MSI 8800GTS 320MB vs Sapphire HD2900Pro 512MB battleground

Status
Not open for further replies.
Ok people, have a little treat for you.
With the introduction of 2900Pro from DAMMIT's stable the graphic card scene just became interesting with finally a card to compete against 8800GTS 320MB which has become extremely popular with mid budget enthusiast who cant spend fortune on 8800GTX/Ultra or even 8800GTS 640MB.
This review will be somewhat different than all others you see on the internet. It wont be run on your daily plain jane pc. The test setup will be something that will show how these cards perform when CPU and memory bottleneck is done away with and I will only concentrate at high resolution gaming. With the increasing popularity of 20 and 22 inch screens 1680x1050 has become popular choice for gamers and many other will be thinking of switching to these LCDs but dont exactly know how these cards will perform at this resolution. Not many reviews concentrate at this resolution for reason which i fail to understand.

Anyway lets begin.

ati-vs-nvidia.jpg

[BREAK=MSI 8800GTS 320MB]

MSI 8800GTS 320MB

I received MSI 8800GTS 320MB for testing purpose.
8800GTS by default is clocked at 500Mhz core and 800Mhz GDDR3 memory with 96 shader processors clocked at 1200Mhz.
Memory interface is 320Bit.
MSI also has the OC edition which comes factory overclocked at 575Mhz core, 850Mhz memory and 1320Mhz shader clock.

The card is pretty much standard 8800GTS. Nothing special here.
[BREAK=Sapphire HD2900Pro]

Sapphire HD2900Pro

ATI 2900Pro is latest kid on the block in GPU market. This card makes a very lucrative purchase as its nothing but the underclocked 2900XT.

It comes with same 320 steam processors, 512bit memory interface which is same as 2900XT.
Its clocked at 600Mhz core and 800Mhz memory. 2900XT on the other hand is clocked at 725Mhz core and 825Mhz memory.
Physically the card looks identical to any 2900XT.

Feature wise this card has something special to offer. It comes with HDMI dongle. And unlike Nvidia's offering, this is able to process audio and pass it over the HDMI interface which is great if you are looking to hookup your PC to HDMI equipped LCD / Plasma TVs. Just 1 cable will carry your audio and video signals.
You do not require extra sound card for this. The card has secondary audio functionality and is capable of producing 5.1 surround sound output over HDMI.
Here is a quick comparison between the features offered by both the cards.

[BREAK=Photoshoot MSI 8800GTS 320MB]
320MB
Photoshoot MSI 8800GTS


100_2715.JPG


100_2716.JPG

100_2717.JPG

100_2718.JPG

100_2719.JPG

100_2720.JPG

[BREAK=Photoshoot Sapphire HD2900Pro]

Photoshoot Sapphire HD2900Pro

100_2702.JPG

100_2703.JPG

100_2704.JPG

100_2705.JPG

100_2707.JPG

100_2709.JPG

100_2710.JPG

100_2711.JPG
[BREAK=System Setup]

System Setup

Here is the system on which these cards were tested.
CPU : Intel Core2Quad E6600 @ 3.6Ghz (overclocked)
Motherboard : ABIT IP35 Pro
Ram : 4GB PDP DDR2 800
HDD : 4x320GB
PSU : OCZ Gamexstream 700W
Sound Card : Creative SBLive 24Bit (The weak point in the system ;) )
Operating System : Windows Vista Ultimate 64Bit

Drivers used

ATI Catalyst 7.10 Final
Nvidia Forceware Vista 163.75

System Preparation

The care was taken to ensure the system resembles a regular gaming pc environment and not some highly tweaked stripped down windows with many features disabled.
Clean install of Windows Vista Ultimate X64 retail was used.
Windows Vista was updated with all critical patches and DX9 and DX10 were updated to latest release. Motherboard and Sound drivers were also updated to latest WHQL versions available from the manufacturer.
Avast Free home edition antivirus was installed as well and was allowed to start its service at boot but its On Access Protection was disabled while testing to endure that the whatever amount of processor time used by this does not affect the benchmarks.
No other items were allowed to start at boot.
Aero interface was enabled and kept on while benchmarking. This somewhat decreases the scores but there is no point in turning everything off.

Softwares Used

Rivatuner 2.05
Fraps
HardwareOC benchmarking tool

[BREAK=How I tested]

How I tested


Drivers

I left the drivers on both the cards to their default settings. Only Vsync was turned off. ATI driver is by default set to maximum quality. Same is not the case with Nvidia forceware which is by default to 1 notch down from the highest quality settings. But it was left alone.

