- Joined
- Mar 24, 2004
PRODUCT ARRIVED. ORANGE SCORES ARE MY x1950XTX AND GREEN SCORES ARE 2900XT!!!!!!
PICTURES: http://public.fotki.com/vixro/computer-related/sapphire-hd-2900xt/
*click on a thumbnail and then click on "get original uploaded photo" for the hi-res version*
Test System:
I left programs open to give a better idea of how your performance is in a real world setting. You don't turn down all your graphics and close every program you have before you play your games (at least, I don't).
Programs open:
NOD32
AIM
MSN Live
Quickcam Pro
Gmail Notifier
X-Fi Control Panel
CCC Traybar
Vista Sidebar with a couple gadgets.
This is done to show performance on a normal system that has programs running in the background instead of a completely clean test system that gives unfair results.
Intel Core 2 Duo e6600 @ 3.24ghz
G.Skill BHZ PC2-6400 2x1gb @ 540mhz(1080) 5-5-5-12-2T
Asus Commando v1.00g with 1001 bios.
X-Fi XtremeMusic
Catalyst AI on normal for all tests.
Unless otherwise noted, all tests run with maximum quality graphics 1680x1050res, AAA disabled.
connect3D x1950XTX with Cat. 7.4
Sapphire HD 2900 XT with (8.37)
Windows Vista 32bit
3dMark06
1280x1024
No AAA.
Optimal filtering.
High performance mipmap.
Catalyst AI advanced.
High-quality AF disabled.
Benchmark settings (default 3dmark06 settings) run.
SM 2.0 - 2608 / 4638 (77.8% increase)
SM 3.0 - 2935 // 4989 (69.9% increase)
CPU - 2764 / 2794
Score - 6925 / 10857 (56.7% increase)
Advanced scores:
Fill Rate Single-Texturing - 6132.795 MTexels/s / 8057.992 MTexels/s (31.4% increase)
Fill Rate Multi-Texturing - 10308.639 MTexels/s / 11786.722 MTexels/s (14.3% increase)
Pixel Shader - 293.541fps / 314.037fps (.07% increase)
Vertex Shader Simple - 187.335 MVertices/s / 242.458 MVertices/s (29.4% increase)
Vertex Shader Complex - 62.026 MVertices/s / 191.640 MVertices/s (208.9% increase)
Shader Particles (SM 3.0) - N/A / 129.735fps
Perlin Noise (SM 3.0) - 94.840fps / 172.011fps (81.3% increase)
3dMark06 - High Quality
1680x1050
4x AA
16x AF
Very high mipmap.
Catalyst AI normal.
High-quality AF enabled.
SM 2.0 - 1716 / 2000 (16.5% increase)
SM 3.0 - 1393 // 2439 (75.1% increase)
CPU - 2768 / 2801
Score - 4159 / 5727 (37.7% increase)
3dMark05
1024x768
No AAA.
Optimal filtering.
High performance mipmap.
Catalyst AI advanced.
High-quality AF disabled.
Benchmark settings (default 3dmark05 settings) run.
Score - 13801 / 19376 (40.4% increase)
Advanced scores:
Fill Rate Single-Texturing - 6277.0 MTexels/s / 8063.5 MTexels/s (28.4% increase)
Fill Rate Multi-Texturing - 10159.9 MTexels/s / 11739.2 MTexels/s (15.5% increase)
Pixel Shader - 510.5fps / 546.5fps (.07% increase)
Vertex Shader Simple - 149.4 MVertices/s / 228.4 MVertices/s (52.8% increase)
Vertex Shader Complex - 62.1 MVertices/s / 192.9 MVertices/s (210.6% increase)
3dMark05 - High Quality
1680x1050
4x AA
16x AF
Very high mipmap.
Catalyst AI normal.
High-quality AF enabled.
Scores - 8069 / 10894 (35% increase)
Guild Wars Nightfall*:
1680x1050
4xAA
16xAF
1680x1050
Everything maxed possible.
