Can't decide on what card to buy!!

[sleepdeprived82]

They make a x1600 pro for AGP

BUT

it would never outperform the X800gto

x1600 pro has 12 pipes, 5 vertex 128bit bus and 12.8gbps bandwith 500/800 core/mem DDR2

x800gto 12 pipes, 6 vertex 256 bit bus and 31.4gbs bandwith 400/980 core/mem DDR3

The gto is a fair bit better and will easily oc to the 500 core speed of the x1600.

 

[unbiased4u]

How about this card?

http://www.ebuyer.com/customer/products/index.html?action=c2hvd19wcm9kdWN0X292ZXJ2aWV3&product_uid=107115

I'm sure that's worth the £125 price tag.

It'd be nice if someone gave their view 'cause i'm considering buying one of those 😛

It's pretty good...512MB GDDR RAM is overkill though...

 

[cleeve]

x1600 pro has 12 pipes, 5 vertex 128bit bus and 12.8gbps bandwith 500/800 core/mem DDR2

Actually, the X1600 has 12 pixel processors and 12 pixel shaders, but they work in banks of 3... so it can only output 4 pixels per clock.

It's more like a 4-pipeline architecture than a 12-pipe. That's why it performs so inconsistantly.

 

[sleepdeprived82]

Nope that card its awful

Dont touch the 6800 512mb it has been crippled with a 128bit interface and 512mb of slow ram. The extra ram is just marketing.

A 6600gt would outperform that card

 

[sleepdeprived82]

Ok confused the hell out of me but I believe you, its still a not as good a the GTO

 

[TheGreatGrapeApe]

Forgive me if this was mentioned by you or someone else...

Sorta mentioned as a caveat, in passing.... "Just be sure that the X800GTO isn't crippled in other ways with slow memory and slow clocks. "

But you're fogiven... this time! :twisted:

 

[TheGreatGrapeApe]

Ok confused the hell out of me but I believe you, its still a not as good a the GTO

Good view of the X1600 architecture and how it's layed out;
http://www.beyond3d.com/reviews/ati/rv5xx/

Think of it like this, when complex shaders requiring multiple passes or multiple source targets are required then the X1600 does better than a pure 4 pipeline design (like the R9600/X600) because they have to go through the whole pipleline 2-3 times for each pixel whereas it can stagger the load and essentially do it in 3 times the speed, but if the pixel shader load is light then the max it can still output is the same as the simpler design because both are still limited by having 4 ROPs. They also both have only 4 texture units.

So under easiest conditions it'd never perform better than a regular 4 pipeline design, but under the most gruelling pixel processing situation theoretically it could come close to matching a 12 pipeline card that needed to do 3 transformations/clocks for each pixel operation.

Realistically though it currently performs closer to a slightly underperforming 8 pipeline design, but in the future the performance results should look closer to an overperforming 8 pipe design.

 

[sleepdeprived82]

Ok Understand now, explains its unpredictabiliy

Cheers