Is the HD 4850 worth it? Find out!

Is the HD 4850 worth it? Find out! · 23-07-2008, 02:50 AM

Finally got around to installing my brand new ASUS HD 4850. The card looks amazing and is loaded with copper. I, however, have reservations with the small fan and inadequate cooling at default settings. In this test today I am putting a BFG 9600GT OC2 against the HD 4850 to see if the upgrade was worth the amount of money spent. I have added a synthetic benchmark, a popular and most demanding strategy game, an action adventure game, most demanding shooter, most demanding 3rd person game, most demanding RPG, and ofcourse the most demanding racing game.

System Setup

Intel Q6600 2.4 GHz @ 3.2 GHz (1600 FSB)
4 GB G.Skill DDR 2 800 @ 1000 MHz
Gigabyte GA-P35-DS3L
250 GB Western Digital 16MB Cache 7200 RPM
Compaq V1100 21“ Monitor (1600*1200 @ 85 Hz)

Drivers
nVidia WHQL 175.19 Drivers for 9600GT
ATi Hotfix 8.6 Drivers for HD 4850

Operating System
Windows XP 32 Bit SP2

The Cards
BFG 9600GT OC2 @ 720 core / 1826 shader / 2100 memory
ASUS HD 4850 @ Stock

3dMark06 Default Benchmark
9600GT – 12829
HD 4850 @ Stock. 3dmark06 score of 12920
14,001 with HD 4850 @ 700 core and 2200 memory.

World In Conflict V 1.007 DX9 Max Everything

Benchmarking was done using the in game benchmark. I used it because I think it is most representative of the actual gameplay within the game and also has some nice effects to boot.

1280*1024 4xAA 16xAF
Min ---- Avg ---- Max
BFG 9600GT ---- 23 ---- 40 ---- 66
ASUS HD 4850 ---- 26 ---- 44 ---- 66
% Improvement 10%

1280*1024 8xAA 16xAF
Min ---- Avg ---- Max
BFG 9600GT ---- 23 ----- 39 ---- 64
ASUS HD 4850 ---- No Option to Run 8xAA within game

1600*1200 4xAA 16xAF
Min ---- Avg ---- Max
BFG 9600GT ---- 19 ---- 32 ---- 51
ASUS HD 4850 ---- 21 ---- 35 ---- 52
% Improvement 9.4%

1600*1200 8xAA 16xAF
Min ---- Avg ---- Max
BFG 9600GT ---- 18 ---- 31 ---- 49
ASUS HD 4850 ---- No Option to Run 8xAA within game

Observations
Game ran a bit better than 9600GT and overall I could not tell the difference in either IQ or performance.

Devil May Cry 4 Full Retail DX9

Benchmarking was done using the in game benchmark. It is almost the same as the actual game so it was convenient and worthwhile to use it as is.

1600*1200 8xAA 16xAF
BFG 9600GT
Scene 1 61.95
Scene 2 43.20
Scene 3 74.73
Scene 4 47.15

ASUS HD 4850
Scene 1 80.89
Scene 2 57.20
Scene 3 97.7
Scene 4 64.44

% Improvement
Scene 1 30.6%
Scene 2 32.4%
Scene 3 30.7%
Scene 4 36.7%

Observations
Game benchmarked faster than 9600GT yet the difference was not noticeable without looking at the fps counter. It is possible since the fps were already very high on the 9600GT that improvement from HD 4850 were not visible to my eye. However, I felt that the shadows on the 9600GT looked better than the HD 4850 overall. HD 4850 had a lot of flickering shadows (it is a game problem but more noticeable on 4850 than 9600GT).

Crysis V1.2 DX9 All High

Benchmarking was done using the GPU benchmark available in the Bin32 folder of Crysis. The benchmark tests the fastest version of the game yet still has ample scenarios to actually be representative of actual gameplay. My personal observation is that whatever is the average score in the benchmark, you can expect the game to run within 3-5 fps of the benchmark average for most of the time (exclusive of the last level of the game).