Overclocking
Overclocking was done using Rivatuner 2.05.
On ATI its very simple, just Core and memory speeds are changed.
On 8800GTS however there is shader clock as well and with this version of Rivatuner you can change that in Sync with the core or separately. I set it to overclock in sync with the core clock. This is what matters more. overclocking shader clock benefits the 8800GTS more than anything else.
3Dmark 2006

3DMark 2006 was updated with latest patch and was allowed to run at its default settings. This is purely synthetic benchmark so I did not spend too much time on this.
Bioshock

bioshock%20dvd.jpg

Bioshock retail version did not come with any benchmarking tool, neither it had any provision for demo recording. This made benching this game difficult. I cleared the first level of all the enemies and then ran though it. Fraps was used to record the framerates. This means these are not very accurate results but the recording was made for long time to ensure each part of the level was accessed. That is best I was able to do.
The game was set to max details @ resolution of 1680x1050.
Prey

prey%20dvd.jpg

This is another very popular game widely used for GPU benchmarking. To benchmark this I used HardwareOC benchmarking tool. Game was set to max details @ 1680x1050 resolution. Guru3D demo built into the benchmarking tool was used. 4XAA and 16XAF settings was used.
Serious Sam 2

serious_sam_2.jpg

Another very popular arcade game. Huge open areas, HDR, plenty of action on screen makes it ideal game to benchmark the cards. Framerate is usually very high without the Anti Aliasing and default auto detech game settings. But with max details 4XAA and 16XAF the game can be really hard on even modern GPUs.
Game was updated to latest 2.070 patch.
HardwareOC benchmarking tool proved to be little annoying with this game as the settings inputed in the tool did not stick when the game actually ran resulting in very high framerate. So I used built in demos of the game with the Fraps.
World In Conflict

WIC%20dvd.jpg

This is highly anticipated new game release which bring out the real eyecandy and DX10 really does make difference in this game. DX10 mode is very hard on the GPU and you will find that out in the results. Game was unplayable on 8800GTS @ stock speeds.
I used max details in both DX9 and DX10 mode and games built in benchmarking tool.

Unreal Tournament 3 Demo

ut3.jpg

This is probably what most of you will be interested in. The demo is simply stunning and shows what to expect from this game. The gameplay was simply amazing and even though low resolution textures were used for ground the game was simply great looking and ran very well.
All in game settings were set to max. No AA was used as its not supported in the demo and there is no way to force it on the 2900Pro. So to keep the gamefield even it was left out.
Game was ran at 1680x1050. This is one of the few games which is really using the multi core CPUs. All 4 cores were partially loaded during the gameplay.

Enough talks, lets head to the numbers.

[BREAK=3DMark 2006 results]

3DMark 2006 results

3DMARK.jpg

There is no competition here. HD2900Pro clearly outperforms 8800GTS 320MB in this benchmark. but lets not give too much importance to this. Its just synthetic benchmark. Lets move on to gaming benchmarks.

[BREAK=Bioshock results]

Bioshock results

bioshock.jpg
8800GTS has clear advantage in this game as well. But game was again very smooth on both cards. The performance of this game is steadily increasing on ATI cards but they still have some catching up to do.

[BREAK=Prey results]

Prey results

Prey.jpg
As you can see the 8800GTS has clear advantage at stock as well as overclocked speeds. The game is not that heavy on the cards so as far as gameplay goes, neither card has any issues in this game.
[BREAK=Serious Sam 2 results]

Serious Sam 2 results

SS2.jpg

The results came as surprise to me as well. I double and triple checked them again to ensure it was correct and even tried older forcewares. Results were consistent. I again went and checked results with some popular sites like guru3d to see if I made some error but then noticed none of them had manually set the game settings to max and enforced AA. So it seems results are good and with AA enabled ATI has clear advantage and in games like SS2 high framerate is required. Action is fast and you need high FPS and this affects the gameplay. Without AA the frame rate is very high and Nvidia has slight advantage but with AA the figures speaks themself.

HD2900Pro comprehensively outperforms 8800GTS in this game.

[BREAK=World In conflict results]

World In conflict results

WIC%20chart.jpg
Keep in mind the prerecorded demo used by this game for benchmarking has plenty of things going on and camera always moving to the action. This game is RTS and thats not exactly how we play RTS. So usually in game the frame rate was higher than that in the benchmark but still 8800GTS did very very bad in this game. It was somewhat playable when oced but this game made card soo hot that at one point it reached 98C in Air conditioning when other games hardly took it beyond 85C.

