Originally Posted by yahooadam the 7950 GX2 seems to give margianlly (very marginal) over the ATI cards, a x1900XT will set you back £352 (www.overclock.co.uk) whereas a 7950 will set you back £408
For ~£50 more you get like 5% performance gain (maybe) but its a better cooler, and it uses less power (i believe) and its not using the catalist software
But you have the hassle of SLI profiles and such
Generally speaking, in the more graphic-intense games, the performance of GeForce 7950 GX2 is in another league to Radeon X1900XTX. ;)
Originally Posted by zr_ox The price vs performance is the things that I'm really interested in because I'm pretty sure that X1900XTX in CF mode will own the 7950 GX2. The interesting thing is by how much and the relative £ per frame.
it cant be easy sourcing all of the hrdware and performing industry leading benchmarking and editorials. Out of interest, you dont have a little scoop on DirectX 10 hardware yet do you? There is an article over at http://www.extremetech.com/article2/0,1697,1982032,00.asp where they picked the brains of Microsofts Chris Donahue. He said that we should expect DirectX 10 hardware within two years :|. Thats all he could give because now its down to the hardware manufacturers. Since you guys have your finger on the pulse...have you heard anything...at all. The community needs to know!
Unfortunately, I am bound by Non-Disclosure Agreements at the moment - as soon as I am allowed to talk about next-generation hardware, I will. ;)
I am still looking to do something on the whole Vista/DirectX 10 subject, but I've been over-hauled by an incredible volume of hardware. Having lots of hardware is cool, but having to sit down and spend enough time with everything in that pile for long enough to be confident in my own opinions. Ultimately, that means a lot of man hours right now. ;)
Originally Posted by Tim S Unfortunately, I am bound by Non-Disclosure Agreements at the moment - as soon as I am allowed to talk about next-generation hardware, I will. ;)
I am still looking to do something on the whole Vista/DirectX 10 subject, but I've been over-hauled by an incredible volume of hardware. Having lots of hardware is cool, but having to sit down and spend enough time with everything in that pile for long enough to be confident in my own opinions. Ultimately, that means a lot of man hours right now. ;)
I understand.
But maybe I can leand a helping hand. My PC went up in smoke and I'm stuck with a 700 Mhz PIII laptop. I can help testing that mountain of hardware and help reduce your workload.
im never sure on things like fraps, i wonder which is more accurate, ingame counters and fraps dont allways give the same result ....
So which is more accurate, generally i like to stick to the ingame ones, but fraps does show on my G15 LCD .. which is really useful IMO (as you can for eg see in the graph how much your FPS dropped in a battle)
Originally Posted by yahooadam im never sure on things like fraps, i wonder which is more accurate, ingame counters and fraps dont allways give the same result ....
So which is more accurate, generally i like to stick to the ingame ones, but fraps does show on my G15 LCD .. which is really useful IMO (as you can for eg see in the graph how much your FPS dropped in a battle)
FRAPS is very good, based on my experiences with the tool dating back to 2004. The frame rate is generally a pretty good indication of the actual frame rate. :)
how do you check the high and low - just by eye or a plugin or somthing ?
do you use another screen so that fraps shows frame rate graph and stuff - just wondering how accurate the bit-tech benchies are :)
I sit and play games with both FRAPS and the in-game FPS counter (if available) running before starting to record frame rates. Once I am happy that the frame rates are 'where they should be', I then start investigating performance at various settings. Going through a checkbox list, to some extent.
i.e. Can I achieve smooth gameplay at 1920x1200 with 4xAA. If so, can I do it with transparency supersampling/adaptive AA enabled. If not, what about 2xAA with TSS/AAA.
Only when I am happy that the frame rates are consistently smooth over a good portion of the game being tested to I sit down and start recording manual FRAPS runs using the pre-determined 'benchmark hotspots' (typically between 3-5 minutes of intense gameplay). Providing the card behaves, three runs are recorded to remove the differences between runs (because they are run manually, there is an element of human error). The differences are typically less than 1fps, as each manual run through follows the same path in-game, at the same speed.
Most games take around 1-1 1/2 hours per card to benchmark per resolution. For this review, because there were two resolutions evaluated, you can double the time - it was taking over a day to benchmark the cards tested. Although the four different 'reference clocked' cards were essentially the same, I spent a day testing them all across a few games to ensure that they performed as well as one and other so that my reported benchmarks could be amalgamated into one.
