-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
X1900 = X1800 Killer?
Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
as has been implied in many threads.....this will be very interesting.
I dont know wether Nvidias method or ATI´s method of implementing shaders and textures will be more effective. This is going to be the biggest refresh for both companies......ever i think.
It is strange that the Inquirer is not speculating on higher memory speeds for the ATI card. as surely higher colcking memory will be more available soon.
Maybee we are all wrong and there will be another refresh before the R600 with gddr4.
As for the X1800 killer. it might survive as a lower end hgh end part. the 9800 pro was still avaiable by the side of the 9800xt albeit with the same chip.
we could get some crazy things this next round.
I dont know wether Nvidias method or ATI´s method of implementing shaders and textures will be more effective. This is going to be the biggest refresh for both companies......ever i think.
It is strange that the Inquirer is not speculating on higher memory speeds for the ATI card. as surely higher colcking memory will be more available soon.
Maybee we are all wrong and there will be another refresh before the R600 with gddr4.
As for the X1800 killer. it might survive as a lower end hgh end part. the 9800 pro was still avaiable by the side of the 9800xt albeit with the same chip.
we could get some crazy things this next round.
JackBurton
Lifer
WTF? Why the fvck would someone get banned for saying someone is an AEG fanboy? Or even saying they work for AEG? And what the hell is AEG?
Matthias99
Diamond Member
You evidently missed the whole "NVIDIA viral marketing" thing. Please search for several VERY long threads about NVIDIA's use of AEG (a marketing firm that specializes in so-called 'viral' marketing techniques) to supposedly influence forum members by, essentially, bribing them with free stuff.
apoppin
Lifer
i find that hard to believe
did you threaten to sue ATF . . . again 😉
:disgust:
and i don't work for AEG . . . either 😛
:roll:
does that mean we can't even 'think' some of the posers here do work for AEG? . . . it is very likely - actually imo it is indesputable
:shocked:
:laugh:
Here's that AEG thread:
AEG and NVIDIA's "Viral"'outreach programme' targets Forums
JackBurton
Lifer
Apparently I'm missing out on a lot of drama here in the video forum.
JackBurton
Lifer
What is going on with the "do not post names" crap in that thread?
Matthias99
Diamond Member
People kept posting, essentially, "well, that explains <insert forum member's name>'s posts.", either in jest or seriously. Without any sort of proof, that's just a baseless personal attack/flame, and is against forum rules.
Matthias99
Diamond Member
AFAIK, nobody was "banned" (at least in that initial thread; not sure who Rollo is referring to above or what, if any, moderation action was taken). Suspecting is one thing; accusing them without proof is another. Such posts are nothing but flames and/or personal attacks, which are against forum policy. The posts were edited out, and I suspect a few people were warned (and maybe even given vacations).
Duh. Hence why McCarthyistic finger-pointing is not allowed.
QFT
And you're right, my bad, the "smart" witch hunters were vacationed. (not banned)
The whole thing is pathetic. I said I had a friend that went to work for nVidia years ago. (I wasn't pulling that SM3, PVP, shimmer, and 512GTX info out of my ass)
I've never been asked to post anything by anyone, and getting a discount on some engineering samples occasionally and some info from time to time isn't going to make me post "Gee whiz! These are the best thing since sliced bread!" :roll:
If I got a FREE X850 Crossfire set, I'd still post it sucked.
I haven't had problems with SLI on my rigs, with the games I play, so I post "I like SLI". I think 95% of you would probably enjoy SLI as well.
Keysplayr
Elite Member
Matthias, there were two people (could be more but I only know 2 for a fact) were given 2 week vacations for blatant AEG accusations and some wicked flaming. I don't know which of these factors got them these vacations, but only know they both insinuated AEG affiliations toward another member here. But, they definately got those vacations, and one of them is back, campaigning for yet another one. Some never learn. I mention no names in order not to provoke these two members. Why start something from nothing.
Keys
On Topic for RobertR1:
I think the R580 will be an incredible card far overshadowing the R520. And it looks like it will have some good availability. The only thing I really don't care for is that it will still be 16 pipes. Yes, redesigned I suppose, but there is just something about it that makes me automatically not want to buy it. I know better. Pipes arent everything, but in the back of my mind, I can't shake the 16 pipe "inferiority complex". Don't judge me for this as I am just being honest. I do look forward to seeing how it performs. If it severely kicks bootay, I might consider having one in one of my rigs.