-- 67-80fps ~170,000-180,000 Triangles / 65-110fps ~170,000-190,000 Triangles
-Score determined by triangles. Higher triangles means more processing. The low end frames match the high end triangle. e.g. 67fps @ 180,000 triangles for x1950xtx.
Counter-Strike Source:
1680x1050
4xAA
16xAF
HDR Full
Relfect All
High Details
-- 176.94fps / 154.97fps (12.4% decrease)
Half Life 2: Lost Coast:
1680x1050
4xAA
16xAF
HDR Full
Relfect All
High Details
-- 82.41fps / 94.86fps (15.1% increase)
Half Life 2: Episode 1 (HOC benchmark 1.1.0.0):
Resolution: 1680 × 1050
Demo: PC_Space demo
Detail: Maximum quality
High Dynamic Range: Full (if available)
Heapsize: 1024 MB
Antialising mode: 4×
Filtering mode: Anisotropic 16×
Fast Z: Disabled
Color Correction: Enabled
-- 68fps / 84fps (23.5% increase)
Quake 4 1.4.1 (HOC Benchmark 1.5):
Resolution: 1680 × 1050
Demo: HOCdemo.demo
Quality: Ultra
Aspect Ratio: [16:9]
Antialiasing: 4×
Anisotropic filtering: 16×
Symmetric MultiProcessing (SMP): enabled
Force Ambient Light: disabled
-- 33fps / 48fps (45.4% increase)
* - Game does not support benchmark. Test is hard to replicate.
Conclusions: The card is what I expected out of the box. I haven't tested the overclocking capabilities and I think I will probably wait until the drivers are a bit better before attempting anything fierce. I can feel the heat coming out the back of my computer and it's a bit hotter than my x1950xtx. The fan runs at a higher idle speed than my old card, probably because of the extra heat it is producing. I can't give an exact number, but I am pretty sure that it runs a bit hotter. Once I find a better GPU temperature monitoring device, I can post exactly how much heat that is. The card is "noisy" at load, but not any more noisy than an x1950xtx. It's louder than the nvidia cards I tested, but in the ATI line it runs about average. Same as an x1950xtx, perhaps slightly louder or less loud depending on the situation and much quieter than an x1900.
Temperatures, equal to that of my x1950xtx. I was concerned that it was running hotter, but according to AMD GPU Clock Tool I average 50-55c at idle, and 70-85c at load depending how long it's been on. The fan kicks on higher speed at 70c, and I do not know the next level of fan increase because I didn't reach it in my short testing period (1 hour).
As for speeds, I was impressed, then less and less impressed. The speeds of newer games and benchmarks is fantastic. I got what I expected out of the card and even more in Quake and in 3dmark06. My Source based games (HL2, lost coast, ep1) weren't as good as I expected. I noticed a DECREASE in performance on CS:S my favorite First Person Shooter. I reran the test, changed options, and did everything I could think of to increase the performance without modifying the test too much. I found similar speeds and less speeds than my x1950xtx but never more speeds. Either valve needs to make an update to the Source engine to accomidate these new cards, or AMD needs to release some better drivers for these type of games. The image quality was similar to that of my x1900 in these directx 9 based games, but that is definitely not a fair comparison since this card is DX10 based. I am hopeful that newer drivers will increase the speeds of these DX9 based games without a problem and know that I won't be let down in this field. Other than 3dmark06 and 05, which I had to set to not detect hardware, I didn't really notice any problems within games. It seems 6xAA is not supported in games, only 2x 4x and 8x. I didn't test any of the special AA modes, as I hadn't found the need or the time to go that far. Perhaps I will eventually delve further into those options once newer games are released.
