ATi 4xxx Series Review

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
The 4850 seems like a pretty decent card. I can't wait to see how the 4870 fairs. For the most part, with high resolution and AA, it appears to be on par with the 9800GTX. But, this is only one review so far... Can't wait to see more pop up.
 

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,001
126
I am quite sure I want a 4870, but those numbers at 1920x1200 look really good. I am only running a 22" (1680x1050) monitor... probably not a whole lot to gain by going with more power. Decisions decisions. :)
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
across all those games tested, it was BETTER than 9800gtx and, in fact, was better than an ultra! I'd like to see more reviews before rushing to judgement, but that is some pretty impressive stuff. If the AA numbers were correct then I think that we're safe in saying "AA bug? WHAT AA bug?"
 

Hauk

Platinum Member
Nov 22, 2001
2,806
0
0
4850 CROSSFIRE

IMPRESSIONS
These cards feature good build quality, with a single slot solid copper heatsink. In one pic below you'll notice exaust vents on the side as well as the back of the heatsink. Pics we've seen don't easily reveal these exaust slots. A good bit of exaust blows thru these directed towards the side panel. You'll see I mounted a 120mm fan near the front of the card avoiding the exaust slots. The fans never spun up past default speed which is low and quiet. There are no fan speed or shader adjustments in CCC.

With good airflow they're idling around 63c, with load temps about 74c. Under load, touching the heatsink will burn your finger after about two seconds. The cards run pretty warm thanks to the single slot cooler, but they run very stable with zero lockups or issues. The driver appears to throttle down the cards to 500 core 750 mem at idle. See CCC pic below under current clock settings. GPU-Z also reports the drop. So far I'm impressed with 4850's in crossfire.

I pulled the heatsinks off for inspection. On the GPU they use the standard grey thermal paste that looks silver based but isn't. Application patch was on par with every other heatsink I've pulled. They used the rubbery thermal pads for memory and the thermal paste/fiber layer stuff for the PWM modules. I used Arctic Ceramique on the gpu and reused everything other TIM's as they were in perfect condition. There were no significant drops in idle or load temps.

OVERCLOCKING
CCC's auto-tune feature was accurate in finding optimal overclocks. Anything higher produced lockups. Benchmarks were ran at 660 core 1050 memory, but the overclock shown is fine with gaming:

Default Clocks 625 core 933 memory
Auto-Tune Overclock 690 core 1153 memory
Take a look at the core temp during Auto tuning. Toasty!



TEST SYSTEM

- Dell 24" Ultrasharp 2408WFP
- Antec P182
- C2D E8400 @ 3.80
- Asus Rampage Formula x48
- 8GB Muskin Redline DDR2 1000 @ 1015 5-5-5-12, 7 performance value (per AT Rampage board review)
- PowerColor 4850's in CF clocked to 660 core 1050 mem
- Thermaltake Toughpower 850W PSU
- Catalyst 8.6 Betas
- Vista 64 Premium SP1


PICS
4850
Twins
Exaust Vents
Installed in Rig
Installed in Rig
Overkill Airflow
CCC
CPU & GPU Screens


BENCHIES

3DMARK06:
18059 3DMarks

*Edit* a few extra marks with Catalyst 8.6 Final and 680core 1100mem
18375 3DMarks


VANTAGE build 1.0.1:
P10431
GPU = 11666
CPU = 7916

X5314
GPU = 5226
CPU = 7831


MASS EFFECT:
FRAPS = 61fps
1920 x 1200, max detail settings, vsync disabled, avg 61fps


ASSASSIN'S CREED:
1920 X 1200, default driver settings, game settings maxed
184 fps
173 fps
162 fps


CRYSIS:
1024 x 768, dx9, high, no AA
Min = 28.68
Max = 96.29
Avg = 69.44

1280 x 1024, dx9, high, no AA
Min = 32.59
Max = 86.31
Avg = 62.67

1680 x 1050, dx9, high, no AA
Min = 26.83
Max = 56.83
Avg = 43.84

1900 x 1200, dx9, high, no AA
Min = 24.93
Max = 56.83
Avg = 43.84

1024 x 768 dx9, high 4x AA
Min = 26.28
Max = 81.06
Avg = 59.21

1280 x 1024 dx9, high, 4x AA
Min = 25.47
Max = 65.51
Avg = 49.07

1680 x 1050 dx9, high, 4x AA
Min = 19.70
Max = 47.53
Avg = 36.39

1900 x 1200 dx9, high, 4x AA
Min = 16.16
Max = 37.83
Avg = 29.97
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
nobody has been able to get the demo to run in dx10 with xfire 4850's yet to the best of my knowledge. maybe the cat 8.6 changed that, however.
 

