New ATi 4870 vs. Nvidia GTX 280

This topic is locked from further discussion.

Avatar image for pknyo
pknyo

214

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#1 pknyo
Member since 2007 • 214 Posts
I know its early, but next month will see the launch of the next gen cards from both Nvidia and Ati. I admit I`m a Nvidia man, but those new ATi specs do look pretty mean, ATi has faster clocks and more SPs but Nvidia has double the memory bandwith and built in Physx. ATi has had faster clocks and Sps this gen as well and nvdia still outpreformed though.
Avatar image for LordEC911
LordEC911

9972

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#2 LordEC911
Member since 2004 • 9972 Posts

I know its early, but next month will see the launch of the next gen cards from both Nvidia and Ati. I admit I`m a Nvidia man, but those new ATi specs do look pretty mean, ATi has faster clocks and more SPs but Nvidia has double the memory bandwith and built in Physx. ATi has had faster clocks and Sps this gen as well and nvdia still outpreformed though.pknyo

How is 122Gbps vs 141Gbps double the bandwidth?
It should be an interesting generation...

Avatar image for Daytona_178
Daytona_178

14962

Forum Posts

0

Wiki Points

0

Followers

Reviews: 4

User Lists: 0

#3 Daytona_178
Member since 2005 • 14962 Posts
Just to let you know that the NVidia 8 & 9 series are capable of physics aswel!
Avatar image for bakalhau90
bakalhau90

2264

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#4 bakalhau90
Member since 2004 • 2264 Posts
How much new gen are these? What's the difference between the 9-series and this one compared to the upgrade that was the 8xxx's to 9xxx's (and ATI's equivalent)?
Avatar image for teddyrob
teddyrob

4557

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#5 teddyrob
Member since 2004 • 4557 Posts

I know its early, but next month will see the launch of the next gen cards from both Nvidia and Ati. I admit I`m a Nvidia man, but those new ATi specs do look pretty mean, ATi has faster clocks and more SPs but Nvidia has double the memory bandwith and built in Physx. ATi has had faster clocks and Sps this gen as well and nvdia still outpreformed though.pknyo

ATI shader clocks are slower.

Avatar image for Beaglesniffer
Beaglesniffer

707

Forum Posts

0

Wiki Points

0

Followers

Reviews: 8

User Lists: 0

#6 Beaglesniffer
Member since 2006 • 707 Posts

ATI shader clocks are the same as there core clocks thats whats limiting them

theres obviosuly a problem that limits them from upping them seperateky otherwise they would have done it by now and if they did there cards would fly ahead.

ATI and nvdidia both handle physics processing but why do we want are GPU's doing physics processing, we want them to concentrate on processing graphics not physics. and most people have powerful CPU's which are generally not put to good use in games they really need to concentrate with physics on CPU's and visuals with GPU's. why make things more complicated, although i did hear of some nvidia physics running better than the new 8 core from intel or something??? cant remeber exactly but if its true then WTF! lol

Avatar image for teddyrob
teddyrob

4557

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#8 teddyrob
Member since 2004 • 4557 Posts

ATI shader clocks are the same as there core clocks thats whats limiting them

theres obviosuly a problem that limits them from upping them seperateky otherwise they would have done it by now and if they did there cards would fly ahead.

Beaglesniffer

The info on the 4870 is that clock speeds are - 850MHz/1050Mhz/1940MHz core/shader/memory. So they seem to have achieved that now but it is still way slower that Nvidia shader clocks who are pushing their shader clocks way higher. Hopefully it won't matter too much on this card.

Avatar image for Nikodemus87
Nikodemus87

27

Forum Posts

0

Wiki Points

0

Followers

Reviews: 3

User Lists: 0

#9 Nikodemus87
Member since 2007 • 27 Posts

I find it distrurbing that "our" GPUs should handle physics, sure physics is "cool" but they(the GPUs) should be used for graphics; not calculating if that barrel or box will hit the Spy´s head in Team Fortress 3

:)

Avatar image for darkmagician06
darkmagician06

6060

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#10 darkmagician06
Member since 2003 • 6060 Posts
How much new gen are these? What's the difference between the 9-series and this one compared to the upgrade that was the 8xxx's to 9xxx's (and ATI's equivalent)?
bakalhau90
well...i know that the gtx 260/280 will use the new gt200 chip...instead of the g80 or whatever they are using now
Avatar image for Alejo17
Alejo17

690

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#11 Alejo17
Member since 2005 • 690 Posts
[QUOTE="Beaglesniffer"]

ATI shader clocks are the same as there core clocks thats whats limiting them

theres obviosuly a problem that limits them from upping them seperateky otherwise they would have done it by now and if they did there cards would fly ahead.

teddyrob

The info on the 4870 is that clock speeds are - 850MHz/1050Mhz/1940MHz core/shader/memory. So they seem to have achieved that now but it is still way slower that Nvidia shader clocks who are pushing their shader clocks way higher. Hopefully it won't matter too much on this card.

