and more interested in projects such as DirectFB, and I have also had good results running my SDL apps on the ‘standard’ Linux framebuffer, outside of X entirely.
Sometimes I wonder if X-Windows is the wrong place to implement accelerated drivers, and whether XFree86 should simply run on an accelerated framebuffer e.g. XDirectFB, or the project that uses OpenGL to emulate a framebuffer driver on which X runs.
XDirectFB provides support for proper double-buffered windows and transparent windows using existing XFree86 code, (it is a patch to build a framebuffer XFree86 Server on top of DirectFB) and it really only held back from being a viable solution by lack of accelerated drivers for more video cards. – why is it a problem for XFree86 itself to implement this stuff?
If graphics were implemented on Linux like this, then it would be easier to produce alternatives to XFree86 that worked without having to reinvent the driver wheel each time.
Gaming, multimedia, embedded applications and general GUI ‘innovation’ would be so much easier if the X protocol was not the only reliable way to get graphics on the screen of a Linux box.
Obviously X has been around longer than the Linux framebuffer, and the future of the framebuffer API is a little murky, but I believe it will be rather important in future to unify the video driver layers.
This is somewhat off-topic, and it is good to see companies providing products to add real value to Open Source core components.
> Gaming, multimedia, embedded applications and general GUI >’innovation’ would be so much easier if the X protocol was >not the only reliable way to get graphics on the screen of a >Linux box.
Why ? Do you have anything _at all_ to back this up?
For embedded devices, sure. For linux running the desktop of an everyday user no. X is fine, noone has proved otherwise.
And this shows that XFree86 can be done better. Much better, so please stop whining about “X is is bad/old”. Its good enough, and can be done _much_ better by improving toolkits/drivers.
Some of us don’t give a flying duck about Linux. What about supporting the BSD’s?
The SNAP drivers will probably be closed source and require payment. But if the price is comparable to the OS/2 drivers, I think $40 is well worth it. I’ve been able to upgrade my video card on my OS/2 machine twice without ever upgrading or changing drivers. Simply remove the card, insert new card, boot up and done.
This is something I’ve always thought XFree86 & Linux needed. Windows always felt snappier when opening windows.
Great job SciTech!
Not open source?
What’s the point?
I can upgrade XFree86 in less than 5 minutes on an old distribution to get it working with the latest hardware. I never needed SciTech’s drivers before and I don’t think I ever will.
too little too late.
I have a valid complaint, about a company that is refusing to provide necessary documentation to allow SciTech to build drivers.
ATI has actually been one of the easiest companies to work with IMHO. We are regularly supplied with early examples of their HW and also have access to key dev support to boot. Now if were speaking about nVidia, well thats another story all together…
Will there at least be a demo to try first to make sure it’s all it’s cracked up to be?
SciTech has always provided free 21-day, fully functional trial version of its products. This allows you to evaluate them on your own machine prior to being asked to part with your hard earned $ and we will continue to do this with SciTech SNAP Graphics for Linux – our simple belief is that if it works as expected then the value is self evident if not then…
Did they finally provide IGP drivers? I was under the impression that the IGP department wasn’t responding to requests.
Did they finally provide IGP drivers?
ATI does not supply us with drivers for any product (as that might prove to be some what redundant) they do however provide the required support that allows us to more easily develop our own accelerated drivers.
Speaking directly to your IGP issue, most of our focus has been in the add-in-card area and not on the integrated side of things (in relation to ATI) as this is relatively a new business unit at ATI. I can tell you that literally no more than 5minutes ago I spoke to ATI about this very project – thanks for the reminder
.
The following Distributions have passed SciTech certification and will be fully supported in the final release of SciTech SNAP Graphics for Linux.
Red Hat Linux (7.1, 7.2, 7.3, 8.0, 9.0)
SuSe Linux (8.1)
Mandrake (9.0, 9.1)
Turbo Linux (7.0)
Slackware (8.1, 9.0)
Debian Linux (3.0r1)
Xandros (1.0)
I ment specs, not drivers. My bad.
Good, because unlike whatever ignorant people are always saying, the main bottleneck in X is *not* the network transparency/protocol/whatever, but the driver.
