Or Nvidia still has problems with the memory controllers...Originally Posted by Iconyu
Or maybe they need to stay within a certain TDP...
Or Nvidia still has problems with the memory controllers...
Or maybe they need to stay within a certain TDP...
Originally Posted by motown_steve
Every genocide that was committed during the 20th century has been preceded by the disarmament of the target population. Once the government outlaws your guns your life becomes a luxury afforded to you by the state. You become a tool to benefit the state. Should you cease to benefit the state or even worse become an annoyance or even a hindrance to the state then your life becomes more trouble than it is worth.
Once the government outlaws your guns your life is forfeit. You're already dead, it's just a question of when they are going to get around to you.
Because Nvidia is deliberately allowing this information to leak out for hype purposes. Their card does better at 2560x1600 and high AA due to the VRAM advantage. It is very very likely that the 470 does worse at 1920x1200 8X aa or less. They want the card shown in the best light possible leading up to its launch. I don't see the point though as when NDA expires and the full story is out they'll be caught. Although NDA is still up, I am quite sure that their is a good reason we have only seen 2650x1600 high AA comparisons so far.
Feedanator 7.0
CASE:R5|PSU:850G2|CPU:i7 6850K|MB:x99 Ultra|RAM:8x4 2666|GPU:980TI|SSD:BPX256/Evo500|SOUND:2i4/HS8
LCD:XB271HU|OS:Win10|INPUT:G900/K70 |HS/F:H115i
Tom Clancy's H.A.W.X. is one of the less demanding games, I'd say it's playable even with this settings. 50FPS not enough? Guys I played it on less than 30 FPS and no problem though
I think we should start a new "Fermi part <InsertNumberHere>" thread each time it's delayed in this fashion!Heck, I think we should start a whole new forum dedicated to hardware delays.
better min frames again. GTX 480 should be around 14% faster with full 512 shaders. Perhaps lightly higher clocks but i think <700 Mhz.
Im sure we'll see the GTX 480 around 15% over the HD5870 / GTX 470. But like in past, some more stable framerates than on ATI cards. Min frames seem to be a bit better. Perhaps in new games with DX11 we'll se a tresselation advantage of the GF100. Does anybody knew if the two tresselation components on one HD5970 can work together?
Last edited by Krautmaster; 03-11-2010 at 12:35 AM.
Intel 990X ES @ 6x4,60Ghz@1,440V klick / DFI UT x58@ MIPS HD6950@70@1Ghz / 256GB M4 SSD / X-Fi / HK3.0 - Laing Ultra - Thermochill PA120.3 + TFC 120er + Cape Cora Ultra 8x/ Aquaero Steuerung / Lian Li A7110B / Seasonic Platinum 1kw / 2xDell IPS 23" / Win 7 + Server 2008 R2/ DSL 16K
HTPC + HTPC Wiki + sysProfile + Clarkdale @<15W + Eigenbau-Ambilight + Ambilight PreOrder
It's tesselation
And GTX 480 apparently only has 480 sp's now
i dont think tdp is whats holding their imc and memory clocks down...
to have the same bandwidth? or to have the same perf? in what resolution? there is no one correct answer to this question...
i suspect that nvidia sent out 470 cards to partners that had artificially low memory clocks of only 850mhz... each card probably had different memory clocks and gpu clocks so they could trace who leaked what, and they could trick ati into believing that their perf is lower than it actually is, so they can surprise them...
unless nvidia has problems with the imc or pcb, im sure they will clock their 470s at least at 1ghz, ie 4ghz effective, if not higher... thats a 15% memory bandwidth perf boost and will make them shine even more in high res with high AA, and should give them at least a 5% boost at lower resolutions as well.
a 5970 scores much higher in unigen heaven, so my guess is yes?
It depends on what you are calling to "work together".
HD5970 works in AFR mode (Alternate Frame Rendering), so each GPU is rendering a different frame at the same time. So each GPU works the same way than a stand alone GPU in its own frame, regardless what the other one is doing.
Obviously that means that the tesselation engine of both GPU's (as any other component of the GPU) will be working in parallel to the other, one on the frame n, and the other in the frame n+1.
Same performance of course. Having the same bandwidth doesn't tell you anything. And there is no correct answer until we see hard numbers which is why I was asking why you think they need higher memory clocks. Maybe 160GB/s is more than enough. At least it's way higher than the previously rumoured 128GB/s.
INTEL Core i7 920 // ASUS P6T Deluxe V2 // OCZ 3G1600 6GB // POWERCOLOR HD5970 // Cooler Master HAF 932 // Thermalright Ultra 120 Extreme // SAMSUNG T260 26"Has anyone really been far even as decided to use even go want to do look more like?
