Poll: GT300 VS. HD5870

[alcoholbob]

All AMD would have to do is some cherry picking for 5890 parts. The RAM can be a problem for some, but 5870 runs cool enough that a 50mhz core OC barely makes a dent in the temperature, and you get a good 4% performance boost.

 

[dguy6789]

Originally posted by: Wreckage

Originally posted by: RussianSensation

> 18%[/b] check out Firingsquad for example.

Once again...... the graph i linked used several sites and games. Certainly you could cherrypick a site or game. But I would assume it would be more prudent to look beyond just 1 source. Having read through the 5870 thread and other threads on multiple forums, it looks like around 20% seems fairly accurate.

And just how many of those benchmarks that were included in the 20% average had cpu limited settings like 1680x1050 with no aa and no af? 20% is extremely conservative and borderline inaccurate. Anandtech has the 5870 averaging 34% faster than the GTX 285.

 

[Jacen]

Comparing speculation-based hardware to in-the-hand gpus is pretty absurd. We really have little idea what is going to happen with the GT300 though I am still hearing the rumors that it won't be out until CES.

 

[Idontcare]

Originally posted by: Nemesis 1
I really don't understand this thread,

I assume that if the NV385 isn't launched by Jan . A simular poll between the ATI 6870 vs the NV 385 would be just as relavent as this one. Honestly these kinda threads are NV hype threads comparring non existant parts against real parts .

But it has been allowed so When the and IF the NV385 launches ATIs 6870 will be as relavent as this BS hype thread. Don't BUY the NV385 because ATIs soon to be launched 6870 slaps it silly. LOL this is a joke right? No responsiable enthusiast is falling for this BS hype on a product thats 6 months away.

Which then begs the question - where is the harm then in having threads like this?

If no one is falling for the hype, as you state, then why go to the trouble to post and vent about a thread's existence? Who needs to read the venting?

 

[Idontcare]

Originally posted by: MODEL3

Originally posted by: Shaq
Also the FS benchmarks were all at 2560 with 4AA in some cases. It would make more sense to bench at 1680 and 1920 and average the results because the vast majority of gamers have a 22" or 24" monitor. Just showing the highest resolutions and high levels of AA is trying to skew the results in favor of ATI. It is established that the 5870 is better with high AA and high resolutions. But does it really matter if a 285 gets 60 FPS and the 5870 gets 75 FPS in a certain game? They are both very playable. It comes down to cost and availability at that point and for some people power consumption and heat. The 285 can be found for $296 versus the $379 5870 if you can find it. That is 25% cheaper and about how much faster the 5870 is compared to the 285 at common resolutions.

I don't have a problem with 2560 res testing, although i would like 1920 testing also.
For this kind of cards 1680 testing (at launch) is not essential imo.

I don't like 8X AA testing because it favors ATI.
It's like if you want to favor NV, do test at 0X AA.
(of cource i prefer 8X AA testing than 0X AA testing)
The thing is 4X AA produce very good results and the difference between 4X AA and 0X AA is big imo but the difference between 8X AA and 4X AA is much smaller imo regarding quality perception.

---------------------------------------------------------
Regarding prices, you forgot $259 5850.

The obvious answer is you (a responsible reviewer) tests all three combinations at all relevant resolutions so the readers who aim to become buyers have the data that applies to them (resolution and likely AF/AA usage preferences) so they can pare down the data to make just the comparisons they feel represent their likely gaming usage.

People who want to create one-dimensional metrics of comparison (like 3Dmark) are interested in what you write about, and what does practically everyone say about those types of benchmarks?

 

[Tempered81]

Originally posted by: Wreckage

Originally posted by: RussianSensation

> 18%[/b] check out Firingsquad for example.

Once again...... the graph i linked used several sites and games. Certainly you could cherrypick a site or game. But I would assume it would be more prudent to look beyond just 1 source. Having read through the 5870 thread and other threads on multiple forums, it looks like around 20% seems fairly accurate.

lol.

Originally posted by: dguy6789
And just how many of those benchmarks that were included in the 20% average had cpu limited settings like 1680x1050 with no aa and no af?

Exactly.

 

[Qbah]

Originally posted by: dguy6789

Originally posted by: Wreckage

Originally posted by: RussianSensation

> 18%[/b] check out Firingsquad for example.