To add further to this, the term 'smooth gameplay' is quite a loose one, and it ultimately depends on the title. These can only be decided after playing the games and getting a feel for how the games react to low frame rates. F.E.A.R. is one that doesn't suffer greatly from a low frame rate - that's because the frame rate mainly drops very low when SloMo is enabled. When that is turned on, the low frame rate doesn't impact gameplay, as it feels like the whole engine has slowed down.
hmmm - i find the problem is, as your going to be playing the game, they during the action bits (when FPS drops) you will be playing, and not recording the FPS - which is the exact points when you need to record them
And if you move out the way of the action to look at them, then you are going to get the wrong figure (as your not in the action) or you will die, and then that screws up the benchmark ...
Quote:
Originally Posted by customholle Do you get paid to do all this???
good question, they probably do, i would expect they get money from the supplying companies or something, or that adverts pay for it (altho adverts have to pay for the site as well) but they have to get money for their DS's and ipods from somewhere ey ;)
Originally Posted by customholle Do you get paid to do all this???
Yes, there are three of us working full time to bring you bit-tech content every day. :)
Without the advertising, the site would not be able to pay for us to spend all of our time doing this. We don't get money from people sending us samples - advertising and editorial is split completely separately. The sales guys handle sales and commercial relationships, while the editorial team is free to write what it wants. This means that we're bringing you guys independant and honest content.
Originally Posted by yahooadam hmmm - i find the problem is, as your going to be playing the game, they during the action bits (when FPS drops) you will be playing, and not recording the FPS - which is the exact points when you need to record them
And if you move out the way of the action to look at them, then you are going to get the wrong figure (as your not in the action) or you will die, and then that screws up the benchmark ...
We look at the frame rates in real time before we actually 'record'. We use the FRAPS 'record' frame rate function - we shoot guns, throw grenades and play the game like any normal gamer would. It's very accurate and many times more useful than a timedemo, since a timedemo doesn't actually represent what a gamer does in a game, unless you run a timedemo of course. ;)
Great roundup, Tim ! First "on the market" :) I've just registered after reading your article...
I have a few questions Tim, if you don't mind. Thanks a lot !
1. Is it true that *all* these cards follow 100% nVIDIA specs?
2. I've realized you didn't test the eVGA solution. I was about to switch to eVGA, but no... I am going to stay with Leadtek(like my last 3 cards). How is the packaging of Leadtek compared to the rest. I am going to send this card oversees in the 'cargo bay' :)
3. What is the Leadtek memory brand ? And what are the timings? (this might be related to Q1) Can you read from the BGA chips?
I am moving from GTX to the GX2 (Long story short..2 leadtek 7900gtx cards crapped out on me and i didnt want a 3rd :( so went for the gx2 ><)
Anyway my only thing that bugs me about the GX2 is the heat. It dosen't dump the heat via the back of the case but rather dump it back into the case..so i guess setting the Sidefan (or having the side case open) is probably the best thing to do here. I also plan to setup a intake in the front of the case...but i am wondering how the heat was in these cards.
My case is a Antec 1080AMG, using Asus A8N SLI Premium Mobo. I got my PCI-E slot pretty much in line with my HDD..so if the way the the heatsink works on this card..its gonna pretty much grab hot air in the case and cycle it through..which would be a pain considering these cards sit very very close to each other.
Originally Posted by nVIDIA Great roundup, Tim ! First "on the market" :) I've just registered after reading your article...
I have a few questions Tim, if you don't mind. Thanks a lot !
1. Is it true that *all* these cards follow 100% nVIDIA specs?
2. I've realized you didn't test the eVGA solution. I was about to switch to eVGA, but no... I am going to stay with Leadtek(like my last 3 cards). How is the packaging of Leadtek compared to the rest. I am going to send this card oversees in the 'cargo bay' :)
3. What is the Leadtek memory brand ? And what are the timings? (this might be related to Q1) Can you read from the BGA chips?
nVIDIA.
Hi there and welcome to the forums. ;)
I'll try to answer your questions as best I can:
1. All of the current GeForce 7950 GX2's, with the exception of two XFX models (one at 520/1300 and one at 570/1550), run at NVIDIA's reference 500/1200 clock speeds. The cards tested here ran at 500/1200, except XFX's XXX Edition, which runs at 570/1550.
2. The packing on Leadtek's card is 'as good' as any other Leadtek card. The card itself is packed in a padded anti-static bag in the centre of the box. I will try and grab a picture of the box layout (it's typical Leadtek, just a little bigger).
3. All of the cards are manufactured at the same place at the moment, so I would imagine that they're all using the same DRAMs - the DRAM is a Samsung BC14 chip - full code K4352324QC BC14. Hope this helps.
