CrossFire Slipping?

[CJ]

Supposedly Catalyst 5.7 already has the "Multi-GPU Catalyst Features"...

Can someone with Cat 5.7 somehow look inside the code and see if SuperAA (8x, 10x, 12x, 14x) has been enabled as well as check if Catalyst AI has been updated to "determine the optimal Multi-VPU solution for both Direct 3D and OpenGL applications"?

 

[Mendel]

At least there aren't any new cards such as master cards listed in the installation inf. (C2_24085.inf)

 

[Geo]

Boy Wonder has an. . .err, odd, take on CrossFire here:

http://www.anandtech.com/weblog/default.aspx

Odd a couple ways, particularly his analysis of what is really going on with recent XFire benchmarks. Frankly, his explanation looks like a mess to me, and either makes ATI "ignore the man behind the curtain" liars (only it turns out it doesn't matter, in his take --in mine it would be *very* troubling), or he doesn't really understand what he was told.

I wonder which?

I mean, really, if you are Anand, why would you believe a vendor which told you they faked their results with a driver trick, but not to worry --the real thing, which they haven't been able to ship yet, and by your own analysis won't ship anywhere close to where they promised, will turn out just the same??

Ish. I usually try to be kinder than this, but I really hope he has his head up his *ss on this one. Actually, I can't figure out any way in which he *doesn't*, so I guess I hope he does in the way which doesn't make ATI guilty of a fairly serious sin re the Computex and subesquent XFire benchies.

 

[digitalwanderer]

I read it more as he's misunderstanding it. I'm not sure if it should be counted as a driver "trick" to "only" render every other frame...they're actually rendering each frame but since there is no composite chip they only display every other frame.

That could actually be included in the drivers for Crossfire to work, just it still works without the compositing chip 'cept it won't display half of them.

I'm chalking it up to confusion for now, but it's something to keep an eye on.

 

[Geo]

As I said, I hope that's what we're looking at, and from what I understand from the other articles I've read, possible. I hope the ATI boys straighten him out, and he corrects it --as much traffic as he gets it would be very unfortunate to leave it the way it is now.

It's a pretty simple calculation he set up there: ("Designed to simulate") - (full disclosure) = *LLPOF

On the more troubling side, it *would* explain why the HKEPC X850 benchies were dead-on (within margin of error, from what I could see) using "real" XFire vs two-slave XFire.



*Liar, Liar, Pants on Fire

 

[digitalwanderer]

On the more troubling side, it *would* explain why the HKEPC X850 benchies were dead-on (within margin of error, from what I could see) using "real" XFire vs two-slave XFire.

But if the composite chip works the way ATi says it works then in practice it will give the same results...it will just also display those frames as well as render them...so it may not be faked.

 

[Geo]

Yeah, and I hope that's what happened. . .rather than someone saying, "well, theoretically, when we really have this thing working, it will turn out this way. . .so let's just hack it for now, and let Anand run his tests."

I suppose we might be able to figure out by replacing the second card with something really much slower, hacking the id, and running. . .that would tell you whether the second card is actually doing anything at all.

Tho, damn it, you'd need a real XFire master to test that and the drivers they used.

 

[BRiT]

Boy Wonder is a Twit...

 

[Geo]

Ah. Occam's Razor cuts again.

 

[trinibwoy]

^^

 

[Kombatant]

I am certain the target date has slipped.. by how much, we'll see I guess.

 

[EntombeD]

Gorgefest of ATI products doesn't sound like a good idea.

How much would you pay for a R4x0 master card when R520 cards are out there?

€ 4,20 would be the max

 

[Sinistar]

With this being on ATI's site
http://ati.com/buy/promotions/crossfireready/index.html
you would think crossfire would be right around the corner.

 

[Geo]

Damn, that was pretty interesting. Nice offer.

Found this one even more interesting: http://ati.com/buy/promotions/crossfireready/motherboards.html

So, where the heck are they??

 

[Geo]

According to Inq, "SLI AA" will be available tomorrow. Way to go, ATI. . .innovate and still get beaten to market. <sigh>

http://www.theinquirer.net/?article=24720

 

[trinibwoy]

According to Inq, "SLI AA" will be available tomorrow. Way to go, ATI. . .innovate and still get beaten to market. <sigh>

http://www.theinquirer.net/?article=24720

I hope they have some more surprises because it looks like Nvidia took those SLI/Crossfire comparisons to heart and are slowly chipping away at anything they considered lacking. They've matched the default SLI mode and now superAA - hopefully ATi has the performance to beat SLI cause the feature advantage (on paper) seems to be gone.

 

[ChrisRay]

According to Inq, "SLI AA" will be available tomorrow. Way to go, ATI. . .innovate and still get beaten to market. <sigh>

http://www.theinquirer.net/?article=24720

Well. If they had previewed the product later when they had it. This likely wouldnt of happened.

 

[Geo]

True enuf. Tho if they'd even hit their initial schedule of review hardware to the reviewers by the end of June. . .

 

[neliz]

Damn, that was pretty interesting. Nice offer.

Found this one even more interesting: http://ati.com/buy/promotions/crossfireready/motherboards.html

So, where the heck are they??

You know what I'm wondering? if "R520" is still supposed to launch AFTER crossfire.
(giving the definitive date of September 29th in that x8*0 CE offer.. it would be R520 within 2 months..

 

[ondaedg]

I am just going to forget about the R520. If I wake up one morning and see all these reviews hitting the net, then it'll be one of those nice pleasant surprises.