Oh my god, I still haven't got a working desktop, still waiting for RV770, now this?
Wow I'm in /fail territory
At this rate I'll be dead before I get a proper desktop going
It's kind of crazy when you think about it, but M$ is such a freaking monopoly that the effect of many of their decisions goes from tiny ripples within their own industry to tidal waves in others. M$ could have easily supported DX10 on XP, but decided that Vista needed all the help it could get... Are any of the various versions of OpenGL limited to a specific OS? I'm not giving them(M$) another dime until they offer a truly compelling OS.
The PC gaming industry has enough headache without M$ coming in and everyone.
Don't forget, nVidia has access to the same 40nm process as well. They'll have plenty of time to release a GT200 refresh that is more than just a die shrink.
OpenGL, so they are OS/platform independent, unlike DirectX.
core i7 920 2.66(for now, will OC to 4GHz with my CM V8 soon)
rampage 2 extreme bios 0901
corsair 3x2gb DDR3 1600 c8
2 GTX 260 core 216
2 WD 750gb RE3 raid 0
1200 odin psu
vista premium 64-bit
lian-li v2010
Case-Coolermaster Cosmos S
MoBo- ASUS Crosshair IV
Graphics Card-XFX R9 280X [out for RMA] using HD5870
Hard Drive-Kingston 240Gig V300 master Seagate 160Gb slave Seagate 250Gb slave Seagate 500Gb slave Western Digital 500Gb
CPU-AMD FX-8320 5Ghz
RAM 8Gig Corshair c8
Logitech 5.1 Z5500 BOOST22
300Gb of MUSICA!!
Steam ID: alphamonkeywoman
http://www.techpowerup.com/gpuz/933ab/
Dilemmas:
With respect to Die Shrink:
ATI- None, 55nm -> next new process (40nm) is logical
Nvidia - Bring out much anticipated 55nm parts in September/October or go all out 40nm to catch up to ATI's nex gen process?
With respect to Multi-GPU:
ATI - None, 3870X2 was a benchmark monster but with performance issues in games. 4870X2 will fix a lot of those issues, but some will remain, but no dilemmas here, other than maybe wanting to release a 4850X2 down the road or not.
Nvidia - Go multi GPU with 9800GX2+ or not? Go Multi-GPU on GT 200b or wait to do it on nex gen chips? Stick with single gpu strategy for GT 300 so as not to have a GT 200 X2 challenge a future GT 300 series just like 9800GX2 performance caused concerns for GTX 280 competitiveness?
With respect to next architecture:
ATI - Decide whether or not move to 512bit is necessary considering cheap manufacturing cost of RV770 shader-based approach and better than expected competitiveness. Tape out 5870 before end of year for release in early 2009.
Nvidia - Make another 512-bit monster and hope costs go down by q1 2009 due to economies of scale while making sure to price it competitive with ATI and still make a profit? Bring out mainstream GT 200-based products now or allow brand new 55nm G92 products to live on till 2009?
Just a run down of issues both companies face, and how Nvidia has far more dilemma's right now with respect to any certainty on future products EVEN WITH access to the same 40nm process as ATI.
(Ask me before posting the above on other forums btw).
Perkam
Last edited by perkam; 08-03-2008 at 04:21 AM.
It is widely speculated that nvidia will adopt 40nm around the same time as ATi. Nobody expects nvidia to bump the high end to 40nm right away, but anything is possible. The GT200 series was made to be shrunk.
In terms of pure performance ATi can't compete chip for chip with nvidia now, with the playing field leveled at 40nm they wont stand a chance. Obviously the best thing that can happen for ATi is a slow and painful transition to 40nm for nvidia.
The new lil Dragon will be quite nice though and probably priced well to match.
Making games are easy, and the solution for pc are more online games.
MMOG, fps shooters, that are well designed and thought out, single player simply seldom makes it worthwhile to buy.
I still play BF2 even when its not supported and some cod4.
that just about it.
some civ4 and CoH then and now.
I miss a great tank battle game, with 50 tanks running and smaking bootoms.
4670k 4.6ghz 1.22v watercooled CPU/GPU - Asus Z87-A - 290 1155mhz/1250mhz - Kingston Hyper Blu 8gb -crucial 128gb ssd - EyeFunity 5040x1050 120hz - CM atcs840 - Corsair 750w -sennheiser hd600 headphones - Asus essence stx - G400 and steelseries 6v2 -windows 8 Pro 64bit Best OS used - - 9500p 3dmark11 (one of the 26% that isnt confused on xtreme forums)
i am expecting them to bump their high end to 40nm. market is not just that slow anymore to "make your high end shrink your old get experience shrink your high end" strategy. if nvidia is not feeling ready they can follow this again but this means that nvidia accepted ati will be leading in that series too.