I am moving from GTX to the GX2 (Long story short..2 leadtek 7900gtx cards crapped out on me and i didnt want a 3rd :( so went for the gx2 ><)
Anyway my only thing that bugs me about the GX2 is the heat. It dosen't dump the heat via the back of the case but rather dump it back into the case..so i guess setting the Sidefan (or having the side case open) is probably the best thing to do here. I also plan to setup a intake in the front of the case...but i am wondering how the heat was in these cards.
My case is a Antec 1080AMG, using Asus A8N SLI Premium Mobo. I got my PCI-E slot pretty much in line with my HDD..so if the way the the heatsink works on this card..its gonna pretty much grab hot air in the case and cycle it through..which would be a pain considering these cards sit very very close to each other.
PS: Getting the Leadtek 7950 Card
Hi, welcome to the forums ;)
The air is blown out through the back of the case on the bottom card (hence the grille on the bracket). The cards get 'warm', but probably no warmer than your average Radeon X1900.
I don't think there will be much of an issue - the heatsink/fan solutions are quite efficient. :)
Originally Posted by Tim S Hi, welcome to the forums ;)
The air is blown out through the back of the case on the bottom card (hence the grille on the bracket). The cards get 'warm', but probably no warmer than your average Radeon X1900.
I don't think there will be much of an issue - the heatsink/fan solutions are quite efficient. :)
Wow i didnt know about the other card venting heat outside. Thats pretty cool actually.
I think i should also move my HDD more up if i can since hot air rises that way i will have less heat on the gpu sucking.
Cant wait to get the GX2 ><. Shame to see the Leadtek is a really ****** overclocker lol but who cares..they run so fast that a high overlock is so pointless..it might be more usefull to just overclock the CPU and get good RAM than trying to overclock the GX2..from the results it seems the overclock dosen't really help with the min frame rate..and i guess thats big important thing.
Cheers for the info ><.
PS: The Leadtek GTX card i got also came with only 1 DVI --> VGA Connector so the GX2 probably also is meant to have 1..which is very silly for such a expensive product to be missing something liek this :P
if your like me you probably have like 5 adaptors lying around lol, a lot more screens are starting to use DVI, and on such a high end card, i would expect a high end screen, and then you should definitely have DVI, in fact i don't understand why people are still using VGA (on the screens)
Hey that's a point, on the GX2, can you run 2 screens with both cards working, because SLI doesn't let you, and if you turn off SLI on those cards, it would be like cutting it down to a 7900GT or something because they are under clocked
Originally Posted by K007 Wow i didnt know about the other card venting heat outside. Thats pretty cool actually.
I think i should also move my HDD more up if i can since hot air rises that way i will have less heat on the gpu sucking.
Cant wait to get the GX2 ><. Shame to see the Leadtek is a really ****** overclocker lol but who cares..they run so fast that a high overlock is so pointless..it might be more usefull to just overclock the CPU and get good RAM than trying to overclock the GX2..from the results it seems the overclock dosen't really help with the min frame rate..and i guess thats big important thing.
Cheers for the info ><.
PS: The Leadtek GTX card i got also came with only 1 DVI --> VGA Connector so the GX2 probably also is meant to have 1..which is very silly for such a expensive product to be missing something liek this :P
Overclocking is a bit of a lottery - we may just have got a poor overclocker, you may get similar clocks to the other GX2s we tested and overclocked. :)
Comments 26 to 50 of 50
I am still looking to do something on the whole Vista/DirectX 10 subject, but I've been over-hauled by an incredible volume of hardware. Having lots of hardware is cool, but having to sit down and spend enough time with everything in that pile for long enough to be confident in my own opinions. Ultimately, that means a lot of man hours right now. ;)
I understand.
But maybe I can leand a helping hand. My PC went up in smoke and I'm stuck with a 700 Mhz PIII laptop. I can help testing that mountain of hardware and help reduce your workload.
You can get the free version of the frame rate recorder that we use from http://www.fraps.com
;)
So which is more accurate, generally i like to stick to the ingame ones, but fraps does show on my G15 LCD .. which is really useful IMO (as you can for eg see in the graph how much your FPS dropped in a battle)
do you use another screen so that fraps shows frame rate graph and stuff - just wondering how accurate the bit-tech benchies are :)
i.e. Can I achieve smooth gameplay at 1920x1200 with 4xAA. If so, can I do it with transparency supersampling/adaptive AA enabled. If not, what about 2xAA with TSS/AAA.
Only when I am happy that the frame rates are consistently smooth over a good portion of the game being tested to I sit down and start recording manual FRAPS runs using the pre-determined 'benchmark hotspots' (typically between 3-5 minutes of intense gameplay). Providing the card behaves, three runs are recorded to remove the differences between runs (because they are run manually, there is an element of human error). The differences are typically less than 1fps, as each manual run through follows the same path in-game, at the same speed.