Once again...... the graph i linked used several sites and games. Certainly you could cherrypick a site or game. But I would assume it would be more prudent to look beyond just 1 source. Having read through the 5870 thread and other threads on multiple forums, it looks like around 20% seems fairly accurate.

And just how many of those benchmarks that were included in the 20% average had cpu limited settings like 1680x1050 with no aa and no af? 20% is extremely conservative and borderline inaccurate. Anandtech has the 5870 averaging 34% faster than the GTX 285.

So I did some calculations myself. At 1920x1200m AAx4, AAx2 or no AA - depending on the review, mostly maxed details - based on all those reviews:

Crysis Warhead: HD5870 is 32% faster than a GTX285 in AVG FPS
FarCry 2: HD5870 is 31% faster than a GTX285 in AVG FPS
Fallout 3: HD5870 is 18% faster than a GTX285 in AVG FPS

Some of those links use 1680x1050 and lower. One of the reviews used only 1440x900 and 1680x1050 and the GTX285 was winning there (talk about bullshit - useless results!). Some links had no numbers, just an article. Crysis was in 13 reviews, FarCry 2 in 10 and Fallout 3 in three only. Fallout 3 is a severely CPU limited even at that resolution.

We have no idea what was added, how they used the information - there's nothing described in what they mean "on average". Useless information.


EDIT: Not to mention the GTX275 is a hair away from the GTX285 based on their charts, but GTX285 in SLI totally kills the GTX275 in SLI? Load of bull, that's what this chart is.

 

[Janooo]

Originally posted by: MODEL3

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo
5870 delivers 2.6x4890 in CoH at the top resolution and settings.

Like i said, i know the results.
In this particular case the 2,6X figure is for 2560X1600 with 8X AA & 16XAF.
If you drop the antialiasing at 4X the difference at 2560X1600 16XAF is 1,95X.
And the DX10 codepath of CoH is extremely pixel shader oriented.
All the above are predictable and natural results.

I am not here to state the obvious.
I am trying to figure out what are the limitations of the Cypress because it shows according to my perseption strange results.
My analysis is complex and to write down it will take me more time than it took to figure out the situation.
Sorry i am not willing to lose time.
If you disagree with me that the 5870 has (except the obvious driver and bandwidth related issues) the following issues:

Originally posted by: MODEL3
1.Geometry/vertex performance issues (in the sense that the classic vertex unit cannot generate 2X geometry in relation with 4890) (my main assumption)

or/and

2.Geometry/vertex shading performance issues (in the sense that the geometry shader [GS] cannot achieve 2X speed in relation with 4890)(another possible assumption)

My assumption is that ATI though that many of the future DX11 games will use vertex techniques based on the tesselator unit.

Then i respect your opinion, but i am not willing to do anything about it.

I just thought to point out what the reviewers with all their knowledge,year in this business, technical background, briefings from ATI, 5870 cards at hand, failed to point out.

Sure i don't have technical background, nor i was at ATI briefings and of cource i don't have a 5870 card in order to be able to test my theory.
So with all these disadvatntages i may be wrong.
The funny thing will be, to be right about it.
And no, it will be no luck at all, trust me.

What part of silicon performs geometry? Is it doubled from 4890 to 5870 or not?
Geometry is not an issue. The BW is the first in line.

Certainly BW in general is first in line, when we examine the potential reasons.
I said that a 800-850MHz 32ROPS/1600SP design with 256bit memory controller (with the current GDDR5 ICs) will be bandwidth limited before 1,5 freakin month.
This is extremely easy to figure out.
But this kind of performance behavior isn't only BW related.
According to my perception Geometry is an issue, so i guess we have to agree to disagree about this matter.

Geometry is done in SP's. They were doubled. It's not an issue unless SP's are starved at the point of doing it. Hence the BW is the first suspect.
5870 shows its potential at the top resolution with the fastest i7. Dual core setups are not good enough.

 

[Keysplayr]

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo
5870 delivers 2.6x4890 in CoH at the top resolution and settings.

Like i said, i know the results.
In this particular case the 2,6X figure is for 2560X1600 with 8X AA & 16XAF.
If you drop the antialiasing at 4X the difference at 2560X1600 16XAF is 1,95X.
And the DX10 codepath of CoH is extremely pixel shader oriented.
All the above are predictable and natural results.