For price, I am happy. It is not much more than a x1950xtx and if you search in the right places you may even get it for the same price. If you're in the market for a new AMD card, I would say go for the 2900XT because it's great for the money compared to other AMD based cards. I have NOT tested any of the 8xxx series cards, so I cannot say if those are better or worse than this new card. My comparisons are based only on the latest AMD based cards and my usage of a 7900GT from last summer. What I state are mostly opinions with a few facts from my own experience, so take it with a grain of salt as a tester that kept all options real and gave his honest opinion. If you are currently on an x1900xt, x1900xtx, 1950---, I would not recommend upgrading unless you want a bit of extra power or you play the games with the larger increase, the hardware you have now is quite fine until driver updates are released and the card reaches it's true potential. If you have an x1800 or lower and you're ready to upgrade to top of the line now. I can see no better choice than the x1950 or the 2900XT on the AMD side. Once again, I have NOT tested the 8xxx series from Nvidia, they may be better or worse, my opinion solely lies on AMD based cards.
I am going to further tweak my RAM and my CPU to make sure I have the best performance I can possibly get. I know I could have had higher points or higher frames with programs closed, a better overclock of my RAM/cpu, and other factors but I wanted to give true life examples of how the card may run for the average user that plays games and may only have an average overclock and nothing extreme.
edit: Original Post----- 5-12-07
http://www.zipzoomfly.com/jsp/ProductDetail.jsp?ProductCode=10005202
I just ordered a Sapphire HD X2900XT 512MB from ZZF for 399.99 + 14 for overnight shipping. (3-5day free, 2day 2.99)
If you use this link, you can find that more and more retailers are starting to equip these cards. I couldn't find any at newegg so I just went with ZZF (where I purchased my x1950xtx when they first came out). I am pretty excited and am going against what I said about being a Guinea pig, but I couldn't resist when I saw a notable etailer with it for a decent price. I wouldn't be surprised if newegg completely sells out of them and then loads the price up (remember the 8800GTX 512MB prices at newegg?).
Anyway, this thread is a placeholder for the results which I will start posting on Tuesday when I install this card into my rig. I will be selling my x1950xtx in the classifieds as soon as I get this card up and running. I can't wait!
Were you waiting, sitting on the fence? I will be here to post the results and take tons of high res pictures for all those interested. If anyone else decides to hop on one, you can share all your results in this thread too. Some gaming results, synthetic benchmarks, and overall performance.
PICTURES: http://public.fotki.com/vixro/computer-related/sapphire-hd-2900xt/
*click on a thumbnail and then click on "get original uploaded photo" for the hi-res version*
Test System:
I left programs open to give a better idea of how your performance is in a real world setting. You don't turn down all your graphics and close every program you have before you play your games (at least, I don't).
Programs open:
NOD32
AIM
MSN Live
Quickcam Pro
Gmail Notifier
X-Fi Control Panel
CCC Traybar
Vista Sidebar with a couple gadgets.
This is done to show performance on a normal system that has programs running in the background instead of a completely clean test system that gives unfair results.
Intel Core 2 Duo e6600 @ 3.24ghz
G.Skill BHZ PC2-6400 2x1gb @ 540mhz(1080) 5-5-5-12-2T
Asus Commando v1.00g with 1001 bios.
X-Fi XtremeMusic
Catalyst AI on normal for all tests.
Unless otherwise noted, all tests run with maximum quality graphics 1680x1050res, AAA disabled.
connect3D x1950XTX with Cat. 7.4
Sapphire HD 2900 XT with (8.37)
Windows Vista 32bit
3dMark06
1280x1024
No AAA.
Optimal filtering.
High performance mipmap.
Catalyst AI advanced.
High-quality AF disabled.
Benchmark settings (default 3dmark06 settings) run.
SM 2.0 - 2608 / 4638 (77.8% increase)
SM 3.0 - 2935 // 4989 (69.9% increase)
CPU - 2764 / 2794
Score - 6925 / 10857 (56.7% increase)
Advanced scores:
Fill Rate Single-Texturing - 6132.795 MTexels/s / 8057.992 MTexels/s (31.4% increase)
Fill Rate Multi-Texturing - 10308.639 MTexels/s / 11786.722 MTexels/s (14.3% increase)
Pixel Shader - 293.541fps / 314.037fps (.07% increase)
Vertex Shader Simple - 187.335 MVertices/s / 242.458 MVertices/s (29.4% increase)
Vertex Shader Complex - 62.026 MVertices/s / 191.640 MVertices/s (208.9% increase)
Shader Particles (SM 3.0) - N/A / 129.735fps
Perlin Noise (SM 3.0) - 94.840fps / 172.011fps (81.3% increase)
3dMark06 - High Quality
1680x1050
4x AA
16x AF
Very high mipmap.
