The Official RV630/RV610 Rumours & Speculation Thread

[Arun]

Please continue UVD discussion in the following thread. All posts in this thread will be moved to a new thread and locked, so that if there's any intelligent discussion left to be done, it can be in the new thread.
EDIT: Oh, and please resume discussing RV610 and RV630, their performance relative to G84/G86, the power consumption, PCBs/fans, and whatever else crosses your mind that I'm not thinking of right now.

 

[AnarchX]

MSI HD 2600XT GDDR4 "test"

http://forum.coolaler.com/showpost.php?p=1776021&postcount=28

 

[pakotlar]

wow those fear scores are pathetic. ati dropped the ball this round.

 

[Evildeus]

Yeah not great....

 

[Pressure]

wow those fear scores are pathetic. ati dropped the ball this round.

Yeah not great....

It still seems to be better than the Geforce 8600GTS.

Company of Heroes, 1024x768 scores:

Geforce 8600GTS: 45 FPS
Radeon HD 2600XT: 49.6 FPS

1280x1024 scores:

Geforce 8600GTS: 34.3 FPS
Radeon HD 2600XT: 39 FPS

Can't compare the F.E.A.R scores from any reviews because nv6800 has Soft Shadows enabled (and we know how Soft Shadows totally kills the performance).

 

[trinibwoy]

Where are the 8600GTS numbers from?

 

[Evildeus]

On a Intel Core 2 Duo E6600...
Company Of Heros : 1280x1024 max:

http://www.tt-hardware.com/modules.php?name=News&file=article&sid=10389

 

[Pressure]

Where are the 8600GTS numbers from?

From Anandtech's review of the G84.

They used pretty much the same hardware,

Intel Core 2 Extreme X6800 (2.93GHz/4MB)
P965 (nv6800) vs 680i SLI (Anandtech)
DDR2-HK-800 4x512MB with 5-5-5-18 timings (nv6800) vs Corsair 2x1GB DDR2 with 4-4-4-12 timings (Anandtech)

A quick and dirty comparison but granted, that is all we have to play with so far.

 

[seahawk]

So roughly X1950PRO level. Not muchz better then 8600GTS either.

 

[Evildeus]

Doesn't Anandtech put AF by default?

 

[_xxx_]

Gee, so it seems like all hope is lost for my SFF and I'll be skipping this gen alltogether unless nV does the half-G80 thingy soon. Meh.

 

[trinibwoy]

From Anandtech's review of the G84.

A quick and dirty comparison but granted, that is all we have to play with so far.

Thanks. But it seems other sites got better numbers - tweaktown/bjorn3d for example. Guess we'll have to wait for the official reviews to see how they line up but it doesnt seem like there'll be much between them at all.

 

[Pressure]

Gee, so it seems like all hope is lost for my SFF and I'll be skipping this gen alltogether unless nV does the half-G80 thingy soon. Meh.

It should be pretty obvious that both NVIDIA and ATI could not resist the cash cow that the mobile segment has become. These chips are clearly made for that market in mind.

 

[_xxx_]

It should be pretty obvious that both NVIDIA and ATI could not resist the cash cow that the mobile segment has become. These chips are clearly made for that market in mind.

Seems so, but still I don't think that neglecting a large portion of the market looking for upper-midrange single-slot parts was such a great idea. Ok, ATI had some struggles with R6xx, but nV surely could have built a part like that if so wished. Oh well.

 

[Geeforcer]

It still seems to be better than the Geforce 8600GTS.

Company of Heroes, 1024x768 scores:

Geforce 8600GTS: 45 FPS
Radeon HD 2600XT: 49.6 FPS

1280x1024 scores:

Geforce 8600GTS: 34.3 FPS
Radeon HD 2600XT: 39 FPS

Can't compare the F.E.A.R scores from any reviews because nv6800 has Soft Shadows enabled (and we know how Soft Shadows totally kills the performance).

Did you see how many graphics options there are in CoH? It review-to-review comparisons largely useless unless it's explicitly stated which ones have been enabled in each one.

 

[Silent_Buddha]

In other words, wait for proper reviews.

Personally I'm still scratching my head about NV leaving out decode acceleration for VC-1. Supporting h.264 and mpeg2 decode is fine. However, anyone planning on using either Blu-Ray or HD-DVD is going to need all three.

Then again until ATI can demonstrate that it has proper acceleration of VC-1, I guess it's not a deal breaker.

I'm wondering if both NV and ATI decided to just target HTPC with their mid-range parts this generation. It doesn't appear (yet) that either company is going to have a compelling upgrade to previous generation mid-range parts if all you consider is performance in current generation games.

DX10 is a large unknown and I'd imagine a mid-range part is going to be even more short lived compared to a high end part when it comes to upgrades for anyone interested in DX10 gaming. After all is it actually reasonable to expect that a mid-range part would offer compelling performance in future DX10 games?

Regards,
SB

 

[nicolasb]

Personally I'm still scratching my head about NV leaving out decode acceleration for VC-1. Supporting h.264 and mpeg2 decode is fine. However, anyone planning on using either Blu-Ray or HD-DVD is going to need all three.

To be fair, software VC-1 decoding is a lot less demanding of CPU resources than than software h.264 decoding. But I agree it's an odd omission. HTPC users who employ a passively cooled RV630 or RV610 card will probably be able to scale back their CPU speed to the point where that can be passively cooled as well. Combine that with a fanless PSU and some quiet 120mm case fans and damped hard-drive mountings, and you should have an impressively low noise-floor.

 

[Anarchist4000]

If AMD is still having driver problems with the R600 wouldn't it translate into the low and mid range parts?

 

[3dilettante]

Someone spoke somewhere in some thread about possible issues utilizing the ring bus efficiently.

If we assume for a moment that this is at least one issue with R600, then could we expect better for the smaller cores?

R600 has 30+ clients on its internal bus.
How many would a chip with only a fraction of the SIMD units, fewer memory channels, and texture units have?

What is the ring bus width for the lower variants?
If the width is otherwise unaffected, we could see better utilization of the ring bus on the lower end chips.

edit:
better utlization of the bus or fewer cases where the bus or bus latency is a limiting factor

 

[Aerows]

MSI HD 2600XT GDDR4 "test"

http://forum.coolaler.com/showpost.php?p=1776021&postcount=28

Ugh...worse than 8600GTS. CPU scores seem low in 3DMark, though. Not tremendously, but still a bit low. The drivers must not be very efficient yet. They will likely drag a bit more performance out of them before they get to reviewers and the street. Inefficient use of the CPU would cause bad FEAR scores.

If AMD is still having driver problems with the R600 wouldn't it translate into the low and mid range parts?

I think it might impact them more than the higher end. The CPU scores are not what I would expect from that processor.