MIX'10: Internet Explorer 9 Platform Preview Info

Why does MS stick to the long development cycles for IE? I'm not saying Firefox is better, but at least they release a new version every six months or so. IE8 has been out for a year and we're just now getting an alpha version of IE9. In the fast-changing world of web technologies, how can they afford to take so much time between major releases?

"I'm not saying Firefox is better, but at least they release a new version every six months or so. IE8 has been out for a year and we're just now getting an alpha version of IE9. In the fast-changing world of web technologies, how can they afford to take so much time between major releases?" There's a huge difference between consumer-targeting companies like Google and Apple that want to differentiate by offering bleeding-edge technologies to the ADHD tech crowd before they become a standard, and Microsoft, who wants to give customers a stable platform to use for a while without having to rip everything out in a year's time - because that's what they ask for. Even software publishers want to sit back for a bit after doing a release to cut back on active R&D costs (and build up revenue from support) before customers start clamoring for the latest and greatest. Software publishers do like to have a set release schedule for major releases (excluding minor bug fixes) just as much as hardware companies. Software is often built around new hardware releases.

Meanwhile, many companies still can't get off IE 6.

No IE9 isn't going to use Webkit

This post was mentioned on Twitter by ICTRacunalo: [News] Internet Explorer 9 Platform Preview Info http://bit.ly/bQAu85

@redu: From what I've read, the 785G only supports a 3200/3300 GPU. Anything above that, and the onboard GPU is disabled automatically by the drivers. None of the 4000 series cards (like the 4330) are supported in Hybrid Crossfire with a 785 chipset. Similarly, the 790GX chipset has a 3300 onboard, and I think Hybrid Crossfire for that only supports 2400's. I dunno if any of this is similar with the new 890GX chipset boards though, but the only documentation that mentions Hybrid CrossfireX for those ones mentions the 5450 only. You can put a 5450 in and it'll support Hybrid CrossfireX, but I also want to know if you can put 2 of them in to an 890GX motherboard, since they all 890GX motherboards support 2 discrete cards. The 5450 GPU's also support standard CrossfireX, so you can use 2 discrete 5450's together. However, I'd like to know if the 4290 stays on in such a configuration. There really shouldn't be any reason why it shouldn't be on, since the 5450 parts run at approximately the same speed as the 4290 (except that the 5450 is a DX11 part), so having all 3 GPU's enabled would make a lot of sense. There's been a lot of issues regarding using mismatched GPU's with Crossfire though, so I wouldn't be surprised if AMD just allowed 1 discrete GPU with the onboard video (it may be a limitation of the number of PCI-e lanes too - where there aren't enough to run the onboard with 2 discrete cards). They always talked about CrossfireX supporting mismatched GPU's, but they still have to be within 1 or 2 GPU families for it to work. They cite synchronization issues as being the reason for the strict requirements, but I think they need to work on doing better load balancing. The drivers should be able to divide the load up between the different GPU's based on a relative scale of performance of each GPU from the whole. They should know the performance factors of one GPU over another based on the number of shaders, RAM amount and speed, etc., more than anybody since it's their hardware. It shouldn't be so hard as to say "if a 5770 is 3x the performance of the onboard 4290, serve 3 frames worth of information to the 5770 for every 1 frame served to the 4290". That might be a gross oversimplication, but I'm sure there's some way to readily buffer data between cards. Perhaps that's something that would be best suited to using a dedicated Hypertransport bus between cards, bypassing the PCI-e bus altogether. Maybe a ribbon cable between cards like NVIDIA uses for syncing SLI cards? If you could merge the memory between cards using a high-speed interconnect, both discrete cards could be working from a singular frame buffer and texture pool. Once that's accomplished, the drivers could get the GPU's communicating with the merged frame buffer in an asynchronous method based on each's relative performance of one another. The lower card would fetch data only a percentage of the time of the faster card. So why is that so hard?