Keys
On Topic for RobertR1:
I think the R580 will be an incredible card far overshadowing the R520. And it looks like it will have some good availability. The only thing I really don't care for is that it will still be 16 pipes. Yes, redesigned I suppose, but there is just something about it that makes me automatically not want to buy it. I know better. Pipes arent everything, but in the back of my mind, I can't shake the 16 pipe "inferiority complex". Don't judge me for this as I am just being honest. I do look forward to seeing how it performs. If it severely kicks bootay, I might consider having one in one of my rigs.
keys, it's getting more complicated than mere pipes. Here's my relatively simplistic and probably somewhat wrong way to think of some GPUs based on (rumored) specs:
R520: 8 VSs -> 16*1.5 PS ALUs + 16 TMUs -> 16 ROPs
G70: 8 VSs -> [24*2 PS ALUs + 0 TMUs *OR* 24 PS ALUs + 24 TMUs] -> 16 ROPs
R580: 8? VSs -> 48*1.5 ALUs + 16 TMUs -> 16 ROPs
G71: 10? VSs ->[32*2 ALUs + 0 TMUs *OR* 32*1 ALUs + 32 TMUs] -> 16 ROPs
AFAIK, G7x is more flexible in that it can vary its pixel shader load b/w those two extremes. Its TMU is built into (or shares logic with) the first PS ALU, so if you use it you can't use that ALU for that clock, just the second ALU. So, per clock, it has potentially more TMUs, but at the expense of some shader power. This assumes you can actually conduct 32 texture accesses per clock, which may be too much for a 256-bit bus.
I guess the idea behind ATI's new "ring bus" memory controller is to squeeze more efficiency out of the TMUs they have. It certainly seems to help their AA numbers, but I'm not sure about AF efficiency relative to TMU count. I Exceled some of Rage3D's 6800GS review's benchmark numbers with AF off and on and saw that the X1600XT's 4 TMUs gave it a bigger hit with AF than other cards in most games--except in FEAR, where it tracked with the other cards. This may be encouraging WRT future games, where you may not need as many TMUs when shaders get longer and more complex. How does this relate to R580 and G71? I don't know if "only" 16 TMUs (compared to G71's theoretical 32) is as much of a limit compared to RV530's 4, but it may not be with newer games like FEAR, and probably future games in general.
The picture is getting more complicated, and I'm omitting and probably overlooking a lot of small details (like the average MULs/ADDs/MADDs per shader, how often specialty functions like FP16 NRM or SINCOS are used, how AF affects TMUs usage, etc.).
I'd really like to see how R520 and R580 compare in two punishing titles like CoD2 and FEAR.
Edit: Thanks, hope it helps.
R520: 8 VSs -> 16*1.5 PS ALUs + 16 TMUs -> 16 ROPs
G70: 8 VSs -> [24*2 PS ALUs + 0 TMUs *OR* 24 PS ALUs + 24 TMUs] -> 16 ROPs
R580: 8? VSs -> 48*1.5 ALUs + 16 TMUs -> 16 ROPs
G71: 10? VSs ->[32*2 ALUs + 0 TMUs *OR* 32*1 ALUs + 32 TMUs] -> 16 ROPs
AFAIK, G7x is more flexible in that it can vary its pixel shader load b/w those two extremes. Its TMU is built into (or shares logic with) the first PS ALU, so if you use it you can't use that ALU for that clock, just the second ALU. So, per clock, it has potentially more TMUs, but at the expense of some shader power. This assumes you can actually conduct 32 texture accesses per clock, which may be too much for a 256-bit bus.
I guess the idea behind ATI's new "ring bus" memory controller is to squeeze more efficiency out of the TMUs they have. It certainly seems to help their AA numbers, but I'm not sure about AF efficiency relative to TMU count. I Exceled some of Rage3D's 6800GS review's benchmark numbers with AF off and on and saw that the X1600XT's 4 TMUs gave it a bigger hit with AF than other cards in most games--except in FEAR, where it tracked with the other cards. This may be encouraging WRT future games, where you may not need as many TMUs when shaders get longer and more complex. How does this relate to R580 and G71? I don't know if "only" 16 TMUs (compared to G71's theoretical 32) is as much of a limit compared to RV530's 4, but it may not be with newer games like FEAR, and probably future games in general.
The picture is getting more complicated, and I'm omitting and probably overlooking a lot of small details (like the average MULs/ADDs/MADDs per shader, how often specialty functions like FP16 NRM or SINCOS are used, how AF affects TMUs usage, etc.).
I'd really like to see how R520 and R580 compare in two punishing titles like CoD2 and FEAR.
Edit: Thanks, hope it helps.
Are you sure only the first ALU is used, AFAIK both ALUs are used for texture addressing (could be wrong so please correct me if I am)?
If it can use the second ALU then it looks like it only has a 50/50 chance of doing so...
You wouldn't happen to know how many threads in flight on the NV40/G70/G71? Or if they decreased the batch size?
Yeah, I don't expect the R580 would blow away the R520 esp. in games where the 16-1-3-1 config. doesn't greatly benefit over the 16-1-1-1 setup of the R520. The benchmarks should be quite interesting. I'm thinking in certain games you'll see small gains, while huge gains in other and the G71 will have consistantly higher gains all across the range. We'll see soon though.
As for pipes. It's similar to the engine mentallity. 4>6>8>10>12 cylinders. We naturally relate more power to higher numbers. Pipelines are being thought of as cylinders and general assumption is, the more the better. It is quite a mental barrier and was one for me also when I an option to pick the x1800xt over the 7800GTX 512 at a much cheaper price. The "16" pipe will be a much tougher sell than the "32" pipe G71 due to that. You can bet that nvidia will be using the word pipeline like no tomorrow during their marketing campaign and rightfully so. It's good for their business.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 24K
-
-
