Sapphire HD 2900 XT
Test Setup and Notes
Published: 9th July 2007 | Source: Sapphire | Price: |
Author: Matt Kemp
Source: Sapphire
Test Setup
To test these high-end cards I set a PC that gives as little of a bottleneck as possible. Please note I am using the numbers from a previous review on the 8800 GTS shootout.
The Test setup used for the 2900XT is:
CPU: Intel Core2Duo E6700 @ 3.33Ghz
Motherboard: XFX 650i Ultra
RAM: 2gb Mushkin XP2-6400 4-4-4-12
Sound: Creative Labs X-Fi fatal1ty FPS
HDD: OS - 160gb Hitachi Deskstar SATA II
HDD: Gaming - 2 x 40gb Hitachi Deskstars in RAID 0
Power: Silverstone Zeus 560w
Case: Silverstone TJ09
Cooling: Scythe Infinity
The setup for the nVidia cards was:
CPU: Intel Core2Duo E6700 @ 3.33Ghz
Motherboard: abit IN9 32x max Motherboard
RAM: 2GB Mushkin XP2-6400 4-4-4-12
Sound: Creative Labs X-Fi fatal1ty FPS
HDD: OS - 160gb Hitachi Deskstar SATA II
HDD: Gaming - 2 x 40gb Hitachi Deskstars in RAID 0
Power: Silverstone Zeus 560w
Case: Silverstone TJ09
Cooling: Scythe Infinity
Games Tested:
Call of Duty 2
F.E.A.R.
Oblivion
Quake 4
Command and Conquer 3: Tiberium Wars (Sapphire HD2900XT only)
I am using a manual run-through of the games listed, exactly the same as in previous reviews, but doing it slightly differently to get some consistant numbers. Notice that for all games I ran through with 4 x AA and 8 x AF @ 1920 x 1200. For all of the 3Dmark benchmarks I did two runs. One run was with the stock settings that the free version of the benchmark comes with and one was with 4 x AA @ 1920 x 1200.
For installation I have installed the card as usual with the normal PCI-E power dongle. Checked that the card is seated correctly and powered on.
I am using a clean install of Windows XP Professional SP2 with all the latest patches.
Both nVidia cards are using the nVidia 97.92 Forceware drivers.
The ATI card was installed with the Catalyst Control Center 7.6's.
Gainward Bliss 8800GTS "Golden Sample". Core Speed: 550. Memory Speeds: 1760 (880). Stream processor clock 1200MHz.
XFX 8800GTS "XXX" Edition. Core Speed 550. Memory speed 1800 (900). Stream processor clock 1500MHz.
Sapphire HD 2900 XT. Core Speed 740. memory speed 1650 (825). Stream processor clock 320MHz.
Most Recent Comments
As always, nice review Kempez!
wow, it kicked the 8800's ass at some things :O
in 3dmark it dominated
nice looking card
in 3dmark it dominated
nice looking card
Tis nice looking, not kind on the ears though
Quote:
|
Originally Posted by name='Kempez'
Tis nice looking, not kind on the ears though
|
Not a fan on the sandwich heatsink. The 8800 Blocks are expensive enough, the 2900s are going too need even more copper = higher price again. Same goes for after-market air coolers.
But
at the oblivion results
But
at the oblivion resultsLooks nice..
But a question, would this be a good upgrade from my x1900xt or should i go for a 2600xt? I am in two minds, as i dont really play a lot of games. But i do run two 1600x1200 monitors, and would like to be able to do dual screen with dx3d apps.
Or would a 8800GTS 640mb be better, given the price differences between them.
But a question, would this be a good upgrade from my x1900xt or should i go for a 2600xt? I am in two minds, as i dont really play a lot of games. But i do run two 1600x1200 monitors, and would like to be able to do dual screen with dx3d apps.
Or would a 8800GTS 640mb be better, given the price differences between them.
Tremendous review Kemp. I wonder why an 8 pin pci-e thingy isn`t in the bundle ? Seems strange to adopt a personalized connection type and not equip the average user with the means to use it.
That aside, bundle scores very nicely - tbh other manufacturers can learn a whole lot from this. Users love `free` stuff, and when it`s closely associated with the product too it`s great. Jeez a user would be happy with an XL "I score with Ati" shirt folded at the bottom of the box
Slightly off topic perhaps, but I see the Oblivion scores are silly-high in comparison with a 8800GTS... leads me to think if the game was heavily dev`d around an Ati engine. Would answer a few questions tbh. I can`t remember whether it boasts a card manuf when u run it.
Over all tho, I think it turned out as u may have previously thought, a good solid competitor for an 8800GTS. And it seems if u`r a really big Oblivion fan, u should seriously consider it.
That aside, bundle scores very nicely - tbh other manufacturers can learn a whole lot from this. Users love `free` stuff, and when it`s closely associated with the product too it`s great. Jeez a user would be happy with an XL "I score with Ati" shirt folded at the bottom of the box
Slightly off topic perhaps, but I see the Oblivion scores are silly-high in comparison with a 8800GTS... leads me to think if the game was heavily dev`d around an Ati engine. Would answer a few questions tbh. I can`t remember whether it boasts a card manuf when u run it.