Aberforth

Golden Member
Oct 12, 2006
1,707
1
0
Originally posted by: bryanW1995
nobody has been able to get the demo to run in dx10 with xfire 4850's yet to the best of my knowledge. maybe the cat 8.6 changed that, however.

I didn't know that, actually I am waiting for 4870 benchmarks, if it beats GTX260 I might buy it.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
Originally posted by: bryanW1995
across all those games tested, it was BETTER than 9800gtx and, in fact, was better than an ultra! I'd like to see more reviews before rushing to judgement, but that is some pretty impressive stuff. If the AA numbers were correct then I think that we're safe in saying "AA bug? WHAT AA bug?"

better across all those games? that is not what I read.
4850 vs 9800GTX
across all those games it was better with AA
On some games it was better without AA. (usually by a fair margin)
On most games it was worse without AA. (usually by a small margin)

Either way it's performance is very close to the 9800GTX... but it does take less power, and it has a lower MSRP.
So overall a better choice. (unless nvidia lowers their price)
 

spittledip

Diamond Member
Apr 23, 2005
4,480
1
81
Originally posted by: taltamir
Originally posted by: bryanW1995
across all those games tested, it was BETTER than 9800gtx and, in fact, was better than an ultra! I'd like to see more reviews before rushing to judgement, but that is some pretty impressive stuff. If the AA numbers were correct then I think that we're safe in saying "AA bug? WHAT AA bug?"

better across all those games? that is not what I read.
4850 vs 9800GTX
across all those games it was better with AA
On some games it was better without AA. (usually by a fair margin)
On most games it was worse without AA. (usually by a small margin)

Either way it's performance is very close to the 9800GTX... but it does take less power, and it has a lower MSRP.
So overall a better choice. (unless nvidia lowers their price)

Except for the dual slot cooler on the 9800GTX if that matters to some
 

Munky

Diamond Member
Feb 5, 2005
9,372
0
76
I don't care which card is faster without AA, that's almost as bad as saying which card is better at medium settings, no shadows, no HDR, etc. I use AA whenever possible, and if the 4850 is faster with AA, then as far as I'm concerned the 4850 is faster, PERIOD.
 

Quiksilver

Diamond Member
Jul 3, 2005
4,725
0
71
Originally posted by: munky
I don't care which card is faster without AA, that's almost as bad as saying which card is better at medium settings, no shadows, no HDR, etc. I use AA whenever possible, and if the 4850 is faster with AA, then as far as I'm concerned the 4850 is faster, PERIOD.

QFT.
 

nitromullet

Diamond Member
Jan 7, 2004
9,031
36
91
Nice job, Steel Six. :beer:

Anyone know what's up with some of the missing DX10 4xAA benchmarks in the hardware.fr article? I thought with being DX10.1 compliant it would have to support AA in at least any titles the GeForce cards could do AA in...
 

chewietobbacca

Senior member
Jun 10, 2007
291
0
0
Those were UT3 engined games and apparently ATI cards can't get AA enabled for them in DX10 but Nvidia can force it through their control panel
 

Avalon

Diamond Member
Jul 16, 2001
7,567
156
106
With a 4850 at stock and an E8400 @ 4Ghz, I got 12403 in 3dmark 06. Working on benches at the moment, but so far looks promising. The gains in Crysis over my 3850 are very nice.
 

nitromullet

Diamond Member
Jan 7, 2004
9,031
36
91
Disappointing for ATI about UT3 titles, I didn't know that. STALKER isn't a UT3 engine game though... On the brighter side though, those Oblivion and RaceDriver GRID numbers are through the roof for a $199 card.
 

