geforce 8800 poll.

rocks or not?

  • Total voters
    190
  • Poll closed .
It should rock, if the specs are even close to what they are rumored to be.
 
None of the above. I'll reserve judgement until I see the reviews, see the performance and new features.
 
I'm neutral until I see actual shipping cards get tested with publicly available drivers.
 
i'll reserve the judgement just the person above, as we don't want the 2 vid card company pull a 7900 or x1900 after a short release of the 7800 and x1800.
 
All we know about it is rumors...
Honestly, do we need another thread about this?
 
I will pass since it's the first technology with DX10 and there is bound to have bugs and issues. I will wait at least till the first refresh and see a wider choice before giving any judgement like it's the case with each big new technology step.
 
I want to say rocks, but theres just nothing to base that on, so i have to go neutral. looks good on paper, but we need to see performance comparisons to current cards.
 
R600 will be faster but arrive a few months later.

So I definitely won't get a GeForce 8800 until I see everything the R600 brings to the table. No sense in wasting cash on an 8800 if the R600 is as good as it's rumored to be, most peeps are doing the same thing.

Most sites are also leaning towards the R600, dunno why but it seems ATi is going to take this generation by storm.
 
I have been putting off my new computer build for DirectX 10 & Conroe. Conroe is here and if NVIDIA can come through in a few weeks.....woo hoo. To bad ATI could not launch at the same time, so we could have one big loud next gen love fest.
 
PRIME1 said:
I have been putting off my new computer build for DirectX 10 & Conroe. Conroe is here and if NVIDIA can come through in a few weeks.....woo hoo. To bad ATI could not launch at the same time, so we could have one big loud next gen love fest.
Click to expand...

Impossible.. R600 is too advanced to launch at the same time. It's the largest most complex processor ever conceived (not just in the graphics industry all over spanning any industry).

Maybe ATi is taking too large of a chance. We'll see.
 
ElMoIsEviL said:
R600 will be faster but arrive a few months later.

So I definitely won't get a GeForce 8800 until I see everything the R600 brings to the table. No sense in wasting cash on an 8800 if the R600 is as good as it's rumored to be, most peeps are doing the same thing.

Most sites are also leaning towards the R600, dunno why but it seems ATi is going to take this generation by storm.
Click to expand...


Can you post a couple of links to those sites? I wanna read about that since i haven't read much about R600 :)
 
gorilla show said:
Can you post a couple of links to those sites? I wanna read about that since i haven't read much about R600 :)
Click to expand...
TotallyPulledItOuttaHisButt.com :p
 
Neutral. NVidia hasn't wowed me since the Ti4200 overclocking days...and those days were short (9500Pro FTW)
 
How could anyone possibly rate a card that isn't out yet? You have not read one single review and the specs of the card have not been confirmed and you are voting that the card Rocks? Just goes to show you the average age and IQ level on these boards are that of a 12 year old child. :(
 
PRIME1 said:
TotallyPulledItOuttaHisButt.com :p
Click to expand...
Ummm no. If you just read comments from the most knowledgeable peeps in the Video Card community (mainly Beyond3D members) you'll see that most have this sneeking suspicion with regards to R600. Mainly because R600 is soo secretive and the words coming out of Orton's mouth.

96 Unified Complex Shaders, 512-bit memory interface and over 500GFlops (Half a TeraFlop). These alone would push R600 ahead. This is what Orton hinted about.

Then the power Requirments. R600 takes almost twice the power of G80 and has been declared as being the most advanced and LARGEST GPU/VPU ever conceived.

I didn't pull any of this out of my butt.
CLICKME!

Why here talking about this instead of five years ago? These problems are becoming more relevant to a broader audience. Research has been done. GPU has evolved to being multiple processors. 48 today, general FP engines. Direction now is to move to common elements for geometry, raster processing. Today, it's 48, future, maybe it's 96.

CPUs are now talking about two, four cores.

We have 375 gigaflops, 64 GB/s. We're about a third of a teraflop. Next generation will exceed half a teraflop on a single GPU.

This class of problems scales very well for parallel processors.
Click to expand...

Wikipedia R600
The R600 is estimated to be the largest GPU ever made. Production based on an 80 nm process with ATI having 65 nm clearly in sights
Click to expand...

G80 will be the fastest GPU for a few months... when R600 strikes I doubt G80 will still be the fastest. All of ATi's shortcomings have also been fixed. Native Crossfire support with no dongle and integrated Compositing engine are also part of the package.
 
Passin' on this one....it's not even out yet so nobody can even make an educated guess as to how well it'll perform.

Though, it does look very promising, I still think ATI will 'win' this round of video cards.
 
Every ATi supporter says "ATi has learned from there mistakes.." which is great, but I can promise you nVidia was not sitting around with blinders on. The 6 and 7 series' were no mistake and had no equal until the x19 series. I would expect serious things from nVidia with this new generation, it just might take ATi another 2 years to catch up again and then again maybe not.

Regardless, the more competition the better. Just bring the damned prices down.
 
The 8800 does look nice. Maybe it's time to upgrade from my X700. It doesn't do that well in CoH.
 
ElMoIsEviL said:
Ummm no. If you just read comments from the most knowledgeable peeps in the Video Card community (mainly Beyond3D members) you'll see that most have this sneeking suspicion with regards to R600. Mainly because R600 is soo secretive and the words coming out of Orton's mouth.

96 Unified Complex Shaders, 512-bit memory interface and over 500GFlops (Half a TeraFlop). These alone would push R600 ahead. This is what Orton hinted about.

Then the power Requirments. R600 takes almost twice the power of G80 and has been declared as being the most advanced and LARGEST GPU/VPU ever conceived.

I didn't pull any of this out of my butt.
CLICKME!

