Printable View
I'm done with SLI and Crossfire, I won't buy a system equipped with multi-gpu again (see: microstuttering). I'm positive microstuttering is going to be solved somehow, but not right now.
I'll sell my two 8800GT's and get one 4870. I'm very sure it'll be a lot more consistent and fluid.
30% is hardly the best case scenario. The 4870 has almost twice the bandwidth, and (for some models) a twice larger frame buffer. It's not hard to think of situations where the 4870 will be several times faster than the 4850. Right now that means the 4870 will need high res, aa, and aniso to really pull ahead of the 4850. Sure, at lower settings the difference will be negligible, but that's not what matters (If it runs > 60 FPS who cares what the performance numbers are?). Future games will take advantage of the added memory and bandwidth at lower and lower resolutions. Just look at where the GTS 640 and GTS 320 are today vs. where they were at launch.Quote:
Originally Posted by Tonucci
Quote:
Originally Posted by GGuyZ
Too right, dont get me started on this launch, especially of it in the UK.
:mad::mad:
I was talking about the 512Mb version, forgot about the 1Gb ones...With double frame buffer you may be right, lack of video memory can knock FPS into single digits.Quote:
Originally Posted by hurleybird
But I think its unlikely that double bandwidth and increased GPU frequency alone would translate into 30% faster average performance in games.
yeah not doubting the results, but just wondering why the thai site had lower results.Quote:
Originally Posted by Mad1723
the french site did seem more comprehensive in testing. I m optimistic, those results will be the norm :) As much as my gx2 has served me well, those 4850's and by inference, 4870's are looking amazing.
Wise man, my 8800gtx has given me a much better experience than 2 x 9600gt or 2 x 1900xt do/ did.Quote:
Originally Posted by annihilat0r
Bench's might say 2 x 9600gt are 40% faster than my 8800gtx, but the 8800gtx feels much better to me, I wont be touching crossfire or sli again either.
Looking forward to seeing if the rumours of the HD4870X2 not being based on crossfire are true, members here say it is fact, but I still have my doubts that Ati have tech that complex ( it took games years to use dual core cpu's over single core , HD4870x2 is suppose to be based on similar tech and people think todays software will crossover perfectly )
I am finding that hard to believe..
lets calm the heck down here people!! how many have seen mascaras's review? please everyone go take a look at it before everyone looses there pants. he is an extremely trusted forum source and the closet thing to real numbers we have. and he has done extensive testing in Crysis and different resolutions with and without crossfire. he has done direct comparisons to the 9800gtx and it is slower in every case. that's right it's slower than the 9800gtx by about 1-5FPS with and without AA at High and Very High and at different resolutions.
so people need to take it slow and realize that the card is not going to beat the GTX 260. in fact it will be slower. by ALLOT he has also done crossfire numbers against his 9800GX2. and a HD4850 Crossfire is slower in Crysis than the 9800GX2. and if we remember the GTX 280 is a bit faster than the 9800GX2 in Crysis so then we can say that the HD4850 Crossfire will be slower than the GTX 280. please read the reviews done by trusted sources before posting stuff like "nvidia is teh screwed cuz HD4850 is faster than GTX 260!!1!!!111!!!" HD4850 is on par to be about the same performance as the 8800GTS 512. and knowing this we can say the HD4870 will slot in nicely ahead of the 9800gtx in performance and still be lower than the GTX 260 in price and performance. and thus the saying "you get what you pay for" will be shown in it's full glory.
oh and BTW he did notice microstuddering in Crysis under crossfire at certain points in the game.
Crysis is just one game that ran rather badly,compared to NV's cards,on Radeon HD3xxx.Also,it happens to be one game that scales poorly in CF and SLI.So to say Radeons are not shining in Crysis at any setting is nothing new.
