• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD To Launch RV770 On June 18th

The gap between the GTX280 and GTX260 seems quite large by the going specs. In fact the gap between the GTX260 and the 9800 GTX is smaller which I can't see as a good thing. :\

Let's see here.

Memory bandwidth:

1. GTX280.
2. 4870
3. GTX260

Shader power:

1. 4870
2. GTX280
3. 4850

Texture fillrate:

1. GTX280
2. 9800 GTX
3. 8800 Ultra

Memory amount:

1&2: GTX280 & 4870
3: GTX260.

I hope you found that interesting.
 
Last edited:
By the way, ignore the ROP and TMU values for the nVidia cards on that table, they've swapped rows :D

EDIT: New corrected image

11132493rk8.jpg
 
Last edited:
kylew said:
How does FLOPS relate to actual real world performance? Is it a simple equasion, or is there a lot more to it?
Click to expand...

FLOPS = Floating-Point Operations per Second.

Basically, the theoretical maximum calculation capacity. Since modern games are shader-based, and shaders are (in essence) a series of add-multiply commands on floating point numbers, the total FPU capacity (measured in FLOPS) largely determines the capacity to render scenes. Virtually everything else within the card is to do with getting stored data to the GPU (eg. memory bandwidth), rendering the calculated image, or applying post-process effects.

The simple answer is that the performance of a card is its FPU capacity multiplier by an efficiency-multiplier.

For scientific computing applications, the FPU capacity is paramount, since other factors (such as rendering) are somewhat redundant. A GPU is a massively parallel architecture, highly specialised for parallel floating point computations. This is what makes is often over 200x faster than a high end CPU, but also what makes it more difficult to write general programs for. A CPU is designed for serial flexibilty, and a GPU as a parallel FPU powerhouse.



edit - for those that might not know, a floating point number is a number with a decimal place. "Whole numbers" are called integers, and are handled in a different way by the CPU. The vast majority of practical calculations use floating point numbers.
 
I find this slide interesting:

05rv770gb1.jpg


"TeraScale", wasn't that the name of Intel's research project on many-cored computing?

Edit: Oh wait, no it had a hyphen. "Tera-Scale" :p

I also find it interesting that the slide mentions AI, physics and raytracing calculations.
 
Last edited:
matt100 said:
is it normal for there to be no benchmarks 3 weeks before launch?
Click to expand...

Yep.

Since the whole x800 / 6800 fiasco, where both companies did huge paper launches with no real product availability and got universally slated for it, they both keep a tight lid on their products.

We won't see "official" benchmarks until release day, when we will actually be able to buy the cards. Sometimes some Taiwanese guy (usually someone on XS forums) manages to get hold of a card in the couple of weeks prior to release, but that's about it.

I'd rather have it this way than have papaer launches and no availablity again.
 
I never pay enough attention tbh but I do remember the X800XTPE (paper edition) fiasco :)

surely only the purest fanboy would buy on launch day in that case?

No idea of performance, no proper benchmarks, no idea on reliability, no waterblocks and hiked pricing.

A winning formula in my book!
 
matt100 said:
I never pay enough attention tbh but I do remember the X800XTPE (paper edition) fiasco :)

surely only the purest fanboy would buy on launch day in that case?

No idea of performance, no proper benchmarks, no idea on reliability, no waterblocks and hiked pricing.

A winning formula in my book!
Click to expand...

Meh. It only takes half an hour maximum to look at the reviews and benchmarks from the big, reliable sites (ie anandtech), get a feel for the performance over a wide range of programs, and decide if it's worth the cash.

I mean, you either have the cash and desire for a new card, or you don't. If you do then it doesn't take long to figure out which (if any) of the new cards are worth buying.
 
well some people say that 9800gtx will outperform 4870 so i was thinking what ati said so far :

RV770 family were present to confirm a Radeon HD 4870,4850, ATI by AMD for the past model named the two principles that can be high or low performance, the performance of two products beyond Radeon HD 3870X2
Click to expand...

so even 4850 will perform better then 3870x2 so the cards who perform better then 3870x2 might compete with 4850/4870 ? so just ask your self what cards are better then 3870x2?

i dont think 9800gtx is faster then 3870x2 correct me if im wrong.?

and if 4850 according of ati is faster then 3870x2 ,4870 should be even better. so i would really like to see benchmarks 9800gx2 vs 4870

I think its logical ? what do you think Azza ?
 
You must log in or register to reply here.
Back
Top Bottom