:shocked::shocked::shocked:
:shocked::shocked::shocked:
How much CPU affects the final results of this test???Quote:
Originally Posted by Gaul
HD 6870 - Phenom II X4 @ 4063Mhz
GTX 460 - Core 2 X9650 @ 3007Mhz
To be honest,CPUs are clocked at different frequencies(QC Deneb @ 4Ghz,QC Penryn @ 3Ghz),but that is not nearly enough to offset the more than 2x performance difference between 460 and 6870. Amazing what a GPU like Barts can pull.
edit:
ethomaz ,C2 X9650 is a QC as far as I know.
This is interesting because MSI Kombustor like FurMark is very GPU intensive and the CPU does not affect much the final score. Of course FurMark-like tests are not representative of real 3D tests like in game rendering but it shows the GPU raw power… :up:Quote:
Originally Posted by ethomaz
Also better min frame rates. I don't really think this is a very good real world benchmark, though.Quote:
Originally Posted by Gaul
Yes to both. :yepp:Quote:
Originally Posted by informal
Regarding MSI Kombustor benchs, how much 58xx got for that bench? Seems architecture changes are not that tiny even it is tiny bench :D
Sorry... I read just X and thought it was a dual core... so it's QX9650.Quote:
Originally Posted by informal
Mine...Quote:
Originally Posted by ethomaz
Stock cpu and 775/1125 gpu
http://img229.imageshack.us/img229/4715/komph.jpg
Yeah, right so it's irelevant, 5850 is also twice as GTX 460 in Msi Kombustor...
kombustor made for Ati xD I guess even 48xx series faster than GTX4xx in kombustorQuote:
Originally Posted by xdan
its based on Furmark, which has since the 2900s, been doing very well with the simple shaders on ATI hardware.Quote:
Originally Posted by Dezmen
i still remember the whole fiasco where ATI cards would overheat and shut down, or throttle. and everyone is freaking out, yet none of them ever cared to realize their fps was also 50-100% faster than the competitions.
just did a test on my 4850, got 220fps average, however i only had opengl 2 or 3 to choose from, no d3d10:shrug:
http://www.geeks3d.com/20100810/gpu-...1-3-available/ here you go. same as on ss, 1.1.3 versionQuote:
Originally Posted by Manicdan
thanksQuote:
Originally Posted by Dezmen
heres my OpenGL 2 numbers
212 min
252 avg
292 max
score: 15134
single 4850 @ 675/1150
http://img109.imageshack.us/img109/9810/9436.png
oddly things didnt run at that resolution. the text was a bit blurry and i had a popup from my monitor saying i wasnt running 1080p, not sure how much that affects things
but basically ATI cards do great, and consume massive amounts of heat in these fuzzy programs
now i see that is not good bench to judge
even my 4850 beats 460GTX :D
The difference in scores seems quite close to the difference in GFLOPS between 6870 and 460 (and the other cards posted here).Quote:
Originally Posted by Gaul
Some more FUD info (looks accurate): link
Quote:
We are still getting some last minute details about Barts PRO and XT. This time we can tell you that Barts PRO and XT parts feature a 255mm2 die size. This is quite good as Radeon HD 5850 and 5870 had 334mm2 size. The chip is almost 30 percent smaller and the new one consumes up to 151W in the XT version, which is less than the Radeon HD 5850 series. Radeon HD 5850 had 2.15 billion transistors while the Radeon HD 6850 and 6870 have 1.8 billion.
Radeon HD 6870 has a total memory bandwidth of 134.4 GB/s, 900 million polygon/ sec geometry throughput and 14 SIMD engines, four less than Radeon HD 5850. the total number of stream processors is 1120, while HD 5850 has 1440, the 6870 has 56 texture units, 128 Z/stencil ROPs and 32 color ROPs.
AMD claims that with a good CPU it scores 7730 in 3Dmark Vantage Extreme, most likely with a six core top notch CPU. The maximal board power is as we previously reported 151W while the idle power stays at 19W. The idle power is better than Radeon 5850's 27W while the maximal board power remains the same.
Launch day is as Friday the 22nd, as we've already said a few times. All this comes from the mouth of Eric Demers, CTO of graphics group and a very nice chap. Of course he didn't tell us this directly, but these specs is what he showed to AMD's tame press.
this rename arguing is completely pointless; when did we see completely new architectures?
ATI:
9800-X800 refresh
X1k series completely new
HD2xxx new shader part; same memory controller (2x wider)
HD3xxx tweaked shader part; different mem controller
Hd4xxx same shader, tweaked mem, bigger = refresh
HD5xxx tweaked shader (DX11; tesselation); same MC, same front end
HD6xxx tweaked shader / new shader (barts/cayman); unknown MC; new frontend
so when was the last time we saw a completely new architecture?
i think that was the x1xxx....
going by the logic of some people in here we'd call the next card X1999 XTXXXX edition v2....
if there is a major perf improvement or power efficiency improvement you can give it a new name; which has been the case since the 7xxx days with ati....
nvidia on the other hand had some very inconsistent naming in the last few years....
this post shows why Nvidia has large dies.Quote:
Originally Posted by Aten-Ra
from G80-GT200 you gain 33% larger SIMD array (I think) The node shrink only help so much in design
and from GT200 to GF100 you gain an even larger SIMD array.
They have only kept adding to the SIMD arrays and have not taken anything out to shrink it's design size.
ATI here did this
3870-4870 did away with Ring bus and added more UDV stuff
4870-5870 did away with crossbar in there and add more powerfull tesseltion unit
5870- 6870 did away with 5:1 simd and went to 4:1 and add even more powerful tessellation unit lastly dual Rasterzisted.
(this excludes any AA and AF changes)
Yep I agree with you , and think bout it logically why would the design a new arch over and over again when there is plenty of improvement and tweaks to be had with the old arch , its not like this is the exact same thing, its the old arch with stuff readjusted, its been tweaked it uses less to do more and that is what progress is all about. I mean nvidia arent going to redo fermi from the ground up, they are going to tweak it , redo bits to deal with the issues, but they are going to call next set 5xx.
This whole arch-u-ment is pointless, it doesn't have to be a complete redesign, to be new.
they tweak and make small improvements and offer a better price/performance ratio.
if u run a company, its kinda hard to justify a Fermi approach especially since amd still runs red in the books.
however I do expect the 6970 be a little extra than last gens 5870.
I will be surprised if you can qualitatively tell the difference between 6870 and 5870.
a judge of this will be the normally ridiculous (but well intentioned) hardocp benchmarking system.
if hardocp thinks there isn't enough difference to warrant different visual settings, they will benchmark both cards at the same settings.
Yeah I just about lost it when I saw this too. LOL TWICE AS MUCH POWER!!! :rofl:Quote:
Originally Posted by xdan
Some MOAR pics
http://i1130.photobucket.com/albums/...g?t=1287586369
http://i1130.photobucket.com/albums/...g?t=1287586433
sorry if they were posted earlier