I am not here to state the obvious.
I am trying to figure out what are the limitations of the Cypress because it shows according to my perseption strange results.
My analysis is complex and to write down it will take me more time than it took to figure out the situation.
Sorry i am not willing to lose time.
If you disagree with me that the 5870 has (except the obvious driver and bandwidth related issues) the following issues:

Originally posted by: MODEL3
1.Geometry/vertex performance issues (in the sense that the classic vertex unit cannot generate 2X geometry in relation with 4890) (my main assumption)

or/and

2.Geometry/vertex shading performance issues (in the sense that the geometry shader [GS] cannot achieve 2X speed in relation with 4890)(another possible assumption)

My assumption is that ATI though that many of the future DX11 games will use vertex techniques based on the tesselator unit.

Then i respect your opinion, but i am not willing to do anything about it.

I just thought to point out what the reviewers with all their knowledge,year in this business, technical background, briefings from ATI, 5870 cards at hand, failed to point out.

Sure i don't have technical background, nor i was at ATI briefings and of cource i don't have a 5870 card in order to be able to test my theory.
So with all these disadvatntages i may be wrong.
The funny thing will be, to be right about it.
And no, it will be no luck at all, trust me.

What part of silicon performs geometry? Is it doubled from 4890 to 5870 or not?
Geometry is not an issue. The BW is the first in line.

Certainly BW in general is first in line, when we examine the potential reasons.
I said that a 800-850MHz 32ROPS/1600SP design with 256bit memory controller (with the current GDDR5 ICs) will be bandwidth limited before 1,5 freakin month.
This is extremely easy to figure out.
But this kind of performance behavior isn't only BW related.
According to my perception Geometry is an issue, so i guess we have to agree to disagree about this matter.

Geometry is done in SP's. They were doubled. It's not an issue unless SP's are starved at the point of doing it. Hence the BW is the first suspect.
5870 shows its potential at the top resolution with the fastest i7. Dual core setups are not good enough.

Does quad core benefit in every game? Not really at this point. Maybe many more games next year will be able to utilize as many cores as a system has. That would be excellent.
So, is it safe to say that the 5870 ranges anywhere from 18 to 40% faster than a GTX285 depending on game and res?
Or is this not satisfactory..

 

[Qbah]

Originally posted by: Keysplayr
Does quad core benefit in every game? Not really at this point. Maybe many more games next year will be able to utilize as many cores as a system has. That would be excellent.
So, is it safe to say that the 5870 ranges anywhere from 18 to 40% faster than a GTX285 depending on game and res?
Or is this not satisfactory..

That is pretty much it based on what I've seen. With a note that the lower difference in some games is mostly due to CPU limited resolution - nobody in their right mind will suggest buying a HD5870 to a person running at 1680x1050 or lower Unless they want to max Crysis only

 

[dguy6789]

The limitation on the 5870 is most certainly memory bandwidth. Memory bandwidth is the only category in which the 5870 does not outclass the 4870x2 in.

Radeon HD 5870:

Pixel Fill Rate: 27200 MPixels/sec
Texture Fill Rate: 68000 MTexels/sec
Flops: 2720 GFLOPS
Memory Bandwidth: 153.6GB/sec

Radeon HD 4870x2:

Pixel Fill Rate: 24000 MPixels/sec
Texture Fill Rate: 60000 MTexels/sec
Flops: 2400 GFLOPS
Memory Bandwidth: 230.4GB/sec

 

[SlowSpyder]

Originally posted by: Keysplayr

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo
5870 delivers 2.6x4890 in CoH at the top resolution and settings.

Like i said, i know the results.
In this particular case the 2,6X figure is for 2560X1600 with 8X AA & 16XAF.
If you drop the antialiasing at 4X the difference at 2560X1600 16XAF is 1,95X.
And the DX10 codepath of CoH is extremely pixel shader oriented.
All the above are predictable and natural results.

I am not here to state the obvious.
I am trying to figure out what are the limitations of the Cypress because it shows according to my perseption strange results.
My analysis is complex and to write down it will take me more time than it took to figure out the situation.
Sorry i am not willing to lose time.
If you disagree with me that the 5870 has (except the obvious driver and bandwidth related issues) the following issues:

Originally posted by: MODEL3
1.Geometry/vertex performance issues (in the sense that the classic vertex unit cannot generate 2X geometry in relation with 4890) (my main assumption)

or/and

2.Geometry/vertex shading performance issues (in the sense that the geometry shader [GS] cannot achieve 2X speed in relation with 4890)(another possible assumption)

My assumption is that ATI though that many of the future DX11 games will use vertex techniques based on the tesselator unit.