Over all tho, I think it turned out as u may have previously thought, a good solid competitor for an 8800GTS. And it seems if u`r a really big Oblivion fan, u should seriously consider it.
Nice review but I think the comment about the "needing the extra 2 pins" was unnecessary. If you take a look @ the rear of your 8800GTX, you'll notice there is in fact a 6 pin and an 8 pin. NVIDIA chose to only solder on 6 of those 8 pins though. ATi could've done the same thing, you don't need 8 pins. ATi's Overdrive unlocks when you use an 8 pin and a 6 pin, but some people @ XS have discovered you can use two 6 pins and short the remaining 2 with a paperclip and Overdrive unlocks 
This I know, what I am questioning is why did ATI choose to include this? And why have they chosed not to unlock overdrive if you dont use the 6 PIN. Normal users won't be sticking paperclips into graphics cards and why the hell should they?
I am not just catering for the knowledgable enthusiasts in the reviews, it has to be everyone tbh
I am not just catering for the knowledgable enthusiasts in the reviews, it has to be everyone tbh
long review 
On the quake4 test. Did you do a timedemo
the results seem a bit 'wild'
All of the fps results are a bit wierd how do you get the avg fps etc?
edit: Once again I did three two minute runs on Quake 4 on each card and took the average of all my readings from these.
Sounds like a FRAPS type job?
The oblivion benchmark? is quite interesting.
On the hardware side, the cooler is quite nice
Shame about it being noisy
I thought the review was excellent. I really liked the beginning with the hardware images and little bits of info like the fan being noisy etc
I would be interested to see the HDMI capabilities of the card
On the quake4 test. Did you do a timedemo
the results seem a bit 'wild'All of the fps results are a bit wierd
how do you get the avg fps etc?edit: Once again I did three two minute runs on Quake 4 on each card and took the average of all my readings from these.
Sounds like a FRAPS type job?
The oblivion benchmark? is quite interesting.
On the hardware side, the cooler is quite nice
Shame about it being noisy
I thought the review was excellent. I really liked the beginning with the hardware images and little bits of info like the fan being noisy etc
I would be interested to see the HDMI capabilities of the card
Ye I use FRAPS to record FPS and play a manual run through of the game. Not sure why you say "wild", tis just min, max and average FPS. I know it works well cause the FPS stay the same every time and comparing two similar cards gets similar results. Some cards handle games well staying very steady, some really fluctuate ate certain points. That's where my impressions of the game come in so I can say if it was smooth or not (not just let the graphs do the talking)
Ta mate
I would have tested HDMI but I don't have an HDMI TV/Monitor (yet). Still, I know HDCP works well
Ta mate
I would have tested HDMI but I don't have an HDMI TV/Monitor (yet). Still, I know HDCP works well
There is a flaw in the review and nvidia cards were in disadvantage ..
"Both nVidia cards are using the nVidia !!! 97.92 !!! Forceware drivers.
The ATI card was installed with the Catalyst Control Center 7.6's."
The latest oficial is 158.22 and the latest beta is 163 ... Can the reviewer repeat the tests with latest official if not beta drivers...I hate nvidia very much but the reviews must be fair..
Latest drivers VS latest drivers
7.6 official VS 158.22 official
"Both nVidia cards are using the nVidia !!! 97.92 !!! Forceware drivers.
The ATI card was installed with the Catalyst Control Center 7.6's."
The latest oficial is 158.22 and the latest beta is 163 ... Can the reviewer repeat the tests with latest official if not beta drivers...I hate nvidia very much but the reviews must be fair..
Latest drivers VS latest drivers
7.6 official VS 158.22 official
The review clearly states that these results were taken from a previous review which is the reason for the older drivers.
Yes there would perhaps be a difference, but not to the extent it would skew the results.
I no longer have the 8800 GTS's as they are back with the manufacturers.
Yes there would perhaps be a difference, but not to the extent it would skew the results.
I no longer have the 8800 GTS's as they are back with the manufacturers.
Ya .. sorry..I forgot that the review states that thing.
Hello Kempez.After 7.8 will be released is any chance to update the review.The AA performance has been improved radically.
I no longer have a 2900XT, and unfortunately I can't redo each review after every driver release, I have enough trouble reviewing the new stuff that comes in! 
Quote:
|
Originally Posted by name='badsykes'
Hello Kempez.After 7.8 will be released is any chance to update the review.The AA performance has been improved radically.
|
Quote:
|
Originally Posted by name='Kempez'
I no longer have a 2900XT, and unfortunately I can't redo each review after every driver release, I have enough trouble reviewing the new stuff that comes in!
|
Trouble is this isn't when the graphics cards companies want them done, and we kind of have to live by their schedule I'm afraid
CSS Valid
Sorry about that
Here we go
Enjoy