1024*768 16xAF
Min ---- Max ---- Avg
BFG 9600GT ---- 21.99 ---- 57.40 ---- 39.93
ASUS HD 4850 ---- 25.72 ---- 76.40 ---- 59.43
% Improvement 48.8%

1280*1024 16xAF
Min ---- Max ---- Avg
BFG 9600GT ---- 18.89 ---- 35.27 ---- 27.06
ASUS HD 4850 ---- 22.46 ---- 47.93 ---- 34.81
% Improvement 28.6%

1280*1024 4xAA 16xAF
Min ---- Max ---- Avg
BFG 9600GT ---- 16.45 ---- 29.39 ---- 22.68
ASUS HD 4850 ---- 17.62 ---- 36.52 ---- 27.75
% Improvement 22.4%

Observations
I still do not have a card that can run this game at 1280*1024 4xAA 16xAF hence the disappointment. Overall if I were to continue playing this game at 1024*768 then I am sure I can add 4xAA with the HD 4850 which was not possible with the 9600GT.

Gears of War V1.0 DX9 AA forced via CP

Gears of War was benched by having a fraps run through from start of the game till Marcus enters the chopper and a spider like creature comes out of the earth. The section of the game contains ample explosions and the creature entre culminates usually into low fps scenarios. Game is capped at 61 fps and fraps reports highest fps as 64. It was ensured that all explosions were in view and tutorial was skipped.

1600*1200 4xAA 16xAF
Min ---- Max ---- Avg
BFG 9600GT ---- 28 ---- 64 ---- 56.745
ASUS HD 4850 ---- 22 ---- 64 ---- 59.962
% Improvement 5.7%

Observations
This game is already so fast that a comparison is irrelevant. However, the HD 4850 did show minor stutters where the fps drop was noticeable during gameplay. It happened twice in my benchmark run but it happened nonetheless.

Witcher

Benchmarking was done in Old Vizima after when it is burning and contains approximately 10-15 NPCs on screen. It also has a conversation with Siegfried and battle with 2 Greater Brothers in front of a burning house/cottage. This section of the game was typically very heavy for any video card that I have seen and hence was used for benchmarking.

1600*1200 4xAA 16xAF
Min ---- Max ---- Avg
BFG 9600GT ---- 21 ---- 76 ---- 38.623
ASUS HD 4850 ---- 22 ---- 85 ---- 46.764
% Improvement 21%

Observations
Although the numbers show the HD 4850 providing large improvement in the game, in actual fact the difference in performance was barely noticeable and the framerate dropped at the same place in both games (while viewing the burnt cottage and slashing a Greater Brother).

GRID

Benchmarking was done on San Francisco Short Circuit with 1 lap with a Dodge Viper SRT 10 at the front of the pack. It was an error free 1 lap without any collisions and/or cars in front.

1280*1024 8xAA 16xAF
Min ---- Max ---- Avg
BFG 9600GT ---- 45 ---- 72 ---- 53.979
ASUS HD 4850 ---- 51 ---- 80 ---- 67.168
% Improvement 24.4%

1600*1200 8xAA 16xAF
Min ---- Max ---- Avg
BFG 9600GT ---- 33 ---- 54 ---- 40.275
ASUS HD 4850 ---- 24 ---- 52 ---- 42.789
% Improvement 6.24%

Redone Benches for HD 4850 w/ 8xAA ---- 29 ---- 59 ---- 47.053
Redone Benches for HD 4850 w/ 4xAA ---- 42 ---- 79 ---- 65.032
% Improvement over 8xAA 38.2%

Observations
I redid the 1600*1200 benchmark and still got the same results. This is quite strange and 1600*1200 is still not playable since I do not get 50+ fps with 8xMSAA on with 16xAF on the HD 4850. However, it should be noted that GRID looks crispier and also has better anti aliasing with the HD 4850 compared to the 9600GT.

Initial Impressions (not my final impressions, opinions may change).

After playtesting the card I can honestly not tell if I have gotten any faster card. Most of my game settings remained the same with fps counter ticking a bit faster (not noticeable in game unless fraps was running). I will do some more benches and report back. After 3 hours of testing I must say that I am a bit disappointed overall and the people who were raving about it as being faster than a 9800GTX must have something that I don't in my rig.