When i'm being paid i always do my job through.
Possible yes, but I don't see why they would need to do it again. RV770 is a great chip and going for 2x performance/watt while moving to 40nm points to lowering power consumption first of all.
2000sps just doesn't fit with AMD's strategy of trying to create small and power efficient chips that scale well to multi-chip. We shouldn't dismiss anything at this point though.
//Andreas
I maintain my position that the RV870 is 40-60% faster than RV770 under simillar power load and design methodolagies.
RV870 will not have the luxury of having a mature process like RV770 did, though, so in this area they might have problems with scaling or power usage. But the chips will be darn impressive- ~9600GT size, 2.5x the performance!
My think about this :
RV870 is :
40 nm Tech
960 SP
144mm²
384bits wide
5970 X2 is Two 5970 on same PCB - 350W
5970 is Tri core of RV870 w high frequency - GDDR5 1.5Go - 275W
5950 is Tri core of RV870 w low frequency - GDDR5 768Mo - 245W
5870 is Dual core of RV870 w high frequency - GDDR5 768Mo - 175W
5850 is Dual core of RV870 w low frequency - GDDR5 768Mo - 150W
5670 is single core of RV870 w high frequency - GDDR5/3 768Mo - 80W
5650 is single core of RV870 w low frequency - GDDR5/3 384Mo - 50W
One single chip, for high end, mainstream and low end. Nice, seem very nice
Sure, the GTX280 has the single GPU crown right now, but considering the 4870 isn't that far off from the GTX280 even and that the die is half the size, it tells me that ATI still has a few tricks up their sleeves.
As perkam alluded to, the RV770 finally validated the R600 architecture. Yeah many changes were made, but the basic architecture principles were intact. Furthermore, they did more with a smaller die than Nvidia did with a much larger one, which despite doubling all specs from G92, did not offer anywhere near 2x performance. So either Nvidia is hitting a wall with their G80-derived architecture or ATI's architecture scaled that much better (efficiency wise, at least).
So it'll be interesting to see what nvidia actually does with 40nm. Keep in mind, only a couple years ago ATI was winning the single massive GPU crown so its not like Nvidia is always dominating at the single GPU level.
Multi-core ships aren't confirmed yet, right? And after all, we have 960SP per chip or per core? If it's the second it makes 3x320SP, which puts one more connection in the whole system, one more place to have issues.
And still, what's better with multi-core chip, when the performance and the amount of sp/tmu/rop are the same? The only thing, that come in my mind is that this technology is cheaper. They produce just cores and pads (the thing, where the cores are placed) and combinate them.
That rumor said it would have 2000 SPs and like Delph1 said, that does not fit in with AMD's strategy. That 2000 SPs rumor started as a joke over at Beyond3D, but there were some rumor sites that started stating it as more of a fact a couple of days afterwards and that's why those rumors have little credibility in my eyes at least.
EDIT: I found the link about that joke:
http://forum.beyond3d.com/showthread...42#post1173642
Last edited by Helmore; 08-03-2008 at 03:21 AM.
"When in doubt, C-4!" -- Jamie Hyneman
Silverstone TJ-09 Case | Seasonic X-750 PSU | Intel Core i5 750 CPU | ASUS P7P55D PRO Mobo | OCZ 4GB DDR3 RAM | ATI Radeon 5850 GPU | Intel X-25M 80GB SSD | WD 2TB HDD | Windows 7 x64 | NEC EA23WMi 23" Monitor |Auzentech X-Fi Forte Soundcard | Creative T3 2.1 Speakers | AudioTechnica AD900 Headphone |
384bits wide bus would be great but I greatly doubt it.
Me too, a 256-bit bus on GDDR5 is more than sufficient for the moment, especially because GDDR5 will be more widespread by then and it will feature even higher clocks by then. So I basically expect RV870 to still have 4 64-bit memory controllers with each 4 ROP cluster, but these will probably be fatter than the current cluster.
"When in doubt, C-4!" -- Jamie Hyneman
Silverstone TJ-09 Case | Seasonic X-750 PSU | Intel Core i5 750 CPU | ASUS P7P55D PRO Mobo | OCZ 4GB DDR3 RAM | ATI Radeon 5850 GPU | Intel X-25M 80GB SSD | WD 2TB HDD | Windows 7 x64 | NEC EA23WMi 23" Monitor |Auzentech X-Fi Forte Soundcard | Creative T3 2.1 Speakers | AudioTechnica AD900 Headphone |
Bookmarks