Wikipedia R600


G80 will be the fastest GPU for a few months... when R600 strikes I doubt G80 will still be the fastest. All of ATi's shortcomings have also been fixed. Native Crossfire support with no dongle and integrated Compositing engine are also part of the package.
Click to expand...


exceed half a teraflop
Click to expand...
from ortan's interview so lets just say the r600 will end up around 2 times the performance of an r580. The g80 will do quite a bit more then 500 gigaflops as well, both cards will be close in performance, highly dependent on what clock speeds each GPU is run at of course.

Next time read what they actually say at B3D.

No one there mentioned anything about performance between comparisions of the r600 and g80 the closest to a comparision is this.

I explained the quantities:

The GPU presumably consists of groups of pipelines in order to provide some kind of physical locality, e.g. four groups each of which consists of:
8x quad-scalar ALU processors
3x quad-TMU/ROP fetch/store processors
2x VB/CB fetch/store processors
So that's 32 ALU, 12 TMU/ROP and 2 VB/CB processors per group, with 4 groups, total: 128, 48 and 8 respectively.

It's worth pointing out that you don't really need 3x four separate TMU/ROP fetch processors, you prolly actually have 3 processors, each of which delivers a quad (or multiple thereof) of data. A similar thing goes for the VB/CB processors. They don't necessarily just fetch one item of data, they prolly fetch a block of data, e.g. 32 bytes, 128 bytes, etc. (Something similar to this happens in Xenos with VB/CB.) So the two VB/CB processors will each be able to fetch/store some data in parallel.

As for clock speeds, ALU and VB/CB-fetch are at 1350MHz and TMU/ROP-fetch is at 575MHz. I am just guessing to "match the numbers" here...

The key thing to note in my hypothesis is that a lot of the die that is normally taken up with TMU/ROP pipelining (perhaps half the total area of G71?) is now given over to ALUs, with very compact fetch/store processors which at most only have to perform some simple integer arithmetic, and which have a far less troubled time hiding latency. These fetch/store processors are smaller than TMUs/ROPs because they are only working out what memory addresses to access and then reading/writing. Being dedicated like this, and with the thread scheduler responsible for working out how to hide latency, these units are only minimally pipelined, too.

---

I really do think G80 will be a FLOPzilla due to programmable TMU/ROP, utterly dwarfing R600. Is now the right time for fully programmable TMU and ROP?

One of the intriguing things about this kind of architecture is that whereas R600 has VS, GS and PS threads to switch amongst, G80 would have VS, GS, PS, TMU and ROP threads. The increased quantity of choice might look awkward, but it actually provides a good opportunity for more finely-grained time-slicing of work.

NVidia appears to be building a GPU that prefers not to thread out of order (i.e. it won't thread PS out of order if there's the slightest chance of a problem with overlapping triangles resulting in the wrong results - see my earlier posting about the PIOR flag). ATI's solution to this problem is, I guess, buffering the finished pixels and sorting them back into the correct order.

So, G80 looks like it will process a lot of PS threads in order. Which theoretically presents a bit of a problem when texturing latency, for example, is encountered. Now the set of pixels in flight for threading is constrained to just those pixels that are part of the same "step" (e.g. somewhere else on the same triangle). But, with both TMU and ROP work now available as choices for the ALU pipes to execute, you have a wealth of other work to fill the latency gap. The ALUs can now work on some texture filtering or some AA sample generation. Since TMU and ROP work are both separately queued (and kept in order) everything (PS, TMU and ROP) marches along roughly together and order is maintained, e.g.:
Click to expand...

And its possible in theoretical tests one chip might be better then the other in certain areas but in games things tend to even out with advantages going to both sides.
 
all i have to say is nv30. nvidia basically decided to make their own design a year before dx9 debuted, instead of listening to microsoft and sticking close to them like ati did. ati's 9700 was the dog's bollocks: nv30 floundered even with rigged futurmarks. now dx10 is here and nvidia has decided on a non-unified architecture against the dx10 standard, jumping to the "dx10 compliant" marchitecture. learning from mistakes, indeed.

xbox360, built by ati. ati has already built a successful dx10 gpu and will be releasing their second gen dx10 part with the r600. unified shaders and gddr4 are inclusions that are not rumored; they are included in the r600 and not in nv80. while nvidia has gone through their second tape-out to get their first one right, ati is building ontop of an already working and fast performing model. in all fairness, nvidia has an advantage having the ps3 and its use of openGL SE as well (and incredible linux support), but no matter how much we fight it the pc games market is microsoft's world.

finally, i do not see the point in buying an 8800 when its released. sure, we can say that we are future-proofed for vista and for certain have the fastest card for two months. Why else? dx10 is not coming for at least 3 months, and in that time we could be looking at a refresh of the nv part, or even (dare i say) a price drop in that 3 month timeframe between you buying that huge hunk of silicon and finally getting to use it.

there are just so many reasons, at least for me, to not get an 8800. There is no reason to get a dx10 part yet, since the competition will be intime as well and could be better (i said "could be"!!!). The competition already has experience building a dx10 chip and developers have already been programming for their chips. Finally, nvidia does not have the history of making a great 1st gen directx part, especially when blazing their own path. but that's just my opinion: i could be wrong.
 
http://www.xtremesystems.org/forums/showthread.php?t=120598

victorwang has a g80 and already has posted a ton of info. this card DOES rock! :)
(browse and be amazed. Heh, there are a few people on there pretending to be disappointed)

Plus, from daily tech's blurb not too long ago (backed up by a chinese site) --the g80 DOES have unified shaders (well this is really based on nvidia marketing they got. And well there is the question of what nvidia considers unified. I am taking it a face value though ..g80 has unified shaders. yaya :)).
 
You must log in or register to reply here.
Back
Top