I will do some comparisons with online sites later. For the money it does not pack the performance. If you are on a 9600GT/8800GT/9800GTX series then there is little sense in making this upgrade. I will update more soon.

Summary

Pros:
Copper Cooling
Silent Operation
DX10.1 Support
Some performance improvement in games

Cons:
Runs damn hot (85-88C Load)
Price/performance ratio for 9600GT/8800GT/9800GTX users not there

Verdict

Unless you are upgrading from 7XXX series or HD 3850/3870 series or lower there is not much sense in getting an HD 4850 as a single card solution.

However, if you are in the market for a new card and have 16,500 Rs then HD 4850 is the one to purchase.

Personally getting a used 9600GT for now is good enough till prices on HD 4850 drop further.

corruptd · 23-07-2008, 03:38 AM

Thanks for sharing

superfrog · 23-07-2008, 03:52 AM

I would say the real impression this card is making across any review is when in the Cross Fire mode, thats where it really packs the punch with GTX 260 or perhaps GTX 280. But in a single card solution i dont think in real world situation there is any difference.

ali.tayyab · 23-07-2008, 04:17 AM

i got myself a His 4870, havent done any really deep testing, but it seems to be on par with a highly oc'd 8800gtx as far as 3dmark 06 is concerend.
i am gonna do some more tests to see how it fares up against my old card.

i think something is holding ur card back. it should show greater performance boosts esp in Gears of war.
(edit i just read: the frame cap was in place)

superfrog · 23-07-2008, 04:24 AM

could be the motherboard ???

ali.tayyab · 23-07-2008, 07:43 AM

in all probability the game was locked at 60fps.
thus the closeness of the results!

Kashif · 23-07-2008, 10:13 AM

congrats kac; and ali bhai on th 4870. i was almost ordering the 4850 from abroad, but since it runs hot, things could go bad sooner; and at least there will be warranty buying from here.

since i'll be moving from 7900gs to 4850, the leap will be decent. anyway, will give my thoughts some time as im not in a hurry.

fajoo24 · 23-07-2008, 11:59 AM

nice review there kac....

@kashif...yar heating issue is not a problem anymore...u may consult Shahzad, who has bought his 4850 recently and after tweaking fan speed to full, his idle temps r 55 degrees and load temps are 75 degrees max......

mave_3 · 23-07-2008, 12:07 PM

Congrats to u to @ali..
where are the benchies?
where did u get the card from?
and whats the overclock news?