Due to limited time for which I had the game DVD with me, I was unable to bench this game on ATI at usual 790/990 oced speeds. But even at lower speeds the card was able to outperform the 8800GTS. The 8800GTS card couldnt even complete the DX10 run at stock speeds.

2900Pro had advantage in this game.

[BREAK=Unreal Tournament 3 Demo results]

Unreal Tournament 3 Demo results


UT3-Heatray.jpg


UT3-Shangrila.jpg


UT3-Suspense.jpg

As you can see even though ATI has slight lead its not much. Both cards are able to play this game without any issue. Gameplay was smooth on both cards generally. 8800GTS did stutter in some places when hell broke loose with 4-5 players at same place shooting like madman but it was momentary stutter and it went away. It can be annoying at times but it should not matter much.

[BREAK=Conclusion]

Conclusion

As the results show, the two cards are well matched most of the times. But 8800GTS 320MB failed 1 critical game. That is one game which really shows the eyecandy in DX10 mode. The cinematic experience on ATI card was breathtaking. And it was surprising to see the 8800GTS do soo poorly in that game. This might be because of just 320MB ram which can become issue in near future with Crysis just around the corner.
ATI did fell behind in 2 games but there was not a single game which was unplayable on this card.
I had suspected this result for a while but noone out there tested the games the way I wanted.
Take example of Serious Sam 2. All benchmarks out there showed Nvidia leading by some margin in this game. But did anyone bothered to see if the benchmarking tool is actually benchmarking as it should? When all settings were manually set to max the tables seems to turn in favor of HD2900Pro even with AA enabled. This is sharp contrast to the results from reputed reviewers. And it made me run my tests again and again until i was convinced.

What does this mean? If you go out to buy a card in Rs.17000 range today then get the HD2900Pro and be happy. I am not defending ATI without a reason here. Its a better product. I am not even gonna go into the extra feature set on HD2900Pro. Leave that to individual. 2900Pro is basically 2900XT at 40% less cost. It may not compete against 8800GTX that well but it is a very strong card in its price range.

Its able to play each game out there at this moment smoothly @ 1680X1050. If thats what you want then you know my recommendation.

Nothing against 8800GTS 320MB. It was unchallenged king of mid budget cards till now. Maybe 320MB is what hurting this GPU. But it is hurting it in gfx intensive games.

Rest is upto an individual in the end.
Reviewer's Recommendation : ATI HD2900Pro

Indicative Indian Market Prices

Sapphire HD2900Pro : Rs.17000 approx.

MSI 8800GTS 320MB : Rs.18500 approx.
Special thanks to Mr Anish and Ashish Modi from Mnemonic Systems, Borivali (w), Mumbai who made the MSI 8800GTS available for testing purpose. Without this card this review would not have been possible.
 
  • Like
Reactions: 10 people
greenhorn said:
could you post a few FSX stats please :D
yulp me too :P
29ym7bt.jpg

hey seen different

but some games are hardy u beat in future proof at different setting so ati
could not improve some driver issue
The GeForce 8800 GTS 320 MB finally has some strong competition - so strong, in fact, that it beats it in the price segment. The Radeon 2900 PRO is a barely disguised 2900 XT with slightly lowered clock speeds and very similar performance to its faster brother. With a relatively mild overclock, you'll get 2900 XT performance from this card. The only downside is that the 2900 PRO is probably not going to be around for a long time, so get them while they're hot!

but nvida and ati are both outperform different games :P
 
Its highly unlikely that I would be able to get my hands on 2 2900Pros. I do have crossfire capable board ( Multiple boards actually ) but no way to source another HD2900PRO. And without testing I would not even try to guess ;)

But its highly illogical to have 2x 2900PRO instead of 1x 8800GTX to be honest.

I personally prefer single card environment.
 
Excellent Review Indeed!! The 2900 Pro looks hot.. droolmaal!
P.S : The MSI 8800GTS OC costed me 18,200 all incl. when purchased in Aug '07
well then i am ready to sell my 3 month old MSI 8800GTS - HD - OC for 13,500:cool2: .
i will be in Japan for next 6 months so i wont need this card...
 
phew forgot ati thread but i am not fan of ati sorry for hurt ati :bleh:

me amd-intel fan :P
 
Status
Not open for further replies.