Then i respect your opinion, but i am not willing to do anything about it.

I just thought to point out what the reviewers with all their knowledge,year in this business, technical background, briefings from ATI, 5870 cards at hand, failed to point out.

Sure i don't have technical background, nor i was at ATI briefings and of cource i don't have a 5870 card in order to be able to test my theory.
So with all these disadvatntages i may be wrong.
The funny thing will be, to be right about it.
And no, it will be no luck at all, trust me.

What part of silicon performs geometry? Is it doubled from 4890 to 5870 or not?
Geometry is not an issue. The BW is the first in line.

Certainly BW in general is first in line, when we examine the potential reasons.
I said that a 800-850MHz 32ROPS/1600SP design with 256bit memory controller (with the current GDDR5 ICs) will be bandwidth limited before 1,5 freakin month.
This is extremely easy to figure out.
But this kind of performance behavior isn't only BW related.
According to my perception Geometry is an issue, so i guess we have to agree to disagree about this matter.

Geometry is done in SP's. They were doubled. It's not an issue unless SP's are starved at the point of doing it. Hence the BW is the first suspect.
5870 shows its potential at the top resolution with the fastest i7. Dual core setups are not good enough.

Does quad core benefit in every game? Not really at this point. Maybe many more games next year will be able to utilize as many cores as a system has. That would be excellent.
So, is it safe to say that the 5870 ranges anywhere from 18 to 40% faster than a GTX285 depending on game and res?
Or is this not satisfactory..

Not sure if quads matter yet, but according to what I've read, DX11 will better use multicore setups. I didn't see details on just what DX11 will do or how it will do that, though.

Guys, obviously the 18% is just wrong or at a low resolution with stupid low settings where the GPU isn't the limiting factor more often than not. There are examples of the 5870 being 100%+ faster than the GTX285, they obviously aren't meant or gong to be competitors.

Also, people who have $140 22" monitors generally aren't looking at $379 graphics cards.

 

[dguy6789]

Originally posted by: Keysplayr

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo
5870 delivers 2.6x4890 in CoH at the top resolution and settings.

Like i said, i know the results.
In this particular case the 2,6X figure is for 2560X1600 with 8X AA & 16XAF.
If you drop the antialiasing at 4X the difference at 2560X1600 16XAF is 1,95X.
And the DX10 codepath of CoH is extremely pixel shader oriented.
All the above are predictable and natural results.

I am not here to state the obvious.
I am trying to figure out what are the limitations of the Cypress because it shows according to my perseption strange results.
My analysis is complex and to write down it will take me more time than it took to figure out the situation.
Sorry i am not willing to lose time.
If you disagree with me that the 5870 has (except the obvious driver and bandwidth related issues) the following issues:

Originally posted by: MODEL3
1.Geometry/vertex performance issues (in the sense that the classic vertex unit cannot generate 2X geometry in relation with 4890) (my main assumption)

or/and

2.Geometry/vertex shading performance issues (in the sense that the geometry shader [GS] cannot achieve 2X speed in relation with 4890)(another possible assumption)

My assumption is that ATI though that many of the future DX11 games will use vertex techniques based on the tesselator unit.

Then i respect your opinion, but i am not willing to do anything about it.

I just thought to point out what the reviewers with all their knowledge,year in this business, technical background, briefings from ATI, 5870 cards at hand, failed to point out.

Sure i don't have technical background, nor i was at ATI briefings and of cource i don't have a 5870 card in order to be able to test my theory.
So with all these disadvatntages i may be wrong.
The funny thing will be, to be right about it.
And no, it will be no luck at all, trust me.

What part of silicon performs geometry? Is it doubled from 4890 to 5870 or not?
Geometry is not an issue. The BW is the first in line.

Certainly BW in general is first in line, when we examine the potential reasons.
I said that a 800-850MHz 32ROPS/1600SP design with 256bit memory controller (with the current GDDR5 ICs) will be bandwidth limited before 1,5 freakin month.
This is extremely easy to figure out.
But this kind of performance behavior isn't only BW related.
According to my perception Geometry is an issue, so i guess we have to agree to disagree about this matter.

Geometry is done in SP's. They were doubled. It's not an issue unless SP's are starved at the point of doing it. Hence the BW is the first suspect.
5870 shows its potential at the top resolution with the fastest i7. Dual core setups are not good enough.

