English | 中文 | Русский | Français

2,598 Posts served

8,348 Conversations started

Welcome to G45! Better (but still imperfect...)

By Aaron Brezenski (Intel) (5 posts) on August 14, 2008 at 11:34 am

The first couple weeks of motherboard availability on Intel's latest and greatest integrated graphics chipset have been tumultous.

First, our competition threw together a demo booth which stated baldly that HP laptops with G(M)45 did not accelerate Blu-ray at all while theirs, naturally, did. Of course, the Intel system was running Vista AeroGlass while the competitive system was not, and no specifics about which memory or CPUs each was using were given. It was clearly not an apples-to-apples comparison... but the fact that the Intel CPU was pegged at 100% is just as clearly an indication hardware acceleration was not working. I'll not try to puzzle out why (the competition is of course going to try to put themselves in the best light possible), but it was unfortunate FUD.

Next, Extremetech did a very unflattering review on the Intel DG45ID board. While most of their criticisms concentrated on gaming, they threw in a paragraph about Blu-ray playback as well. I'll be honest: I skipped everything regarding gaming; it's not and never will be my focus. But they brought up a couple of troublesome issues: they, too, did not see hardware accleration on G45 Blu-ray playback. They updated this a couple of days later with a correction (presumably someone from Intel got involved to tell them about the proper BIOS settings) showing the acceleration working now, but a less-than-stellar benchmark (the only one they ran) remains: on the HD HQV test, Intel scored a paltry 30 out of 100.

While I question the value of some of the HD HQV tests when evaluating Blu-ray (a topic for another time), they are in fact valid tests. My sources (I do have some-- really!) tell me that Intel's less than stellar scores are due to a player software issue: properly configured advanced de-interlacing will result in scores 20+ points higher. Still not perfect scores, but coming within the realm of workable, and my hope is that subsequent driver tweaks will improve this even further.

Finally, G45 boards started hitting the Audio/Video Science Forum, where the real home theater geeks hang out. There it's been, as usual, a mixed bag.

The Good

On the plus side, use of the Arcsoft Total Media Theater player yielded immediate results. Blu-ray CPU utilization on a Core 2 Duo was lower than 20% on all material: a clear indication that hardware acceleration is working properly, as anyone who's tried to run Blu-ray on an Intel G(M)965, G33, or G35 could tell you. PowerDVD results came in similar, if a bit less efficient-- that's not uncommon for the difference between these two pieces of software in general. The image was clean and 7.1 channel LPCM audio worked over HDMI alongside standard Dolby Digital and DTS. I give kudos to the folks in Design, and in the software, validation, ISV, and Marketing teams who made this happen.

Celestial choirs are singing, dogs and cats holding hands and smiling at one another; there is peace on earth.

Except.

From the enthusiast perspective, nothing's been fixed.

The two biggest issues on G965 and G35 which prevented widescale adoption among enthusiasts when we had the advantage of HDMI 7.1 audio and no one else did, are still there.

The Bad

G965 and G35 suffered from stuttering every 15 seconds or so at 24 Hz display refresh rate on Blu-ray before support for that refresh over HDMI was unceremoniously pulled in the 15.8 drivers. Some information implied (at the time) this may have been an issue with the SDVO chips used at the time for HDMI with Intel graphics, but the fact that the issue is still there on G45 tells us that this was not an accurate assessment. More and more TVs are supporting this refresh rate and even using it as advertising copy in their features list. But currently on G45, 24 Hz refresh Blu-ray playback, which is of interest to those who want the best video quality, is still broken.

The Ugly

Next-- and I would assert more importantly-- trying to send HDMI signals directly to an HDCP TV works, but repeater mode-- required if you want to get 7.1 sound or even just to play Blu-ray through a receiver-- is still broken. I understand the reason: the standard COPP protocols that all of these players use do not support the use of repeaters (even though our competition appears to be ignoring these strictures), and the software players don't yet use OPM (which does support repeaters). This is all very well and good, and we're probably in the right on this regarding supporting the spec: but the end user doesn't see this. They see our competition's graphics working flawlessly with their software player on their shiny new Onkyo or Yamaha receiver, and Intel's failing. The successful 7.1 LPCM tests I mentioned above were done using gray-market AACS-stripping software. Still broken.

These are enthusiasts. They're the ones who are consulted when their friends want to buy a TV, a stereo, or a home theater PC. They want to buy Intel, they really do-- no one in the HTPC space contests our CPU supremacy. And the price point is right.

But right now their recommendation is suffering from those two issues-- the second of which is going to bite us in the ass if it isn't fixed soon, because AV receivers are not exactly uncommon and unless I miss my mark anyone buying one of those sexy new $500+ Dell Hybrids with the G965 in them are going to have this same problem if they get the "Blu-ray edition". I don't know what we can do at Intel to implore our partners in the software player companies to get this done, but we need them to implement a fix for this issue ASAP. I know we must be engaged with them on this, I just don't see why it's not implemented yet. It's been at least three months, and it's making us, not them, look bad in the marketplace.

We can get over the crappy demos from the competition by demonstrating it was a rigged or poorly configured system. We can overcome bad initial reviews on video quality by showing software players that do have our ClearVideo features fully enabled, and by continuous improvement in our drivers (G35 went from initial standard-def HQV scores of 70 at launch to a perfect score of 130)-- I know from what I've seen with my own eyes that we have good hardware. But we are sunk in the HTPC space if the home user can't get their equipment to even work with our graphics, or have to buy hacker software to do so. Even if we're not the root cause.

Conclusion

The G45 is really the perfect home theater chipset (at least until such time as it is superceded by one which can transmit Dolby TrueHD and DTS HD-MA... that's a topic for another time) and the motherboards we're putting out using it are elegantly designed (the DG45FC is a thing of beauty). The drivers are even in reasonably good health for this stage of the chipset's release. But it galls me to see such great work marred by two issues which seem correctable but which are still extant.

That's how I see it.

Categories: Visual Computing

Comments (56)

Trackbacks (0)

Leave a comment  

Name (required)*

Email (required; will not be displayed on this page)*

Your URL (optional)


Comment*