So dig out some cash, pay them, get new drivers, enjoy the speed, and stop spreading your ignorant rants.
Being that you had to sign NDA’s is there ANY CHANCE that SCI-TECH will be releasing a GPL version of SNaP? With say–only OPEN SOURCE drivers available?
(For free as in beer stolen from a party at some sophomore college girl’s apartment–because you’re too broke from Student loans and you just maxed out your unsecured credit card to pay for your Calculus & Chemistry & Physics books–and the part time job as a fry cook at the Wendy’s in the student center just isn’t cutting it–but beer rules so you took it anyway–and stolen beer still tastes better than a 40oz of Olde English 800?)
SciTech SNAP has many different components to it – and as such portions may be opened up under various licenses. As for the individual driver binaries I would have to say that the chances are slim that we will ever be in a position to provide all 200(currently supported) drivers under a GPL license.
With that said we have seen some encouraging signs form HW vendor that give me hope that we will be able to open up at least some of the drivers.
hmm, make that “ol’8ball” a Sierra Nevada and serve with Cheerios and you have the makings of a well balanced breakfast – cheers
“Increasingly, I am more and more interested in projects such as DirectFB, and I have also had good results running my SDL apps on the ‘standard’ Linux framebuffer, outside of X entirely.”
If you like DirectFB, then you will really like SciTech SNAP. The SciTech SNAP drivers are not actually XFree86 specific at all, but rather the XFree86 driver is just a small module that interfaces to SciTech SNAP. Once SciTech SNAP is installed on the system, developers can actually code directly to the SciTech SNAP API’s via the SDK, and bypass X entirely. In fact SciTech SNAP drivers have been used by MGL developers on Linux for about 3 years now, entirely without X. X is just the most common component and what most people are interested in.
Also another thing worth noting is that the SciTech SNAP drivers can be used to provide device support for any other graphics API on linux, such as Qt/Embedded, SDL, DirectFB, SVGALib or GGI (as well as the SciTech MGL of course). Right now we only have support for native SNAP apps, SciTech MGL and X11, but we have an almost complete Qt/Embedded module and would like to implement modules for SDL and DirectFB too.
“The SciTech SNAP drivers are not actually XFree86 specific at all, but rather the XFree86 driver is just a small module that interfaces to SciTech SNAP.”
Any chance you can push for this “interface” to become an open standard on Linux/FreeBSD?
It would be great for the whole community to move away from X11 only graphics drivers. If this was an open promoted standard then the open source community could make use of it too.
http://www.scitechsoft.com/products/enterprise/snap_linux.html
As for making SciTech SNAP a standard well that takes a major buy-in from the community (as it should). If you are interested and willing to spread this idea we would be happy to do our part in making it a reality.
Well, right now, with the oh-so-crufty XFree86 drivers I can actually download a tarball of binary drivers (like the ati drivers from GATOS) built on Linux machines, untar it into /usr/X11R6 on FreeBSD, restart X, works. And later I’ll put them on my Solaris/x86, too.
How’s that for compatibility and easy installation?
Of course it’s a rather moot point if chipset vendors won’t play nice with open source driver programmers.
Do you only support the version of Xfree shipped with the various distribution (for example Xfree4.2 with suse 8.2) or do you also support updated xfree versions on “old” distributions (like xfree 4.3 for suse 8.1 from http://ftp.suse.com)?
We support any version of XFree86 (4.0.2 through 4.3.0), whether it came with the distribution or not. The standard tarballs are actually less likely to have problems, since the distros haven’t made their little tweaks to them…
I was not asking for a standard for X11, but rather an independant graphic driver standard for Linux/FreeBSD etc.
So simply making SciTech SNAP a standard is *not* what I meant. This can be *one* implementation which uses a standard driver plugin interface.
Simply downloading XFree86 drivers and copying them into a directory is *not* what I meant. That will only work for X11. What about other graphics architectures like DirectFB and the like? If the graphics driver was an independant standard from XFree86 then we XFree86, DirectFB, individual games etc. could use the drivers as they saw fit. Not everybody wants XFree86.
Ooop’s wrong URL.
http://www.scitechsoft.com/products/embedded/sdk_download/snap_sdk_…