Does quad core benefit in every game? Not really at this point. Maybe many more games next year will be able to utilize as many cores as a system has. That would be excellent.
So, is it safe to say that the 5870 ranges anywhere from 18 to 59% faster than a GTX285 depending on game and res?
Or is this not satisfactory..

Fixed and I'll agree to that. (World of Warcraft, Anandtech Article)

 

[MODEL3]

Originally posted by: Idontcare
The obvious answer is you (a responsible reviewer) tests all three combinations at all relevant resolutions so the readers who aim to become buyers have the data that applies to them (resolution and likely AF/AA usage preferences) so they can pare down the data to make just the comparisons they feel represent their likely gaming usage.

People who want to create one-dimensional metrics of comparison (like 3Dmark) are interested in what you write about, and what does practically everyone say about those types of benchmarks?

It would be nice if sites had all the meaningful combinations.
Sadly many times this is not possible (for example time constrains)
Especially at launch of a new product, this is very difficult.
I guess, what it would be nice is at least to have some follow up articles.

 

[Keysplayr]

Originally posted by: SlowSpyder

Originally posted by: Keysplayr

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo
5870 delivers 2.6x4890 in CoH at the top resolution and settings.

Like i said, i know the results.
In this particular case the 2,6X figure is for 2560X1600 with 8X AA & 16XAF.
If you drop the antialiasing at 4X the difference at 2560X1600 16XAF is 1,95X.
And the DX10 codepath of CoH is extremely pixel shader oriented.
All the above are predictable and natural results.

I am not here to state the obvious.
I am trying to figure out what are the limitations of the Cypress because it shows according to my perseption strange results.
My analysis is complex and to write down it will take me more time than it took to figure out the situation.
Sorry i am not willing to lose time.
If you disagree with me that the 5870 has (except the obvious driver and bandwidth related issues) the following issues:

Originally posted by: MODEL3
1.Geometry/vertex performance issues (in the sense that the classic vertex unit cannot generate 2X geometry in relation with 4890) (my main assumption)

or/and

2.Geometry/vertex shading performance issues (in the sense that the geometry shader [GS] cannot achieve 2X speed in relation with 4890)(another possible assumption)

My assumption is that ATI though that many of the future DX11 games will use vertex techniques based on the tesselator unit.

Then i respect your opinion, but i am not willing to do anything about it.

I just thought to point out what the reviewers with all their knowledge,year in this business, technical background, briefings from ATI, 5870 cards at hand, failed to point out.

Sure i don't have technical background, nor i was at ATI briefings and of cource i don't have a 5870 card in order to be able to test my theory.
So with all these disadvatntages i may be wrong.
The funny thing will be, to be right about it.
And no, it will be no luck at all, trust me.

What part of silicon performs geometry? Is it doubled from 4890 to 5870 or not?
Geometry is not an issue. The BW is the first in line.

Certainly BW in general is first in line, when we examine the potential reasons.
I said that a 800-850MHz 32ROPS/1600SP design with 256bit memory controller (with the current GDDR5 ICs) will be bandwidth limited before 1,5 freakin month.
This is extremely easy to figure out.
But this kind of performance behavior isn't only BW related.
According to my perception Geometry is an issue, so i guess we have to agree to disagree about this matter.

Geometry is done in SP's. They were doubled. It's not an issue unless SP's are starved at the point of doing it. Hence the BW is the first suspect.
5870 shows its potential at the top resolution with the fastest i7. Dual core setups are not good enough.

Does quad core benefit in every game? Not really at this point. Maybe many more games next year will be able to utilize as many cores as a system has. That would be excellent.
So, is it safe to say that the 5870 ranges anywhere from 18 to 40% faster than a GTX285 depending on game and res?
Or is this not satisfactory..

Not sure if quads matter yet, but according to what I've read, DX11 will better use multicore setups. I didn't see details on just what DX11 will do or how it will do that, though.

Guys, obviously the 18% is just wrong or at a low resolution with stupid low settings where the GPU isn't the limiting factor more often than not. There are examples of the 5870 being 100%+ faster than the GTX285, they obviously aren't meant or gong to be competitors.

Also, people who have $140 22" monitors generally aren't looking at $379 graphics cards.