Aint the 4xxx series clock speeds higher than nvidia ones??

Avatar image for Beaglesniffer
Beaglesniffer

707

Forum Posts

0

Wiki Points

0

Followers

Reviews: 8

User Lists: 0

#12 Beaglesniffer
Member since 2006 • 707 Posts

I find it distrurbing that "our" GPUs should handle physics, sure physics is "cool" but they(the GPUs) should be used for graphics; not calculating if that barrel or box will hit the Spy´s head in Team Fortress 3

:)

Nikodemus87

thats what i was saying we have these powerful CPU's but there not really put to good use to be fair

Avatar image for Beaglesniffer
Beaglesniffer

707

Forum Posts

0

Wiki Points

0

Followers

Reviews: 8

User Lists: 0

#13 Beaglesniffer
Member since 2006 • 707 Posts
[QUOTE="Beaglesniffer"]

ATI shader clocks are the same as there core clocks thats whats limiting them

theres obviosuly a problem that limits them from upping them seperateky otherwise they would have done it by now and if they did there cards would fly ahead.

teddyrob

The info on the 4870 is that clock speeds are - 850MHz/1050Mhz/1940MHz core/shader/memory. So they seem to have achieved that now but it is still way slower that Nvidia shader clocks who are pushing their shader clocks way higher. Hopefully it won't matter too much on this card.

really finally lol

i would expect that even though the shader clocks may be slower than nvidia the fact that the core clocks and amount of SP's are generally higher it should balance it all out

looks like the next cards from both of them will be very close in terms of performance

Avatar image for Led___Zeppelin
Led___Zeppelin

1581

Forum Posts

0

Wiki Points

0

Followers

Reviews: 6

User Lists: 0

#14 Led___Zeppelin
Member since 2004 • 1581 Posts

Hey guys, if im planning on buying a new laptop soon, so should i wait for these new cards to come out? I mean my laptop is only gonna be in the 1000-1500 range but are the new cards gonna be in cheap laptops like that too or are the manufacturers going to update their laptops into better but last gen graphics card or will there be no change at all?

Avatar image for Beaglesniffer
Beaglesniffer

707

Forum Posts

0

Wiki Points

0

Followers

Reviews: 8

User Lists: 0

#15 Beaglesniffer
Member since 2006 • 707 Posts

Hey guys, if im planning on buying a new laptop soon, so should i wait for these new cards to come out? I mean my laptop is only gonna be in the 1000-1500 range but are the new cards gonna be in cheap laptops like that too or are the manufacturers going to update their laptops into better but last gen graphics card or will there be no change at all?

Led___Zeppelin

of course laptops are gonna keep updating and becoming better but they will always be a step behind PC bcoz of the power, heat and space limits.

waiting another 6 months could up your laptop performance considerably it just depends what happens in those 6 months. im not sure if they are releasing these into laptop versiosn im sure som1else could tell you

Avatar image for Led___Zeppelin
Led___Zeppelin

1581

Forum Posts

0

Wiki Points

0

Followers

Reviews: 6

User Lists: 0

#16 Led___Zeppelin
Member since 2004 • 1581 Posts
So you're saying that this is gonna have no effect on laptops for at least a few months? if thats the case i'll just buy the one i want now then. Although it will just be a Geforce 8600... :(
Avatar image for Beaglesniffer
Beaglesniffer

707

Forum Posts

0

Wiki Points

0

Followers

Reviews: 8

User Lists: 0

#17 Beaglesniffer
Member since 2006 • 707 Posts

well it wont have an effect tillt hey come out thats if there makign laptop versions they will also be more expensive soo......

a 8600 is pretty good going for a laptop

Avatar image for imprezawrx500
imprezawrx500

19187

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#18 imprezawrx500
Member since 2004 • 19187 Posts
[QUOTE="Beaglesniffer"]

ATI shader clocks are the same as there core clocks thats whats limiting them

theres obviosuly a problem that limits them from upping them seperateky otherwise they would have done it by now and if they did there cards would fly ahead.

teddyrob

The info on the 4870 is that clock speeds are - 850MHz/1050Mhz/1940MHz core/shader/memory. So they seem to have achieved that now but it is still way slower that Nvidia shader clocks who are pushing their shader clocks way higher. Hopefully it won't matter too much on this card.

have they fixed the lack of texture units?

