Hd5870 - $299

[w0mbat]

1.6 times faster than RV770 is 160% faster, 60% would be 1.6 times as fast as RV770.

[Xoulz]

his math is wrong, any of us idiots know, the shirnk to 40nm gave them 60% more for the same area, and this chip is a good 40% bigger than the 770. and my math is so uber i know 40% + 60% = 100% (yes this was half a joke, but if ATI gives us just a 60% increase for being so much better and on a smaller process, they fail big time)

How can a 60% gain be a fail..?

Specially when the 4800 series is roughly 1 year old. When have you've ever been able to leap through such power over a 17 month period?

[AKM]

How can a 60% gain be a fail..?

Specially when the 4800 series is roughly 1 year old. When have you've ever been able to leap through such power over a 17 month period?

If NV manage to get more than 60% perf. improvement out of their new card then a 60% gain for AMD will be a failure.

[LowRun]

If NV manage to get more than 60% perf. improvement out of their new card then a 60% gain for AMD will be a failure.

Can't see how this makes any sense

[Manicdan]

How can a 60% gain be a fail..?

Specially when the 4800 series is roughly 1 year old. When have you've ever been able to leap through such power over a 17 month period?

its chip size is bigger than the r770, and its on a smaller process, even if its the same exact thing just scaled it should still be 80-120% faster due to how much stuff they can pack into that 300mm2+ chip.

this site has some very basic comparison info for the the 3870 to 4870 jump (just check out the table at the bottom)
http://www.cdrinfo.com/sections/news...x?NewsId=23573
and i like it cause it shows how the size of the 4870 is only 35% bigger, yet offers about double the performance. thats a good jump and hats off to ATI.

[SimBy]

I wonder why anyone takes Fuad seriously

[tajoh111]

If NV manage to get more than 60% perf. improvement out of their new card then a 60% gain for AMD will be a failure.

This is true. Becasuse if NV gets only 60 percent more performance, but still has a giant chip. NV will for money, still be the loser this round.

Wombat, I credit faster 1.60x more to bad grammer than anything. If it was indeed 160% percent faster, the price would certainly be higher and the teraflops wold be higher(3+ teraflops).

Conversely if what is true in the article verbatim, than if 1.6 is how much faster it is in the best case scenario, this could be at 2500*1600 with 8aa resolution when a 512 meg card would be starved for memory.

60% percent is perfectly acceptable for the price of admission. At $299, it blows everything out of the water in the high end for price for performance.

The chips size estimates everyone is doing is based on pictures estimates.

Also if the chip is as large as everyone is saying and AMD is making this a hard lauch so soon. It could be that alot of the shaders are deactivated because yields at 40nm are bad. If this was the case, AMD would have another hand to play almost immedietly when the gt300 came out.

As we have seen between g80 and gt200, a linear jump in tflop performance does not equal a linear jump in gaming performance.

[bill_d]

I wonder why anyone takes Fuad seriously

true

this can be anything

We've heard that ATI's soon to lunch DirectX 11 performance card, something that we know as RV870 and something that AMD calls Evergreen, should be roughly 1.6 times faster than the RV770 chip

but what card, rv870 and evergreen is the whole family of cards
only one they have shown so far is juniper
so is this juniper, or 5850 or 5870

fud throws out a speed but no facts on what

it's like saying it will rain next week but not where

[marten_larsson]

1.6 times faster than RV770 is 160% faster, 60% would be 1.6 times as fast as RV770.

Nah, if something is twice (two times, 2x) as fast it's a 100% increase. 1.6 times as fast is 160% of original speed. Still Fudzilla isn't exactly the greatest source and more than 2TFLOPs for the 5850 would probably exceed 1.6 times as fast.

[ripken204]

its 160% of the original speed and 60% faster than the original speed

[w0mbat]

Nah, if something is twice (two times, 2x) as fast it's a 100% increase. 1.6 times as fast is 160% of original speed. Still Fudzilla isn't exactly the greatest source and more than 2TFLOPs for the 5850 would probably exceed 1.6 times as fast.