Sylvanas

Diamond Member
Jan 20, 2004
3,752
0
0
Originally posted by: chewietobbacca
Those were UT3 engined games and apparently ATI cards can't get AA enabled for them in DX10 but Nvidia can force it through their control panel

Nope, AA has been enabled in UE3 games since Cat 8.3. There was then a fix in 8.4 for AA with Crossfire in UE3. AA is to be forced in the CCC for UE3 games.
 

lopri

Elite Member
Jul 27, 2002
13,310
687
126
Why are ATI/AMD cards so strong for Oblivion? It's been that way forever except when G80 debuted. The card (HD 4850) looks very good for the price. I am going to be patient, though, and wait for the 4870.
 

Quiksilver

Diamond Member
Jul 3, 2005
4,725
0
71
Originally posted by: Sylvanas
Originally posted by: chewietobbacca
Those were UT3 engined games and apparently ATI cards can't get AA enabled for them in DX10 but Nvidia can force it through their control panel

Nope, AA has been enabled in UE3 games since Cat 8.3. There was then a fix in 8.4 for AA with Crossfire in UE3. AA is to be forced in the CCC for UE3 games.

I take it they are unaware this was fixed and didn't test it.
 

tuteja1986

Diamond Member
Jun 1, 2005
3,676
0
0
http://www.hardware.fr/article...ti-radeon-hd-4850.html

Well i will end up getting 4870 for $299 next month unless nvidia drops the price of GT280 to $500.

For a card that is going for a $189 not bad result at all. ATI did fix their GRID performance issue but GRID problem with 38XX is still their and so it runs slow on my machine.

Crysis 1920x1200 4xAA
GT280 : 24.7
GT260 : 20.7
4850 : 16.9

GRID 1920x1200 4xAA
GT280 : 82.3
GT260 : 67.8
4850 : 70.0

World in confilict 1920x1200 4xAA
GT280 : 34
GT260 : 30
4850 : 27

Oblivion 1920x1200 4xAA
GT280 : 71.3
GT260 : 60.1
4850 : 66.3
 

Cookie Monster

Diamond Member
May 7, 2005
5,161
32
86
Link

We are excited to announce the new NVIDIA GeForce 9800 GTX+ GPU that sets a new price performance bar in the GeForce 9 family.

The GeForce 9800 GTX+ takes the highly efficient G92-based GeForce 9800 GTX and offers unbeatable value for value conscious gamers.

Key product points:

· GeForce 9800 GTX+ is the best single card price/performance at $229; the original GeForce 9800 GTX will transition to $199
· Processor clock of 1836MHz and graphics clock of 738MHz
· GeForce PhysX support! World?s only physics API for both the CPU and the GPU!
· Support for CUDA-based applications, including Folding@Home and Badaboom video transcoding applications.
· Support for 2-way and 3-way SLI

The GeForce 9800 GTX+ supports CUDA, the industry?s most widely adopted GPU Computing language. CUDA is an extension of C, allowing programmers to quickly port parallel applications to the GPU. With CUDA, the GeForce 9800 GTX can greatly accelerate video transcoding and protein folding. We recommend you check out the Badaboom transcoding application and the Folding@Home client on the FTP site to see the tangible results that GPU Computing provides.

The price wars have officially begun. Good times for the consumer. I dont think the GPU market has been this fierce with price before.

Merged in from another thread

-ViRGE
 

Cookie Monster

Diamond Member
May 7, 2005
5,161
32
86
Impressive performance improvements. Im more interested in how ATi managed this one.

Looks like they are back in the game.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,770
1,475
136
Originally posted by: nitromullet
Disappointing for ATI about UT3 titles, I didn't know that. STALKER isn't a UT3 engine game though.

Both engines use deferred shading, that's what screws AA up. It is possible to force AA in games that use deferred shading, but it's costly. Luckily for AMD, DX10.1 allows them to use deferred shading and AA without the extra cost. Unfortunately, there are a distinct lack of titles running DX10.1

Deferred shading is the future though. After DX10.1, or failing that, DX11 becomes the standard almost all games will begin using deferred shading as it is much more efficient. When that day comes you'll see 4850s beating GTX 280s when AA is >= 4x (unless the 4850 exhausts its video memory of course!)