Looks nice!
Heh, why does the reflection of the 470 box have 480?
Why is written GTX 480 in shadow?
god thats so lame
Heh, why does the reflection of the 470 box have 480?
Project TJ07 WeeMaan edition in progress...
i7 920 @ 4,4ghz, P6T Deluxe, Corsair Dominator 6Gb, HD5870, OCZ Vertex 60gb + Samsung F1 1tb
Heatkiller copper, EK 5870, Thermochill 120.4, DDC 3.2
Heh, why does the reflection of the 470 box have 480?
INTEL Core i7 920 // ASUS P6T Deluxe V2 // OCZ 3G1600 6GB // POWERCOLOR HD5970 // Cooler Master HAF 932 // Thermalright Ultra 120 Extreme // SAMSUNG T260 26"Has anyone really been far even as decided to use even go want to do look more like?
You're probably right, the card could be so screwed that its memory controller just can't handle any more.
not only that, the reflection of the side of the box does not match the box at all as wellHeh, why does the reflection of the 470 box have 480?
where do all these botched box arts come from
1. Asus P5Q-E / Intel Core 2 Quad Q9550 @~3612 MHz (8,5x425) / 2x2GB OCZ Platinum XTC (PC2-8000U, CL5) / EVGA GeForce GTX 570 / Crucial M4 128GB, WD Caviar Blue 640GB, WD Caviar SE16 320GB, WD Caviar SE 160GB / be quiet! Dark Power Pro P7 550W / Thermaltake Tsunami VA3000BWA / LG L227WT / Teufel Concept E Magnum 5.1 // SysProfile
2. Asus A8N-SLI / AMD Athlon 64 4000+ @~2640 MHz (12x220) / 1024 MB Corsair CMX TwinX 3200C2, 2.5-3-3-6 1T / Club3D GeForce 7800GT @463/1120 MHz / Crucial M4 64GB, Hitachi Deskstar 40GB / be quiet! Blackline P5 470W
Nothing is released to public yet, either the GPU or the box
Consider it as a beta-box.
► ASUS P8P67 Deluxe (BIOS 1305)
► 2600K @4.5GHz 1.27v , 1 hour Prime
► Silver Arrow , push/pull
► 2x2GB Crucial 1066MHz CL7 ECC @1600MHz CL9 1.51v
► GTX560 GB OC @910/2400 0.987v
► Crucial C300 v006 64GB OS-disk + F3 1TB + 400MB RAMDisk
► CM Storm Scout + Corsair HX 1000W
+
► EVGA SR-2 , A50
► 2 x Xeon X5650 @3.86GHz(203x19) 1.20v
► Megahalem + Silver Arrow , push/pull
► 3x2GB Corsair XMS3 1600 CL7 + 3x4GB G.SKILL Trident 1600 CL7 = 18GB @1624 7-8-7-20 1.65v
► XFX GTX 295 @650/1200/1402
► Crucial C300 v006 64GB OS-disk + F3 1TB + 2GB RAMDisk
► SilverStone Fortress FT01 + Corsair AX 1200W
seems it will take some time for fermis box to get its true reflection. so don't be suprised when you go home put your gtx 470 on table and see reflection of gtx 295 :P
C'mon guys, spend more than 1 second looking at these pics and you'll see they're probably both photoshops.
The box, yes GTX 480 as people have said.
The card, "Fermi"? That's pre-codename, even the codename is GF100 - they've never written codenames on the product before.
Considering the Tesla versions will have between 3GB and 6GB of GDDR5, I don't think the memory controller is anywhere near an issue.
+1I'm getting tired of all these """"teaser"""" benchmarks. Most probably the results are meaningless anyway
And still a lot more "teasers" to come.
15 days means teasers numbering near 30-50 all with a resolution of 2560x1920 or higher. Where possible tessellation will be maxed out otherwise use of ultra high texture will be done to bottle neck the 1GB memory on the 5870/5970
Coming Soon
My guess is that will be two versions of GTX 480, one with 480SP and another
with 512 SP (maybe Ultra or Plus version)
However, nVidia needs a good 470 with an attractive price IMHO. If not, things will be well tougher ...
We're talking about clocks speed on the memory, not the amount. If I'm missing something here, you're being too subtle for me.Considering the Tesla versions will have between 3GB and 6GB of GDDR5, I don't think the memory controller is anywhere near an issue.
Posting Permissions
Bookmarks