Catalyst AI normal.
High-quality AF enabled.
SM 2.0 - 1716 / 2000 (16.5% increase)
SM 3.0 - 1393 // 2439 (75.1% increase)
CPU - 2768 / 2801
Score - 4159 / 5727 (37.7% increase)
3dMark05
1024x768
No AAA.
Optimal filtering.
High performance mipmap.
Catalyst AI advanced.
High-quality AF disabled.
Benchmark settings (default 3dmark05 settings) run.
Score - 13801 / 19376 (40.4% increase)
Advanced scores:
Fill Rate Single-Texturing - 6277.0 MTexels/s / 8063.5 MTexels/s (28.4% increase)
Fill Rate Multi-Texturing - 10159.9 MTexels/s / 11739.2 MTexels/s (15.5% increase)
Pixel Shader - 510.5fps / 546.5fps (.07% increase)
Vertex Shader Simple - 149.4 MVertices/s / 228.4 MVertices/s (52.8% increase)
Vertex Shader Complex - 62.1 MVertices/s / 192.9 MVertices/s (210.6% increase)
3dMark05 - High Quality
1680x1050
4x AA
16x AF
Very high mipmap.
Catalyst AI normal.
High-quality AF enabled.
Scores - 8069 / 10894 (35% increase)
Guild Wars Nightfall*:
1680x1050
4xAA
16xAF
1680x1050
Everything maxed possible.
-- 67-80fps ~170,000-180,000 Triangles / 65-110fps ~170,000-190,000 Triangles
-Score determined by triangles. Higher triangles means more processing. The low end frames match the high end triangle. e.g. 67fps @ 180,000 triangles for x1950xtx.
Counter-Strike Source:
1680x1050
4xAA
16xAF
HDR Full
Relfect All
High Details
-- 176.94fps / 154.97fps (12.4% decrease)
Half Life 2: Lost Coast:
1680x1050
4xAA
16xAF
HDR Full
Relfect All
High Details
-- 82.41fps / 94.86fps (15.1% increase)
Half Life 2: Episode 1 (HOC benchmark 1.1.0.0):
Resolution: 1680 × 1050
Demo: PC_Space demo
Detail: Maximum quality
High Dynamic Range: Full (if available)
Heapsize: 1024 MB
Antialising mode: 4×
Filtering mode: Anisotropic 16×
Fast Z: Disabled
Color Correction: Enabled
-- 68fps / 84fps (23.5% increase)
Quake 4 1.4.1 (HOC Benchmark 1.5):
Resolution: 1680 × 1050
Demo: HOCdemo.demo
Quality: Ultra
Aspect Ratio: [16:9]
Antialiasing: 4×
Anisotropic filtering: 16×
Symmetric MultiProcessing (SMP): enabled
Force Ambient Light: disabled
-- 33fps / 48fps (45.4% increase)
* - Game does not support benchmark. Test is hard to replicate.
Code:
Formula used to calculate % increase or decrease:
( (Larger number) - (smaller number) ) / (1950xtx result) = (X)*100 = Percentage.
x1950xtx = 33fps
2900xt = 48fps
48-33 = 15
15/33 = .45
.4545*100 = 45.4Conclusions: The card is what I expected out of the box. I haven't tested the overclocking capabilities and I think I will probably wait until the drivers are a bit better before attempting anything fierce. I can feel the heat coming out the back of my computer and it's a bit hotter than my x1950xtx. The fan runs at a higher idle speed than my old card, probably because of the extra heat it is producing. I can't give an exact number, but I am pretty sure that it runs a bit hotter. Once I find a better GPU temperature monitoring device, I can post exactly how much heat that is. The card is "noisy" at load, but not any more noisy than an x1950xtx. It's louder than the nvidia cards I tested, but in the ATI line it runs about average. Same as an x1950xtx, perhaps slightly louder or less loud depending on the situation and much quieter than an x1900.