syed_shahzad786 · 23-07-2008, 01:50 PM

i`m getting much better performance from this....may be ur drivers problems

23-07-2008, 02:50 AM	#1
Is the HD 4850 worth it? Find out! KickAssCop Member Join Date: Sep 2007 Location: Lahore Posts: 112 KickAssCop 23-07-2008, 02:50 AM Rating: () Finally got around to installing my brand new ASUS HD 4850. The card looks amazing and is loaded with copper. I, however, have reservations with the small fan and inadequate cooling at default settings. In this test today I am putting a BFG 9600GT OC2 against the HD 4850 to see if the upgrade was worth the amount of money spent. I have added a synthetic benchmark, a popular and most demanding strategy game, an action adventure game, most demanding shooter, most demanding 3rd person game, most demanding RPG, and ofcourse the most demanding racing game. System Setup Intel Q6600 2.4 GHz @ 3.2 GHz (1600 FSB) 4 GB G.Skill DDR 2 800 @ 1000 MHz Gigabyte GA-P35-DS3L 250 GB Western Digital 16MB Cache 7200 RPM Compaq V1100 21“ Monitor (16001200 @ 85 Hz) Drivers* nVidia WHQL 175.19 Drivers for 9600GT ATi Hotfix 8.6 Drivers for HD 4850 Operating System Windows XP 32 Bit SP2 The Cards BFG 9600GT OC2 @ 720 core / 1826 shader / 2100 memory ASUS HD 4850 @ Stock 3dMark06 Default Benchmark 9600GT – 12829 HD 4850 @ Stock. 3dmark06 score of 12920 14,001 with HD 4850 @ 700 core and 2200 memory. World In Conflict V 1.007 DX9 Max Everything Benchmarking was done using the in game benchmark. I used it because I think it is most representative of the actual gameplay within the game and also has some nice effects to boot. 12801024 4xAA 16xAF Min ---- Avg ---- Max BFG 9600GT ---- 23 ---- 40 ---- 66 ASUS HD 4850 ---- 26 ---- 44 ---- 66 % Improvement 10%* 12801024 8xAA 16xAF Min ---- Avg ---- Max BFG 9600GT ---- 23 ----- 39 ---- 64 ASUS HD 4850 ---- No Option to Run 8xAA within game 16001200 4xAA 16xAF Min ---- Avg ---- Max BFG 9600GT ---- 19 ---- 32 ---- 51 ASUS HD 4850 ---- 21 ---- 35 ---- 52 % Improvement 9.4% 16001200 8xAA 16xAF Min ---- Avg ---- Max BFG 9600GT ---- 18 ---- 31 ---- 49 ASUS HD 4850 ---- No Option to Run 8xAA within game Observations* Game ran a bit better than 9600GT and overall I could not tell the difference in either IQ or performance. Devil May Cry 4 Full Retail DX9 Benchmarking was done using the in game benchmark. It is almost the same as the actual game so it was convenient and worthwhile to use it as is. 16001200 8xAA 16xAF BFG 9600GT Scene 1 61.95 Scene 2 43.20 Scene 3 74.73 Scene 4 47.15 ASUS HD 4850 Scene 1 80.89 Scene 2 57.20 Scene 3 97.7 Scene 4 64.44 % Improvement Scene 1 30.6% Scene 2 32.4% Scene 3 30.7% Scene 4 36.7% Observations* Game benchmarked faster than 9600GT yet the difference was not noticeable without looking at the fps counter. It is possible since the fps were already very high on the 9600GT that improvement from HD 4850 were not visible to my eye. However, I felt that the shadows on the 9600GT looked better than the HD 4850 overall. HD 4850 had a lot of flickering shadows (it is a game problem but more noticeable on 4850 than 9600GT). Crysis V1.2 DX9 All High Benchmarking was done using the GPU benchmark available in the Bin32 folder of Crysis. The benchmark tests the fastest version of the game yet still has ample scenarios to actually be representative of actual gameplay. My personal observation is that whatever is the average score in the benchmark, you can expect the game to run within 3-5 fps of the benchmark average for most of the time (exclusive of the last level of the game). 1024768 16xAF Min ---- Max ---- Avg BFG 9600GT ---- 21.99 ---- 57.40 ---- 39.93 ASUS HD 4850 ---- 25.72 ---- 76.40 ---- 59.43 % Improvement 48.8%* 12801024 16xAF Min ---- Max ---- Avg BFG 9600GT ---- 18.89 ---- 35.27 ---- 27.06 ASUS HD 4850 ---- 22.46 ---- 47.93 ---- 34.81 % Improvement 28.6%* 12801024 4xAA 16xAF Min ---- Max ---- Avg BFG 9600GT ---- 16.45 ---- 29.39 ---- 22.68 ASUS HD 4850 ---- 17.62 ---- 36.52 ---- 27.75 % Improvement 22.4%* *Observations* I still do not have a card that can run this game at 12801024 4xAA 16xAF hence the disappointment. Overall if I were to continue playing this game at 1024768 then I am sure I can add 4xAA with the HD 4850 which was not possible with the 9600GT. Gears of War V1.0 DX9 AA forced via CP Gears of War was benched by having a fraps run through from start of the game till Marcus enters the chopper and a spider like creature comes out of the earth. The section of the game contains ample explosions and the creature entre culminates usually into low fps scenarios. Game is capped at 61 fps and fraps reports highest fps as 64. It was ensured that all explosions were in view and tutorial was skipped. 