Avatar image for HuusAsking
HuusAsking

15270

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#19 HuusAsking
Member since 2006 • 15270 Posts

ATI shader clocks are the same as there core clocks thats whats limiting them

theres obviosuly a problem that limits them from upping them seperateky otherwise they would have done it by now and if they did there cards would fly ahead.

ATI and nvdidia both handle physics processing but why do we want are GPU's doing physics processing, we want them to concentrate on processing graphics not physics. and most people have powerful CPU's which are generally not put to good use in games they really need to concentrate with physics on CPU's and visuals with GPU's. why make things more complicated, although i did hear of some nvidia physics running better than the new 8 core from intel or something??? cant remeber exactly but if its true then WTF! lol

Beaglesniffer
Because as games get more complicated, the CPU will have its hands full with just AI, environmental interaction and pure resource management. A recent test of Crysis showed that as your GPU finally becomes able to handle Crysis' effects full-on at high resolution, the CPU rapidly becomes more and more the bottleneck. And physics calculations are pretty demanding. nVidia didn't buy out Aegia for no reason. Maybe not soon, but perhaps down the road we'll see nVidia boards that have dedicated physics units--perhaps built into the GPU as a separate logic unit, perhaps on a separate chip on the graphics card PCB, but it'll be there eventually because nVidia itself (who helped contribute to the Crysis stress test) knows there will still be a need for CPU offloading.
Avatar image for darkmagician06
darkmagician06

6060

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#20 darkmagician06
Member since 2003 • 6060 Posts

Hey guys, if im planning on buying a new laptop soon, so should i wait for these new cards to come out? I mean my laptop is only gonna be in the 1000-1500 range but are the new cards gonna be in cheap laptops like that too or are the manufacturers going to update their laptops into better but last gen graphics card or will there be no change at all?

Led___Zeppelin
the laptop versions of these cards wont be out for a while...
Avatar image for Mam00th
Mam00th

432

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#21 Mam00th
Member since 2005 • 432 Posts
[QUOTE="Beaglesniffer"]

ATI shader clocks are the same as there core clocks thats whats limiting them

theres obviosuly a problem that limits them from upping them seperateky otherwise they would have done it by now and if they did there cards would fly ahead.

ATI and nvdidia both handle physics processing but why do we want are GPU's doing physics processing, we want them to concentrate on processing graphics not physics. and most people have powerful CPU's which are generally not put to good use in games they really need to concentrate with physics on CPU's and visuals with GPU's. why make things more complicated, although i did hear of some nvidia physics running better than the new 8 core from intel or something??? cant remeber exactly but if its true then WTF! lol

HuusAsking

Because as games get more complicated, the CPU will have its hands full with just AI, environmental interaction and pure resource management. A recent test of Crysis showed that as your GPU finally becomes able to handle Crysis' effects full-on at high resolution, the CPU rapidly becomes more and more the bottleneck. And physics calculations are pretty demanding. nVidia didn't buy out Aegia for no reason. Maybe not soon, but perhaps down the road we'll see nVidia boards that have dedicated physics units--perhaps built into the GPU as a separate logic unit, perhaps on a separate chip on the graphics card PCB, but it'll be there eventually because nVidia itself (who helped contribute to the Crysis stress test) knows there will still be a need for CPU offloading.

First of all, the CPU is not becoming a bottleneck anytime soon, an old core2duo @ 2,66ghz can perform just as well in almost any game then your brand new E8400, because it is no where near 100% or even 80% maximum load.

As for GPU running physics better then an 8 core intel CPU, that is entirely true. Physic processing require a SIMD processor (single instruction multiple data) like graphic processing does. You can do that by putting many many cores (or SP) on one card with a limited number of instruction and then perform massive parallel computation with a limited number of instructions because the result of your calculation does not impact your next one. Intel CPU use a MIMD (multiple instruction mutiple data) processor (with some SIMD elements though but not much) to do most of your day to day processing because MIMD processor are more apt to compute a random set of instruction then a SIMD processor.