Temperatures, equal to that of my x1950xtx. I was concerned that it was running hotter, but according to AMD GPU Clock Tool I average 50-55c at idle, and 70-85c at load depending how long it's been on. The fan kicks on higher speed at 70c, and I do not know the next level of fan increase because I didn't reach it in my short testing period (1 hour).
As for speeds, I was impressed, then less and less impressed. The speeds of newer games and benchmarks is fantastic. I got what I expected out of the card and even more in Quake and in 3dmark06. My Source based games (HL2, lost coast, ep1) weren't as good as I expected. I noticed a DECREASE in performance on CS:S my favorite First Person Shooter. I reran the test, changed options, and did everything I could think of to increase the performance without modifying the test too much. I found similar speeds and less speeds than my x1950xtx but never more speeds. Either valve needs to make an update to the Source engine to accomidate these new cards, or AMD needs to release some better drivers for these type of games. The image quality was similar to that of my x1900 in these directx 9 based games, but that is definitely not a fair comparison since this card is DX10 based. I am hopeful that newer drivers will increase the speeds of these DX9 based games without a problem and know that I won't be let down in this field. Other than 3dmark06 and 05, which I had to set to not detect hardware, I didn't really notice any problems within games. It seems 6xAA is not supported in games, only 2x 4x and 8x. I didn't test any of the special AA modes, as I hadn't found the need or the time to go that far. Perhaps I will eventually delve further into those options once newer games are released.
For price, I am happy. It is not much more than a x1950xtx and if you search in the right places you may even get it for the same price. If you're in the market for a new AMD card, I would say go for the 2900XT because it's great for the money compared to other AMD based cards. I have NOT tested any of the 8xxx series cards, so I cannot say if those are better or worse than this new card. My comparisons are based only on the latest AMD based cards and my usage of a 7900GT from last summer. What I state are mostly opinions with a few facts from my own experience, so take it with a grain of salt as a tester that kept all options real and gave his honest opinion. If you are currently on an x1900xt, x1900xtx, 1950---, I would not recommend upgrading unless you want a bit of extra power or you play the games with the larger increase, the hardware you have now is quite fine until driver updates are released and the card reaches it's true potential. If you have an x1800 or lower and you're ready to upgrade to top of the line now. I can see no better choice than the x1950 or the 2900XT on the AMD side. Once again, I have NOT tested the 8xxx series from Nvidia, they may be better or worse, my opinion solely lies on AMD based cards.
I am going to further tweak my RAM and my CPU to make sure I have the best performance I can possibly get. I know I could have had higher points or higher frames with programs closed, a better overclock of my RAM/cpu, and other factors but I wanted to give true life examples of how the card may run for the average user that plays games and may only have an average overclock and nothing extreme.
edit: Original Post----- 5-12-07
http://www.zipzoomfly.com/jsp/ProductDetail.jsp?ProductCode=10005202
I just ordered a Sapphire HD X2900XT 512MB from ZZF for 399.99 + 14 for overnight shipping. (3-5day free, 2day 2.99)
If you use this link, you can find that more and more retailers are starting to equip these cards. I couldn't find any at newegg so I just went with ZZF (where I purchased my x1950xtx when they first came out). I am pretty excited and am going against what I said about being a Guinea pig, but I couldn't resist when I saw a notable etailer with it for a decent price. I wouldn't be surprised if newegg completely sells out of them and then loads the price up (remember the 8800GTX 512MB prices at newegg?).
Anyway, this thread is a placeholder for the results which I will start posting on Tuesday when I install this card into my rig. I will be selling my x1950xtx in the classifieds as soon as I get this card up and running. I can't wait!
Were you waiting, sitting on the fence? I will be here to post the results and take tons of high res pictures for all those interested. If anyone else decides to hop on one, you can share all your results in this thread too. Some gaming results, synthetic benchmarks, and overall performance.
Last edited:
![]-[itman](/forums/data/avatars/s/9/9309.jpg?1650325391){kind=avatar}