16001200 4xAA 16xAF Min ---- Max ---- Avg BFG 9600GT ---- 28 ---- 64 ---- 56.745 ASUS HD 4850 ---- 22 ---- 64 ---- 59.962 % Improvement 5.7%* *Observations* This game is already so fast that a comparison is irrelevant. However, the HD 4850 did show minor stutters where the fps drop was noticeable during gameplay. It happened twice in my benchmark run but it happened nonetheless. Witcher Benchmarking was done in Old Vizima after when it is burning and contains approximately 10-15 NPCs on screen. It also has a conversation with Siegfried and battle with 2 Greater Brothers in front of a burning house/cottage. This section of the game was typically very heavy for any video card that I have seen and hence was used for benchmarking. 16001200 4xAA 16xAF Min ---- Max ---- Avg BFG 9600GT ---- 21 ---- 76 ---- 38.623 ASUS HD 4850 ---- 22 ---- 85 ---- 46.764 % Improvement 21%* *Observations* Although the numbers show the HD 4850 providing large improvement in the game, in actual fact the difference in performance was barely noticeable and the framerate dropped at the same place in both games (while viewing the burnt cottage and slashing a Greater Brother). GRID Benchmarking was done on San Francisco Short Circuit with 1 lap with a Dodge Viper SRT 10 at the front of the pack. It was an error free 1 lap without any collisions and/or cars in front. 12801024 8xAA 16xAF Min ---- Max ---- Avg BFG 9600GT ---- 45 ---- 72 ---- 53.979 ASUS HD 4850 ---- 51 ---- 80 ---- 67.168 % Improvement 24.4%* 16001200 8xAA 16xAF Min ---- Max ---- Avg BFG 9600GT ---- 33 ---- 54 ---- 40.275 ASUS HD 4850 ---- 24 ---- 52 ---- 42.789 % Improvement 6.24%* Redone Benches for HD 4850 w/ 8xAA ---- 29 ---- 59 ---- 47.053 Redone Benches for HD 4850 w/ 4xAA ---- 42 ---- 79 ---- 65.032 % Improvement over 8xAA 38.2% *Observations* I redid the 16001200 benchmark and still got the same results. This is quite strange and 16001200 is still not playable since I do not get 50+ fps with 8xMSAA on with 16xAF on the HD 4850. However, it should be noted that GRID looks crispier and also has better anti aliasing with the HD 4850 compared to the 9600GT. Initial Impressions (not my final impressions, opinions may change). After playtesting the card I can honestly not tell if I have gotten any faster card. Most of my game settings remained the same with fps counter ticking a bit faster (not noticeable in game unless fraps was running). I will do some more benches and report back. After 3 hours of testing I must say that I am a bit disappointed overall and the people who were raving about it as being faster than a 9800GTX must have something that I don't in my rig. I will do some comparisons with online sites later. For the money it does not pack the performance. If you are on a 9600GT/8800GT/9800GTX series then there is little sense in making this upgrade. I will update more soon. Summary Pros: Copper Cooling Silent Operation DX10.1 Support Some performance improvement in games Cons: Runs damn hot (85-88C Load) Price/performance ratio for 9600GT/8800GT/9800GTX users not there Verdict Unless you are upgrading from 7XXX series or HD 3850/3870 series or lower there is not much sense in getting an HD 4850 as a single card solution. However, if you are in the market for a new card and have 16,500 Rs then HD 4850 is the one to purchase. Personally getting a used 9600GT for now is good enough till prices on HD 4850 drop further. __________________ Intel Q6600 @ 3.2 GHz (400X8 FSB) \| Gigabyte GA-P35-DS3L Rev 2.0 \| Gigabyte Radeon HD 4870 1 GB \| 250 GB WD SATA2 16 MB \| 80 GB Seagate SATA 8 MB \| 4 GB G.Skill DDR2 800 @ 1000 5-5-5-15 \| LG 20X DVD RW \| Samsung 226BW LCD \| Logitech MX 518 \| FSP 600W Quad Rail PSU Last edited by KickAssCop : 25-07-2008 at 01:35 PM.
Views: 969

Sponsor Ads Sponsored links
PK Forum Community Advertisement