AMD: R8xx Speculation

[rpg.314]

Looking at those numbers I would say ATI is in trouble IF nVidia has a G300 card out by November.

Why? gt300 is supposed to compete with 5870x2 and not 5870, remember. At any rate, how much faster are you expecting gt300 to be vis-a-vis 5870?

 

[psolord]

Looking at those numbers I would say ATI is in trouble IF nVidia has a G300 card out by November.

The performance seems to be solid, but we don't know what resolution they used!

It could be 2560X1600 and then the performance difference may not be so high. I don't care about 2560X1600 anyway but still some people will!

I would expect better performance for Crysis anyway! I mean these benchmarks seem like it is equal to the 4870X2 and one should expect better, since Crysis is one of the worst scaling games on the X2!

So, no chance for 60fps Crysis @1080P DX10 Very high on a single 5870. Hopefully the 5870X2 could do it, but what is the point if GT300 could do it single chip?

 

[psolord]

Agreed. But then you should include CSAA 16x vs MSAA 8x numbers there as well.

You have an agreement by me too!

I still haven't even tryed my 16XQAA modes on my GTX 260 but people are arguing that 8XMSAA is equal or even better than Nvidia's 16XAA modes!

I will look into it though!

 

[Slappi]

The performance seems to be solid, but we don't know what resolution they used!

It could be 2560X1600 and then the performance difference may not be so high. I don't care about 2560X1600 anyway but still some people will!

I would expect better performance for Crysis anyway! I mean these benchmarks seem like it is equal to the 4870X2 and one should expect better, since Crysis is one of the worst scaling games on the X2!

So, no chance for 60fps Crysis @1080P DX10 Very high on a single 5870. Hopefully the 5870X2 could do it, but what is the point if GT300 could do it single chip?

True, but I am sure ATI maximized the results, which they should.

 

[Slappi]

Why? gt300 is supposed to compete with 5870x2 and not 5870, remember. At any rate, how much faster are you expecting gt300 to be vis-a-vis 5870?

I have no idea.... by in trouble I just mean based on expectations.... it is still a fast card if those numbers are true.

However, AMD needs to turn a profit or they will be in trouble soon. Even with the great GPUs they have put out in he last year, they ae still losing money on them. If nVidia reaches even parity with the 58x0 lineup and they do it within 2 months.... I think AMD will continue to lose money in their graphics division.

nVidia really needs to get a GT300 card out before end of November or at the very least by EOY to be a threat to ATI's earnings.

I am obviously invested in nVidia or I would actually want AMD to come out on top. I like AMD, just not as an investment. I actually think they will go out of business.... they have too much debt and are losing tons of money.

 

[neliz]

I have no idea.... by in trouble I just mean based on expectations.... it is still a fast card if those numbers are true.

However, AMD needs to turn a profit or they will be in trouble soon. Even with the great GPUs they have put out in he last year, they ae still losing money on them. If nVidia reaches even parity with the 58x0 lineup and they do it within 2 months.... I think AMD will continue to lose money in their graphics division.

nVidia really needs to get a GT300 card out before end of November or at the very least by EOY to be a threat to ATI's earnings.

I am obviously invested in nVidia or I would actually want AMD to come out on top. I like AMD, just not as an investment. I actually think they will go out of business.... they have too much debt and are losing tons of money.

You have got no clue really?

Guess which department within AMD is one of the very few profitable?

 

[Prometheus]

Looking at those numbers I would say ATI is in trouble IF nVidia has a G300 card out by November.

5870 seems to be faster than the GTX295 in the graphs. Why would Ati be in trouble?
Given these premature numbers I'd say that nvidia is in trouble even if they bring GT300 early.

 

[no-X]

Better graph: http://i28.tinypic.com/r9oill.jpg

This one doesn't start in 0, but in -20... that makes the HD5870 results worse, than in reality.

I edited it a bit - this one starts in zero and shows real percents:

 

[Florin]

You have got no clue really?

Guess which department within AMD is one of the very few profitable?

The one selling off fabs?

You can't mean the graphics unit, they reported a loss of 12$ million last quarterly results. Their dump prices really managed to erode margins in the industry.

 

[Slappi]

You have got no clue really?

Guess which department within AMD is one of the very few profitable?

graphics (2) Quarter ending June 27th 2009
Net revenue 251
Operating income (loss) (12 )


Their graphics division lost 12 million dollars last quarter.

That is straight from their press release found here :

http://phx.corporate-ir.net/phoenix.zhtml?c=74093&p=irol-newsArticle&ID=1309675&highlight=



Sounds like you are the one without a clue.

 

[Sound_Card]

x1.6 RV770 might turn out to be a pretty accurate description if the above graphs are true, don't you think?

Those graphs show on average 40% faster then the GTX295. I don't know how that translates to only 60% faster then the HD 4890.

 

[Kaotik]

Agreed. But then you should include CSAA 16x vs MSAA 8x numbers there as well.

Huh? Why?
Only ATI 8xMSAA vs nV 8xMSAA (8xQAA I think the name was?) should be used, since they're the only really comparable scenes.

 

[mczak]

The one selling off fabs?

You can't mean the graphics unit, they reported a loss of 12$ million last quarterly results. Their dump prices really managed to erode margins in the industry.

Well despite reporting a loss, the graphics unit was still the most profitable one . Not that nvidia did any better though. I don't think it really was the "dump" prices hurting them, rather than the recession (for the graphics unit - I think the computing unit is and will be in trouble).

 

[DegustatoR]

Those graphs show on average 40% faster then the GTX295. I don't know how that translates to only 60% faster then the HD 4890.

40% faster with MSAA 8x? MSAA 8x on 4890 has the same performance hit as on G200?

Huh? Why?
Only ATI 8xMSAA vs nV 8xMSAA (8xQAA I think the name was?) should be used, since they're the only really comparable scenes.

Because CSAA 16x is often comparable to MSAA 8x in quality but has almost no performance hit compared to MSAA 4x on GFs.

 

[Ailuros]

Because CSAA 16x is often comparable to MSAA 8x in quality but has almost no performance hit compared to MSAA 4x on GFs.

I strongly disagree to that; while CSAA is definitely better than 4x MSAA it's in no way as good across the scene as 8xMSAA. Even more so if you also enable transparency multisampling on either/or which doesn't cost that much more performance either.

 

[Kaotik]

Because CSAA 16x is often comparable to MSAA 8x in quality but has almost no performance hit compared to MSAA 4x on GFs.

Apples to apples isn't the same as to apples to fruits looking like apples

 

[no-X]

Slappi: ATi numbers were around zero for the last quarters. nVidia and AMD (CPU div.) are in much worse situation. -$12M/quarter is nothing compared to the other two companies, esp. in these days.

 

[Sound_Card]

don't know if this has been posted...
http://translate.google.com/transla...jsou-venku.html&sl=auto&tl=en&history_state0=

HD 5870
From 5% to 155% compared to GTX 285 (tested at 17 games and 40 configurations)
From -25% to +95% compared to GTX 295 (AA 2x/8x, AF 8x/16x in 16 games)

Crossfire
X2 from -5% to +90% compared to GTX 295 SLI
X3 from 10% to 160% compared to GTX 295 (tested in 15 games)

 

[homerdog]

Apples to apples isn't the same as to apples to fruits looking like apples

True, but then where does CSAA fit in? It seems wrong to just ignore it.

 

[no-X]

Maybe CSAA 16x can be compared to ED 12x (?)