--
RV770 will break 1TFLOPS, while the R700, with two RV770, should be capable of twice that, more than 2TFLOPS. This is higher than what he had expected, but then again the core frequencies are also higher than expected. All cores are made by TSMC and their 55nm process. We're still trying to hunt down more reliable information on the design of the "core".
--
The idle power consumptions of all cards are extremely low and even the dual-core Radeon HD 4870X2 will consume less than 25W when idle.
--
The number of TMUs have finally been increased, actually doubled, to 32 with RV770, while RV740 sports 24 TMUs, and the low-end RV710 has 8 TMUs
--
To end this long and speculative post, we would like to say that if not all, most of the variables here are subject to change and should be considered with a "tiny" pinch of salt. We're still 4 month away from the launch.
Well if thats anything like accurate then its back to the drawing board again with my build scheduled for about april, I can remember not so long ago being bothered by the relativel ack of new tec, now i finally decide to build they release a new CPU/GPU every other month. Oh well its all good and if the reported specs are close then the fight is really on with Nvidia.
Guess i will just have to draw the line in the sand some where or the thing will never get built. Definatly time the TMU count went up but i was kind of expecting that it just seemed a logical step to take, would have liked the ROP count up as well but they need to save something for further down the line dont they.
Mactronix
Message edited by mactronix on 02-16-2008 at 12:09:01 AM
Just buy what you can afford. You don't have to wait on the parts every time Nvidia, AMD or intel release something new.
These 4000 cards only compares to Geforce 8 series. Sure it has higher clocks and memory but it's still a 16 rop card with 256bit memory bus. It's not going to make some drastic improvements of what we have now.
Just buy what you can afford. You don't have to wait on the parts every time Nvidia, AMD or intel release something new.
These 4000 cards only compares to Geforce 8 series. Sure it has higher clocks and memory but it's still a 16 rop card with 256bit memory bus. It's not going to make some drastic improvements of what we have now.
How would it not make drastic changes? Can't you see a kick ass midrange here? 4850 with 480 shaders, 32tmu and the 4670 with 256bit bus, 24tmu, 240 shaders. Huge bang for buck if ATI sticks with the pricing. 4670 @ around $100.
How would it not make drastic changes? Can't you see a kick ass midrange here? 4850 with 480 shaders, 32tmu and the 4670 with 256bit bus, 24tmu, 240 shaders. Huge bang for buck if ATI sticks with the pricing. 4670 @ around $100.
30-50% difference? Not much really. From 7 to Geforce 8 was a drastic change. Much as 100+% difference.
These new cards cost more anyway. 8800gt is currently $200 after rebates.
You do know Geforce 8800gt has 16ROP 56TMU right? 480 shaders sounds a lot but current AMD cards with 320 shaders can't compete with Geforce 8 with 128 shader. IMHO 320 AMD shader is only equivalent to Nvidia's 96 shader.
Good price is always good on the low end however. About time 256 bit memory is mainstream.
True, but a 250mhz jump is pretty big for graphics cards
I would imagine that ATI would have tried to get the highest clocks possible on their HD3XX0 series to be more competitive with nvidia now, but then again maybe TSMC has refined their 55nm process to get higher yields of higher clocked GPUs.
Message edited by turboflame on 02-16-2008 at 12:59:58 AM
480 shaders sounds a lot but current AMD cards with 320 shaders can't compete with Geforce 8 with 128 shader. IMHO 320 AMD shader is only equivalent to Nvidia's 96 shader.
If you are referring to the performance level, i tend to agree. The 320 shaders of the 2900/38xx series are, as a fact, only 64 complex shaders. The others can only compute simple instructions and are worthless without proper drivers.
I'd really like to see some information regarding the efficiency of those simply shaders, especially if their transistor count is compared to that of their complex brethren.
The core clock is high then what i would have expected but the memory is not that high for Gddr 5 hell even gddr4 and 3 go that high
I'm not 100% sure about this, but I think they merely listed the base clock speed, in which case you'd have to multiply it by two to get the effective clock speed. So 1800MHz would actually be 3200MHz, which if I recall correctly has some significance, like what .6ns (or something) memory is supposed to run at. Can someone confirm this part?