Let us see how the card(s) perform under a variety of engines ,just like hardware.fr site has done.In their tests it beats the 8800Ultra and 9800GTX with both AA oturned on and off.Also it draws less power and is quieter.
are you talking about crysis 'the nvidia game' ? besides tweaktown said smthng similar that two of this are as powerfull as gtx280 or maybe i am wrongQuote:
Originally Posted by [XC] hipno650
Mascaras seems to be using older drivers than the french review..as well as dual vs quad. to note the differences in testing.Quote:
Originally Posted by [XC] hipno650
Quote:
Originally Posted by villa1n
im using Cat 8.5 provided with CD
i can try install new cat 8.6 ! (but i think they are the same )
edit:
Quote:
Originally Posted by onewingedangel
regards
RAD! :) I would be interested to see if it results in any difference! Appreciate your hard work.:yepp:Quote:
Originally Posted by mascaras
Quote:
Originally Posted by villa1n
tomorow more tests , im very tired , 10 hours non stop benching :p:
and evga GTX280 HC to review also !
tomorow i test the new drivers and also cod4 results
best regards
Hardware.fr is totally reliable and i'd take their results over any one that is not an XS legend any day of the week. They made this early preview because they were suspicious about the fact that the cards were for sale one week before launch, which could have been because of a poor product AMD would wan't to pass under the radar. Obviously it is not.Quote:
Originally Posted by [XC] hipno650
crysis is the one game that we can't run well at settings that most users use. and for games that people need better graphics cards for they are running 2560x1600 and want AA in which case the HD48xx's won't handle it anyway. yes other games matter but crysis is the best GPU becnhmark we have. and i hate to break it to you but most games are coded for Nvidia just as much as crysis is. the only game that is really ATI based is COJ. and not to many people play it. also you say crysis is not a good CF and SLI game. im not sure about you but the performance increase he showed form single card to dual card was extreme just as much as it should be. yes it is impressive that they made a card that can perform similar to the 9800gtx and be single slot and low power but it's not the card that many seem to be thinking it is.Quote:
Originally Posted by informal
Well, it's even more remarkable that the 4850 is beating the 8800ultra and the 9800gtx while being cheaper when like you said most of the games are coded for Nvidia cards. Maybe it's you that is not realizing what the card is.Quote:
Originally Posted by [XC] hipno650
What he is saying is not without merit, different reviews are showing contradictory results. Lets wait for an larger number of articles before concluding
From Hardware.FR review
..................1920 0xAA.............1920 4xAA
Quake Wars:ET (16xAF) Patch 1.4
9800 GTX...........101.5.................65.2 Performance hit 36.2 FPS
Ultra.................101.7.................76.5 Performance hit 25.2 FPS
GTX 260............121.3.................88.7 Performance hit 32.6 FPS
HD 4850.............86.4.................68.0 Performance hit 18.4 FPS
..................1920 0xAA.............1920 4xAA
Quake Wars:ET (16xAF) Patch 1.4
9800 GTX...........101.5.................65.2 Performance hit 36.2 FPS
Ultra.................101.7.................76.5 Performance hit 25.2 FPS
GTX 260............121.3.................88.7 Performance hit 32.6 FPS
HD 4850.............86.4.................68.0 Performance hit 18.4 FPS
Half Life 2: EP2 (16xAF) DX9 only
9800 GTX............98.3.................70.3 Performance hit 28.0 FPS
ultra..................89.2.................72.6 Performance hit 16.6 FPS
GTX 260............109.4.................90.0 Performance hit 19.4 FPS
HD 4850.............91.4.................73.1 Performance hit 18.3 FPS
Stalker (16xAF) deferred rendering, NO AA FOR ATI
9800 GTX............49.6..................2.9
