Diego !!!!
This means, you guys have 64 bit drivers! Already exists! Proof has been made.
Now, yes it was buggy but you guys had to already use it as a base line template code to produce the 64 bit drivers for the graphics core in the newer Intel Atom z whatever which is PowerVR architecture. The exact VERY SAME architecture that is the GMA 3650. In fact, the GMA 3650 is mere a faster and slightly better HARDWARE wise in that it has a higher clock rate and a superior overall technical specifications.
So, why is it that Intel doesn't get the _____ done!
First off.... basic computer science 101 for you, all drivers for a video chip boils down to communicating to machine level and transmitting values to the graphics chip (in this case, integrated onto the same die as the cpu but it is not the cpu, EITHER !!!!) registers. When you write drivers on a multitask OS, you go through APIs which all boils down to ultimately to the registers of the video chip. Since we are talking about a 'GPU' then we are talking about something kind of like an ADVANCED form of the TMS9918 Video Display Processor - an early pioneer of 'Graphic Processors'. Unlike other video display chips of its time and even for several years since its introduction, it was a 'PROCESSOR' vs. simply a memory mapped register switch 'table'. You actually had a 'graphic processor machine language' much like a microprocessor. After all, the TMS 9918 was essentially a microprocessor like the TMS9900 microprocessor with which the 9918 had special registers and instructions and functionality particularly designed and built for graphics. Fast forward some 30+ years and today's graphics are in themselves actual 'processors' and not just a bunch of registers that can be given values and do something. There is an actual graphic processor 'machine language' to these graphic chips. From my understanding, the PowerVR based graphics processor is speaking the same 'machine language' in current intel atom processors and the GMA3650 graphic processor integrated in the Intel D2700. Besides memory mapped registers and GPU processing languages are in most part the same. The drivers should almost work right out of the box for any PowerVR based GPUs.
FWIW: WDDM is simply an abstraction layer. You guys already made WDDM 1.2 compliant drivers for current PowerVR-based graphics. So, what's the deal? Why can't it work? Come on? Your very source code for graphics drivers for Intel z2760 is already 95-99% correct for making the D2700 work. You just need to relabel some IDs so as to match the D2700 'id' and maybe a few modest changes and timing adjustments to deal with the clock rate difference of the D2700's GMA3650 and the z2760.
You may also have a few register memory mapped location differences but come on. Easy to make the changes for.... for a software programmer.
REALLY... what's the barrier. After all, you have 32-bit drivers for Windows 7... oh.. wait.. that's WDDM !!!! WDDM 1.1. Seriously, we all know that there is video cards with WDDM 1.2 compliant drivers that don't make any more uses of features already available and supported in WDDM 1.1.
No more excuses from you guys. Get the drivers ready by Christmas holiday season. It shouldn't take more than maybe 1-2 months to actually get it done and working.
You already have a good working code base. Turn off driver features present in z2760 that isn't present in the D2700. Also, there is already good working CPU processor driver code for Windows 8 for the actual Intel Atom D2700 microprocessor core itself. It is the graphic drivers component. The only other features that I can think of in the z2760 that isn't in the d2700 is irrelevant to graphics itself. I see no practical difference in the GPU for any reason that would prevent it from working on Windows 8. In fact, you are using the SAME graphic processor in all essence and that the d2700's gpu is really a faster version clocked at 640MHz (IIRC) compared to 533 MHz.
Other than a little bit faster clock rate, there is no reason that you guys can't make the graphics drivers for the z2760 work on the d2700 atom. Binarily, they are basically the same processor just like using the drivers codes for programming a TMS9918 would drive the V9958 and driver code to drive the v9958 would largely drive the TMS9918 where they are the same. In the case of the TMS9918/V9958... there is more degree of difference than the d2700's GMA3650 and the PowerVR used in the z2760. Seriously, lets quit the b.s. story and jerking us around. Tell this and forward my message to the powers at be. It should be a simple task.
It would have likely cost less money to have made the drivers than all this time dealing with all these threads.
It is insanely silly that Intel... a multi-BILLION dollar corporation couldn't spend the minute amount of money to make the driver. It would probably cost less than $250,000... less than what it cost to buy a modest house. I seriously would think it would be easily done in less than 1/3 of the price. I doubt more than $50K would even need to be spent.
Either provide the very exact technical reason why it can not be done.... I don't mean some executive jerkoff as the reason. If it is some management/executive that is the problem... give the jerk the pink sheet. If there is no technical reason to reasonably do it then it get it done or provide me the following:
1. Source code to drivers for d2700 and pertinent driver source code for z2760
2. All documents necessary to perform the work.
3. Authorization to distribute the information to parties necessary to organize a driver set development team
4. Access to direct communication with CederTrail & CloverTrail devs including software driver devs for CedarTrail and also CloverTrail so we can resolve the matters of the drivers for the D2700 and possibly a number of other drivers. The Intel team would need to be involved for Quality Assurance testing of the drivers and overall guidance because of their involvement and knowledge of the matters.
Considering it would be a 'derivative work', it would also be CLEARLY understood that the proprietary rights for the source code and new drivers would ultimately belong to Intel.
It would be nice if we get some working cooperation or Intel gets the drivers made.
I can not POSSIBLY find any real hardware reason why d2700's GMA3650 would not work in Windows 8 64-Bit. The cpu was a true 64-bit 'x86-64' cpu after all. Unlike most other Atom's INCLUDING the z2760. It runs Windows 8 Pro - 64 BIT EDITION.
The GMA 3650 should work fine. After all.... z2760 uses a PowerVR SGX545 core. The GMA 3650 is a PowerVR SGX545 based gpu core. What's the difference?
Intel staff.... don't waste my time or anyone's time with pre-canned statements like 'we'll forward it appropriate department for future consideration' or 'We thank you for your feedback, we are aware of the situation...' or pointing us to a particular thread link about Intel official support... blah blah blah OR anything similar to the above. I want serious response from the techs and those with decision making authority. I don't want circle-jerking or anything like that.... yeah... placed in the circular file for future consideration as if anyone ever really looks into the circular file for future consideration. Don't b.s. me with lies. We KNOW better than that. No one looks to the trash can for things of future consideration so seriously... I only know that someone with the authority to make decisions OR part of the technical department if they personally respond specifically to the comment.
I would want these people to stop hiding behind human shields like a bunch of wusses and actually are man enough to face the customers.
联系客服