Yes, we'll have to see what benefits DX11 has in store for multicore setups. It would be great if it could utilize all the power.
No, the 5870 surely wasn't meant to compete with a GTX285. That 100% over GTX285 isn't exactly common either. I think that across most review sites, that 40% performance improvement is generous at this point. Sometimes more, sometimes less.

And I know a lot of people who spent well north of 300.00 on a graphics card who had only 22" monitors. Crysis was the main culprit here. Believe it or not.

 

[Janooo]

Originally posted by: Keysplayr

Originally posted by: Janooo
...
Geometry is done in SP's. They were doubled. It's not an issue unless SP's are starved at the point of doing it. Hence the BW is the first suspect.
5870 shows its potential at the top resolution with the fastest i7. Dual core setups are not good enough.

Does quad core benefit in every game? Not really at this point. Maybe many more games next year will be able to utilize as many cores as a system has. That would be excellent.
So, is it safe to say that the 5870 ranges anywhere from 18 to 40% faster than a GTX285 depending on game and res?
Or is this not satisfactory..

I believe it does. If a game uses two threads it has to compete with OS and drivers as well.
Drivers are multithreaded and that's why Quad is better than Dual core on any given game unless there is an extreme clock advantage towards the Dual core.
If somebody wants to test and compare GPUs it has to be with the best and fastest CPU available.

 

[Keysplayr]

Originally posted by: dguy6789

Originally posted by: Keysplayr

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo
5870 delivers 2.6x4890 in CoH at the top resolution and settings.

Like i said, i know the results.
In this particular case the 2,6X figure is for 2560X1600 with 8X AA & 16XAF.
If you drop the antialiasing at 4X the difference at 2560X1600 16XAF is 1,95X.
And the DX10 codepath of CoH is extremely pixel shader oriented.
All the above are predictable and natural results.

I am not here to state the obvious.
I am trying to figure out what are the limitations of the Cypress because it shows according to my perseption strange results.
My analysis is complex and to write down it will take me more time than it took to figure out the situation.
Sorry i am not willing to lose time.
If you disagree with me that the 5870 has (except the obvious driver and bandwidth related issues) the following issues:

Originally posted by: MODEL3
1.Geometry/vertex performance issues (in the sense that the classic vertex unit cannot generate 2X geometry in relation with 4890) (my main assumption)

or/and

2.Geometry/vertex shading performance issues (in the sense that the geometry shader [GS] cannot achieve 2X speed in relation with 4890)(another possible assumption)

My assumption is that ATI though that many of the future DX11 games will use vertex techniques based on the tesselator unit.

Then i respect your opinion, but i am not willing to do anything about it.

I just thought to point out what the reviewers with all their knowledge,year in this business, technical background, briefings from ATI, 5870 cards at hand, failed to point out.

Sure i don't have technical background, nor i was at ATI briefings and of cource i don't have a 5870 card in order to be able to test my theory.
So with all these disadvatntages i may be wrong.
The funny thing will be, to be right about it.
And no, it will be no luck at all, trust me.

What part of silicon performs geometry? Is it doubled from 4890 to 5870 or not?
Geometry is not an issue. The BW is the first in line.

Certainly BW in general is first in line, when we examine the potential reasons.
I said that a 800-850MHz 32ROPS/1600SP design with 256bit memory controller (with the current GDDR5 ICs) will be bandwidth limited before 1,5 freakin month.
This is extremely easy to figure out.
But this kind of performance behavior isn't only BW related.
According to my perception Geometry is an issue, so i guess we have to agree to disagree about this matter.

Geometry is done in SP's. They were doubled. It's not an issue unless SP's are starved at the point of doing it. Hence the BW is the first suspect.
5870 shows its potential at the top resolution with the fastest i7. Dual core setups are not good enough.

Does quad core benefit in every game? Not really at this point. Maybe many more games next year will be able to utilize as many cores as a system has. That would be excellent.
So, is it safe to say that the 5870 ranges anywhere from 18 to 59% faster than a GTX285 depending on game and res?
Or is this not satisfactory..

Fixed and I'll agree to that. (World of Warcraft, Anandtech Article)

What about HAWX then? 4xAA 19x12 the 5870 is only 10% faster? That lead grows when we move up to 25x16 though. But if we include all benches, we need to include all benches. Even the ones showing 5870 scaling to the stratosphere.