Ultra..................41.0.................14.7
GTX 260.............58.9.................19.1
HD 4850.............49.0..................0.0
Rainbow 6: Vegas (HDR enabled) Shadows set to Low
9800 GTX............53.5.................30.4 Performance hit 23.1 FPS
Ultra..................53.1.................34.0 Performance hit 19.1 FPS
GTX 260.............60.0.................41.1 Performance hit 18.9 FPS
HD 4850.............66.6.................36.9 Performance hit 29.7 FPS
Oblivion (HDR)
9800 GTX............87.6.................51.2 Performance hit 36.4 FPS
Ultra..................87.4.................51.0 Performance hit 36.4 FPS
gTX 260.............94.7.................82.7 Performance hit 12.0 FPS
HD 4850.............80.2.................66.3 Performance hit 13.6 FPS
Race Driver: Grid (1.1 Patch) DX9 only
9800 GTX............73.4.................58.3 Performance hit 15.1 FPS
Ultra..................71.9.................59.5 Performance hit 27.5 FPS
GTX 260.............79.0.................67.8 Performance hit 11.2 FPS
HD 4850.............80.8.................70.0 Performacne hit 10.8 FPS
Bioshock Unreal Engine 3.0, Max Options, (no AA for ATI) DX10
9800 GTX............99.4.................42.2 Performance hit 57.2 FPS
Ultra..................93.9.................52.3 Performance hit 41.6 FPS
GTX 260............115.0.................57.5 Performance hit 57.5 FPS
HD 4850............119.0..................0.0
Company of Heroes DX10 Max Options Test Version 1.72
9800 GTX............51.0.................38.7 Performance hit 12.3 FPS
Ultra..................49.3.................41.5 Performance hit 07.8 FPS
GTX 260.............65.2.................57.2 Performance hit 08.0 FPS
HD 4850.............45.2.................38.3 Performance hit 06.6 FPS
World in Conflict Patch 1.0002, Options @ Max, DX10, 16xAF
9800 GTX............29.0.................19.0 Performance hit 10.0 FPS
Ultra..................32.0.................24.0 Performance hit 08.0 FPS
GTX 260.............36.0.................30.0 Performacne hit 06.0 FPS
HD 4850.............33.0.................27.0 Performance hit 06.0 FPS
Crysis Patch 1.21, High, DX10
9800 GTX............23.8.................11.9 Performance hit 11.9 FPS
Ultra..................24.1.................16.3 Performance hit 07.8 FPS
GTX 260.............26.9.................20.7 Performance hit 06.2 FPS
HD 4850.............22.2.................16.9 Performance hit 05.3 FPS
Test Bed
Intel Core 2 Extreme QX9770 Intel Core 2 Extreme QX9770
Asus Striker II Asus Striker II
4 Go DDR3 1066 4 GB DDR3 1066
Windows Vista SP1 Windows Vista SP1
Forceware 177.34 Forceware 177.34
Catalyst 8.5 Catalyst 8.5
Catalyst beta pour la Radeon HD 4850 Catalyst beta for the Radeon HD 4850
Sidenote:
If I made a mistake, let me know...thanks!
2 x 4850 will look good
I would like to see percentages after each performance hit, please :)Quote:
Originally Posted by Eastcoasthandle
That test system (according to you) has an awesome RAM configuration, but that's just nitpicking :p:. (you basically copied the system specs 1 too much)
BTW, is there something wrong in the drivers of the NVIDIA cards in Race driver: GRID and Bioshock as their numbers seem so low compared to the 4850? The difference is just too good to be true, other games seem more realistic.
Here we go :up:Quote:
Originally Posted by Luka_Aveiro
http://img519.imageshack.us/img519/287/hd3870jt3.jpg
http://img365.imageshack.us/img365/5...board01up5.jpg
Done by God is dead
http://forum.hardware.fr/hfr/Hardwar...5.htm#t6441334
There ya go. :D:D::rolleyes::rolleyes::ROTF::ROTF::ROTF:Quote:
Originally Posted by jas420221
http://img152.imageshack.us/img152/1...x280gx2nv9.jpg
Thank you very much, that was so sweet :p:Quote:
Originally Posted by AbelJemka
But I also would like to have nvidia's side, just for a wider comparison.
Nice work :up:
Nice to see @ a general point, HD4850 loses 10% less performance with AA on, than HD3870. :)