Most games take around 1-1 1/2 hours per card to benchmark per resolution. For this review, because there were two resolutions evaluated, you can double the time - it was taking over a day to benchmark the cards tested. Although the four different 'reference clocked' cards were essentially the same, I spent a day testing them all across a few games to ensure that they performed as well as one and other so that my reported benchmarks could be amalgamated into one.
To add further to this, the term 'smooth gameplay' is quite a loose one, and it ultimately depends on the title. These can only be decided after playing the games and getting a feel for how the games react to low frame rates. F.E.A.R. is one that doesn't suffer greatly from a low frame rate - that's because the frame rate mainly drops very low when SloMo is enabled. When that is turned on, the low frame rate doesn't impact gameplay, as it feels like the whole engine has slowed down.
And if you move out the way of the action to look at them, then you are going to get the wrong figure (as your not in the action) or you will die, and then that screws up the benchmark ...
Without the advertising, the site would not be able to pay for us to spend all of our time doing this. We don't get money from people sending us samples - advertising and editorial is split completely separately. The sales guys handle sales and commercial relationships, while the editorial team is free to write what it wants. This means that we're bringing you guys independant and honest content.
There's more here: http://www.bit-tech.net/news/2005/06/03/bit_tech_giant_leap/
Anyway, you're straying wayyy off topic now. ;)
EDIT: And thanks for the link
I have a few questions Tim, if you don't mind. Thanks a lot !
1. Is it true that *all* these cards follow 100% nVIDIA specs?
2. I've realized you didn't test the eVGA solution. I was about to switch to eVGA, but no... I am going to stay with Leadtek(like my last 3 cards). How is the packaging of Leadtek compared to the rest. I am going to send this card oversees in the 'cargo bay' :)
3. What is the Leadtek memory brand ? And what are the timings? (this might be related to Q1) Can you read from the BGA chips?
nVIDIA.
I am moving from GTX to the GX2 (Long story short..2 leadtek 7900gtx cards crapped out on me and i didnt want a 3rd :( so went for the gx2 ><)
Anyway my only thing that bugs me about the GX2 is the heat. It dosen't dump the heat via the back of the case but rather dump it back into the case..so i guess setting the Sidefan (or having the side case open) is probably the best thing to do here. I also plan to setup a intake in the front of the case...but i am wondering how the heat was in these cards.
My case is a Antec 1080AMG, using Asus A8N SLI Premium Mobo. I got my PCI-E slot pretty much in line with my HDD..so if the way the the heatsink works on this card..its gonna pretty much grab hot air in the case and cycle it through..which would be a pain considering these cards sit very very close to each other.
PS: Getting the Leadtek 7950 Card
I'll try to answer your questions as best I can:
1. All of the current GeForce 7950 GX2's, with the exception of two XFX models (one at 520/1300 and one at 570/1550), run at NVIDIA's reference 500/1200 clock speeds. The cards tested here ran at 500/1200, except XFX's XXX Edition, which runs at 570/1550.
2. The packing on Leadtek's card is 'as good' as any other Leadtek card. The card itself is packed in a padded anti-static bag in the centre of the box. I will try and grab a picture of the box layout (it's typical Leadtek, just a little bigger).
3. All of the cards are manufactured at the same place at the moment, so I would imagine that they're all using the same DRAMs - the DRAM is a Samsung BC14 chip - full code K4352324QC BC14. Hope this helps.
The air is blown out through the back of the case on the bottom card (hence the grille on the bracket). The cards get 'warm', but probably no warmer than your average Radeon X1900.
I don't think there will be much of an issue - the heatsink/fan solutions are quite efficient. :)
Wow i didnt know about the other card venting heat outside. Thats pretty cool actually.
I think i should also move my HDD more up if i can since hot air rises that way i will have less heat on the gpu sucking.
Cant wait to get the GX2 ><. Shame to see the Leadtek is a really ****** overclocker lol but who cares..they run so fast that a high overlock is so pointless..it might be more usefull to just overclock the CPU and get good RAM than trying to overclock the GX2..from the results it seems the overclock dosen't really help with the min frame rate..and i guess thats big important thing.
Cheers for the info ><.
PS: The Leadtek GTX card i got also came with only 1 DVI --> VGA Connector so the GX2 probably also is meant to have 1..which is very silly for such a expensive product to be missing something liek this :P
Hey that's a point, on the GX2, can you run 2 screens with both cards working, because SLI doesn't let you, and if you turn off SLI on those cards, it would be like cutting it down to a 7900GT or something because they are under clocked