No, "1.6 times as fast as rv770" is 160% of the original speed and thast 60% faster. But FUD wrote "1.6 times faster than rv770" and thats 260% of the original speed and thus 160% faster. Im cant say what he ment, but he wrote 160% faster for shure.

its 160% of the original speed and 60% faster than the original speed

No, u got it wrong, atleast what he wrote is different. Look it up, 1 times faster = 2 times as fast.


Just read the B3D r8xx fred, many ppl think like u 2 but its wrong. Not that i think Fudo is right with that, maybe he ment 60% faster and wrote 160%, but thats not the point.

[LordEC911]

The chips size estimates everyone is doing is based on pictures estimates.

Which is a pretty good way to estimate, since they were very close the last couple of times.

Also if the chip is as large as everyone is saying and AMD is making this a hard lauch so soon. It could be that alot of the shaders are deactivated because yields at 40nm are bad. If this was the case, AMD would have another hand to play almost immedietly when the gt300 came out.

Yields are better than ~60%, which compared to early RV740 was around 20%.

[.:Nismo]

It's not that hard to understand. FUD wrote that it is 1.6 times as fast. Just think of it like this:

100 X 1.6 = 160 <---- It increased by 60%. You keep the same performance that the rv770 (that's the 1 in 1.6) and you increase it by 60% (that would be the 6 in 1.6)

[w0mbat]

I perfectly understand and i, too, think that Fudo wanted to write 60% faster than rv770. But its a fact that he wrote the following:

RV870 is 1.6 times faster than RV770

and that means 160% faster than rv770.

Again: 1.6 times faster = 160% faster and 1.6 times as fast as = 60% faster. Its that easy

[TheCarLessDriven]

Do we have a god damn release date yet?? Is it september for sure? Early or late september????

[w0mbat]

22nd

[TheCarLessDriven]

For sure? Was this released by ATi? I havent read it anywhere.

[zalbard]

22nd

Confirmed? Link / source?

[TheCarLessDriven]

if im building a new rig to sell (i7 920, ssd, saphire 4890 1gb) do you think i could still get away with selling it even with the 5890's coming out inside the next month (Although i am doubting they will release in september, more likely october)

[flopper]

For sure? Was this released by ATi? I havent read it anywhere.

Press 10sep and some chibang intro with demos, specifications etc...
nda and such a week or two so after, I seen 24sep as a date.
Now, after 10sep we will most likely see a few leaks going on and get out the information in how the cards perform.

In the shop, after the 24?
with greater availibility in october before windows7.

[BrowncoatGR]

w0mbat is usually spot on with his info

[LordEC911]

w0mbat is usually spot on with his info

Yeah, he doesn't need a source/link.

[Calmatory]

AMD has done just amazing job keeping the stuff secret. Can't wait.

[w0mbat]

w0mbat is usually spot on with his info

Yeah, he doesn't need a source/link.

Waaaaaa, dont hunt me down if im wrong. I cant always be right

Im just writing so much stuff, some of it gotta turn out true...

[jaredpace]

I saw this post over at B3D. It makes the most sense to me -

Seriously who believes that in a real game benchmark typically used, RV790(OC) scores for example 50 FPS. Would you believe RV870 scores 130 FPS or rather 80 FPS.
The 160% faster does not match to any rumoured Vintage scores either, which were reported at around 16-18.000.

http://forum.beyond3d.com/showpost.p...postcount=2141

If COD4 is giving you 50 fps on your 4870 or 4890 in any particular benchmark, the HD5870 is not going to yield 130+ FPS, more like 80-100 FPS. However, going from 3870 --> 4870 almost doubled performance (40 FPS to 80 FPS).




3870 -----> 4870 -------Facts-------
320SP ----> 800SP-------- 2.5x
500 Gflop -> 1200 Gflop---- 2.4x
P5200-----> P10100 ------ 1.95x
39 FPS ---> 83 FPS ------- 2.15x

4870 -----> 5870 -------Rumors------
800SP ----> 1600SP------- 2.0x
1200Gflop -> 2400 Gflop --- 2.0x
P10100----> P18000 ------ 1.8x
50 FPS ----> 90 FPS ------ 1.8x -or-
50 FPS ----> 130 FPS ----- 2.6x

But they've also doubled the ROPs, so who knows what that improvement brings to the table?