 

[Qbah]

Originally posted by: Keysplayr
What about HAWX then? 4xAA 19x12 the 5870 is only 10% faster? That lead grows when we move up to 25x16 though. But if we include all benches, we need to include all benches. Even the ones showing 5870 scaling to the stratosphere.

Well, the HD5870 is 59% faster in NFS:Shift too, at 2560x1600 according to Xbitlabs. Hell, it's 80% faster than a GTX295 in the same test. That's why you really don't take the extreme results They skew the picture. 40% looks about right :thumbsup: +/- 10% depending on game?

 

[Tempered81]

Originally posted by: Keysplayr
if we include all benches, we need to include all benches. Even the ones showing 5870 scaling to the stratosphere.

Well you can find benches where a 285 is faster than a 5870, and you can find some where the 5870 is 240% faster than a 285... So that doesn't tell us much about averages, only about extremes.
On average i'd say %20 for 1680 no AA, 30% for 1920 4xAA, and 45% for 1920 or greater with 8xAA. So overall ~30-35%.

 

[SlowSpyder]

Originally posted by: Keysplayr

Yes, we'll have to see what benefits DX11 has in store for multicore setups. It would be great if it could utilize all the power.
No, the 5870 surely wasn't meant to compete with a GTX285. That 100% over GTX285 isn't exactly common either. I think that across most review sites, that 40% performance improvement is generous at this point. Sometimes more, sometimes less.

And I know a lot of people who spent well north of 300.00 on a graphics card who had only 22" monitors. Crysis was the main culprit here. Believe it or not.

Like I said, 'generally' people who buy cheap monitors aren't looking for expensive graphics cards... at least that would be my belief.

And yes, that 100%+ performance over the GTX285 is an extreme example, in no way was it meant to sound like I feel that's an average. But someone is back and saying things like 18% is an average, I just don't buy that and I think you know that's false too. Maybe if we look at 1440 or 1680 res it's around 20% faster (as the CPU would be limiting more often then not I would think) but clearly the 5870 is much faster then a ~20% advantage at the resoutions it'll likely be used at (1080P, 1920x1200, etc).

 

[MODEL3]

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo

Originally posted by: MODEL3

Originally posted by: Janooo
5870 delivers 2.6x4890 in CoH at the top resolution and settings.

Like i said, i know the results.
In this particular case the 2,6X figure is for 2560X1600 with 8X AA & 16XAF.
If you drop the antialiasing at 4X the difference at 2560X1600 16XAF is 1,95X.
And the DX10 codepath of CoH is extremely pixel shader oriented.
All the above are predictable and natural results.

I am not here to state the obvious.
I am trying to figure out what are the limitations of the Cypress because it shows according to my perseption strange results.
My analysis is complex and to write down it will take me more time than it took to figure out the situation.
Sorry i am not willing to lose time.
If you disagree with me that the 5870 has (except the obvious driver and bandwidth related issues) the following issues:

Originally posted by: MODEL3
1.Geometry/vertex performance issues (in the sense that the classic vertex unit cannot generate 2X geometry in relation with 4890) (my main assumption)

or/and

2.Geometry/vertex shading performance issues (in the sense that the geometry shader [GS] cannot achieve 2X speed in relation with 4890)(another possible assumption)

My assumption is that ATI though that many of the future DX11 games will use vertex techniques based on the tesselator unit.

Then i respect your opinion, but i am not willing to do anything about it.

I just thought to point out what the reviewers with all their knowledge,year in this business, technical background, briefings from ATI, 5870 cards at hand, failed to point out.

Sure i don't have technical background, nor i was at ATI briefings and of cource i don't have a 5870 card in order to be able to test my theory.
So with all these disadvatntages i may be wrong.
The funny thing will be, to be right about it.
And no, it will be no luck at all, trust me.

What part of silicon performs geometry? Is it doubled from 4890 to 5870 or not?
Geometry is not an issue. The BW is the first in line.

Certainly BW in general is first in line, when we examine the potential reasons.
I said that a 800-850MHz 32ROPS/1600SP design with 256bit memory controller (with the current GDDR5 ICs) will be bandwidth limited before 1,5 freakin month.
This is extremely easy to figure out.
But this kind of performance behavior isn't only BW related.
According to my perception Geometry is an issue, so i guess we have to agree to disagree about this matter.

Geometry is done in SP's. They were doubled. It's not an issue unless SP's are starved at the point of doing it. Hence the BW is the first suspect.
5870 shows its potential at the top resolution with the fastest i7. Dual core setups are not good enough.