That is why the cell processor in the PS3 is wonderful at physic calculation, because it has 8 SPE which are basicly SIMD processors to physic calculation and 1 big MIMD processor for other task. But it's one hell of a job to program for this beast.

Avatar image for codezer0
codezer0

15898

Forum Posts

0

Wiki Points

0

Followers

Reviews: 44

User Lists: 0

#22 codezer0
Member since 2004 • 15898 Posts

I find it distrurbing that "our" GPUs should handle physics, sure physics is "cool" but they(the GPUs) should be used for graphics; not calculating if that barrel or box will hit the Spy´s head in Team Fortress 3

:)

Nikodemus87
Well since none of you all would buy the PhysX part, there needs to be another way of accelerating physics with a hardware solution. CPU's are great general-purpose devices, but they would simply get bogged down trying to handle (reasonably) accurate physics anywhere near what the PPU did. The only other architecture that is close enough to be able to accelerate physics processing with the speed and breadth that will be needed in the future is the GPU.
Avatar image for HuusAsking
HuusAsking

15270

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#23 HuusAsking
Member since 2006 • 15270 Posts

First of all, the CPU is not becoming a bottleneck anytime soon, an old core2duo @ 2,66ghz can perform just as well in almost any game then your brand new E8400, because it is no where near 100% or even 80% maximum load.

Mam00th
It will become a bottleneck if you actually try to max out Crysis on very high detail at very high resolution--for example, if you throw in a tri-SLI 9800GTX rig. Which means as the GPUs get faster and better able to handle these demanding games, the more the CPU finds itself no longer waiting for the GPU to do its thing, and since there are a ton of functions that can only be handled by the CPU (I/O, AI, environment manipulation, resource management, the OS, etc.), its time will always be divided. So, if the GPU is able to take on the physics calculations more as it grows faster, the CPU won't feel the strain nearly so much.
Avatar image for prolink47
prolink47

467

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#24 prolink47
Member since 2004 • 467 Posts
there was a bug in the 3870 series and thats why it didnt scale up to the 8800 so i guess they fixed that in the 4 series
Avatar image for slickchris7777
slickchris7777

1610

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#25 slickchris7777
Member since 2005 • 1610 Posts
[QUOTE="Led___Zeppelin"]

Hey guys, if im planning on buying a new laptop soon, so should i wait for these new cards to come out? I mean my laptop is only gonna be in the 1000-1500 range but are the new cards gonna be in cheap laptops like that too or are the manufacturers going to update their laptops into better but last gen graphics card or will there be no change at all?

Beaglesniffer

of course laptops are gonna keep updating and becoming better but they will always be a step behind PC bcoz of the power, heat and space limits.

waiting another 6 months could up your laptop performance considerably it just depends what happens in those 6 months. im not sure if they are releasing these into laptop versiosn im sure som1else could tell you

Within the next few months, Nviida will be launching thier mobile 9 series, and ATI will be launching thier mobile 3000 series.

Avatar image for dalopolo
dalopolo

25

Forum Posts

0

Wiki Points

0

Followers

Reviews: 1

User Lists: 0

#26 dalopolo
Member since 2007 • 25 Posts
I'll just get the best performing card
Avatar image for DISABLED_K3Y
DISABLED_K3Y

54

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#27 DISABLED_K3Y
Member since 2007 • 54 Posts
where did you find this information?
Avatar image for cge10
cge10

25

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#28 cge10
Member since 2008 • 25 Posts

i got 2600 xt now im willing to spend 350 on gpu.

my choices is 2x 3870 iceq3 or a 9800 gtx on a maximus formula se

Avatar image for GuitarFreak2
GuitarFreak2

670

Forum Posts

0

Wiki Points

0

Followers

Reviews: 0

User Lists: 0

#29 GuitarFreak2
Member since 2006 • 670 Posts
Just so you know, the 4870 is supposed to be near the 3870X2 performance wise. And seeing as the 8800GTX has similar performance to the 3870X2, it's not going to come near the GTX 280. The 280's shaders are supposed to be ~50% faster than on the 9 series, which means it'll be quite a bit faster than the ATI 4xxx's.