Sadly,no.it's much more complex than that.

Like I told you i am not talking about BW which is OBVIOUS.

I will post later some geometry synthetic tests that i found.
And i will search for more.
I have to post it first in the comment section at Ryan's review.

LOL

 

[MrK6]

Having used both, the 5870 is equivalent to the GTX295. However, they have markedly different gameplay quality, so to extend the comparison beyond that isn't appropriate.

 

[Qbah]

Originally posted by: SlowSpyder

Originally posted by: Keysplayr

Yes, we'll have to see what benefits DX11 has in store for multicore setups. It would be great if it could utilize all the power.
No, the 5870 surely wasn't meant to compete with a GTX285. That 100% over GTX285 isn't exactly common either. I think that across most review sites, that 40% performance improvement is generous at this point. Sometimes more, sometimes less.

And I know a lot of people who spent well north of 300.00 on a graphics card who had only 22" monitors. Crysis was the main culprit here. Believe it or not.

Like I said, 'generally' people who buy cheap monitors aren't looking for expensive graphics cards... at least that would be my belief.

And yes, that 100%+ performance over the GTX285 is an extreme example, in no way was it meant to sound like I feel that's an average. But someone is back and saying things like 18% is an average, I just don't buy that and I think you know that's false too. Maybe if we look at 1440 or 1680 res it's around 20% faster (as the CPU would be limiting more often then not I would think) but clearly the 5870 is much faster then a ~20% advantage at the resoutions it'll likely be used at (1080P, 1920x1200, etc).

Also, that graph linked earlier shows

HD5870 = 100
GTX285 = 81.6

Hence, based on that silly chart, the HD5870 is ~22.5% faster across all the resolutions and settings, from the low 1024x768 to 2560x1600 in games that both are heavily CPU and GPU limited (lack of basic logic here - something faster than = the slower is the base). Also, according to the same chart, the GTX295 is 35% faster "on average" than a GTX285... yeah right :disgust:

EDIT: Btw. the same chart shows that GTX275 SLI is 20% faster than a single GTX275...

 

[SSChevy2001]

Originally posted by: Keysplayr
What about HAWX then? 4xAA 19x12 the 5870 is only 10% faster? That lead grows when we move up to 25x16 though. But if we include all benches, we need to include all benches. Even the ones showing 5870 scaling to the stratosphere.

HAWX seems to be a strange case where some review sites seem to refuse to use DX10.1 from the looks at the scores, but when it's on ATi should have a nice advantage.

http://www.guru3d.com/article/...hd-5870-review-test/18
1920x1080 4xAA the 5870 is 40% faster and at 2560x1600 55.6% faster.

http://www.firingsquad.com/har...nce_preview/page20.asp
1920x1200 4xAA the 5870 37% faster and at 8xAA 86% faster.

The lead grows just by using 8xAA, which is the point of having such a high end card.

 

[Idontcare]

Originally posted by: dguy6789
The limitation on the 5870 is most certainly memory bandwidth. Memory bandwidth is the only category in which the 5870 does not outclass the 4870x2 in.

Radeon HD 5870:

Pixel Fill Rate: 27200 MPixels/sec
Texture Fill Rate: 68000 MTexels/sec
Flops: 2720 GFLOPS
Memory Bandwidth: 153.6GB/sec

Radeon HD 4870x2:

Pixel Fill Rate: 24000 MPixels/sec
Texture Fill Rate: 60000 MTexels/sec
Flops: 2400 GFLOPS
Memory Bandwidth: 230.4GB/sec

I would think this theory would be fairly easy to test and either prove or disprove by anyone who has the 5870 in hand.

Just underclock the memory. Do stock, underlocked mem by 10-15%, underclocked mem by 20-30%, and plot the benchmark data.

If you get a linear line with nice R^2 and the slope is nice and large (slope of 1 on a % bandwidth delta versus % fps delta would be perfect) then you can easily conclude the performance would be improved by having higher bandwidth above stock.

What is interesting is that no one seems to have done this test in their reviews yet (or did I miss it?).

What would be even more interesting is if the data proved out this theory as surely AMD would have tested this and determined whether or not bandwidth was really going to be the limitation of the 5870 well before the launch. So why would they artificially handicap the 5870 by knowingly starving it of bandwidth? I don't think we should assume ignorance on their part here.