I think you have misunderstood me, I am aware of the lack of AA + HDR on nvidia cards and I agree with the "highest playable settings" testing model, but as you say with the SC:CT point, all the reviews I have read regarding that game highlight the fact that you could only use SM2 on ATI (patched) rather than SM3 (pre X1000 series) and that it would affect the results of the tests, you didn't highlight the drivers' difference in supporting dual cores, also I said FX-57 as that is what you said was the other choice in the poll, I wasn't referring to user base, it is merely dual core versus single core that I was trying to get across. In my opinion the difference in highest playable settings between all the cards tested may have differed on a single core setup. Oh and you can't deny that there are more single core gamers than dual core, there may be more or less than I think, I never stated a number. :D
I've found very little difference between FX-57 and X2 4800+ in all honesty.
I also upgraded the memory this time around too. That made some games react totally differently (like BF2 for example) because of the extra system memory, but I've not found much of a difference between X2 4800+ and FX-57 when I was playing around with 1GB of memory on an X2. All games that are currently available are all single threaded anyway, so I don't believe that the difference is as big as you're making it out to be.
I'm not talking about the difference between the FX-57 and the X2, I'm talking about the fact that the nvidia drivers are using the second core for some vertex processing IIRC, whereas the Ati drivers are using only one core for the game. I know that games are single threaded now, but to highlight the performance difference, here is a post at nvnews.
The Ati card you tested is basically running like the nvidia card in those results with the /ONECPU switch. I would say that those are reasonably significant performance gains.
Sorry I couldn't find more I'm at work right now, but I have read that the performance gains can be even larger than that.
I was just wondering if there was any word on the overclockability of the x1800xl? but if u r having to use beta drivers i am guessing that atitool haven't got anything that will do it yet? would be nice to know what sort of headroom the r520 has in it.
Originally Posted by skank I'm not talking about the difference between the FX-57 and the X2, I'm talking about the fact that the nvidia drivers are using the second core for some vertex processing IIRC, whereas the Ati drivers are using only one core for the game. I know that games are single threaded now, but to highlight the performance difference, here is a post at nvnews.
The Ati card you tested is basically running like the nvidia card in those results with the /ONECPU switch. I would say that those are reasonably significant performance gains.
Sorry I couldn't find more I'm at work right now, but I have read that the performance gains can be even larger than that.
Notice that they're running games that I've not used, and they're also running time demos which do not reflect real world game performance. I'm talking about resolution and detail scaling here, not frame rate scaling. A game is either playable or it isn't and the difference between FX-57 and X2 4800+ does nothing to playability from my personal experiences.
We're using drivers that are publicly available for NVIDIA hardware from an official source and we're using a CPU that was chosen by the majority of bit-tech readers who chose to vote in a recent poll in the hardware forum. You can't actually get the ATI drivers unless you actually buy an X1800XL. That's the only part of the whole review that is not publicly available and I challenge anyone to prove that the settings that we've reported do not fit inside the boundaries that we set out.
In fact, the only one that did was on the Sapphire Radeon X1800XL in F.E.A.R. where the average frame rate was 49.5 fps but the minimum frame rate was 18 fps and the game was a lot smoother than on the XFX 7800 GT Extreme scoring 60.4 fps average and 18 fps minimum. As I said in the review, I'd take the Sapphire's smoothness over the higher average (but noticeably choppier) frame rate that was returned by the XFX card any day.
/ponders Gosh, I must be NVIDIA's PR bitch because I used a dual core CPU.
The clocks are impressive, but the XL and XT have different core revisions AFAIK. I don't have the Sapphire card anymore and I didn't have an overclocking tool, so I can't tell you how well a retail XL clocks at the moment...
I don't know why you keep bringing up the comparison with the X2 and the FX, I don't care about that comparison (merely said FX in my earlier post as it is single core), I'm just saying that you should inform your readers that there is a likely performance gain for the nvidia card on the X2 from the fact that you are using drivers that support dual core from nvidia compared to drivers that don't from Ati, just like readers are normally informed that Ati is using SM1.1 or 2.0 or 2.0+ or whatever which can affect performance and playable settings. Perhaps you could do your own testing to see if there are any real world benefits from the dual core aware drivers? ie whether when you disable one of the cores on the X2 you have to drop some detail? At the end of the day, timedemo or not, higher fps normally indicates that for the same resolution you can enable higher details, maybe go up a level in AA, maintaining the same playability level. I'm not saying you are in nvidia's PR team or anything (really though how much are they paying you?) :)
I just like to see impartiality, and ommitting the fact that nvidia is using both cores in it's drivers (which I think is fantastic by the way) is not painting a fair picture of what is really happening.
Originally Posted by bigz The clocks are impressive, but the XL and XT have different core revisions AFAIK. I don't have the Sapphire card anymore and I didn't have an overclocking tool, so I can't tell you how well a retail XL clocks at the moment...
Originally Posted by skank I just like to see impartiality, and ommitting the fact that nvidia is using both cores in it's drivers (which I think is fantastic by the way) is not painting a fair picture of what is really happening.
As I've said I discussed the CPU and left it open to the readers to choose from two choices - I went with what the majority of them said they'd buy if they were given the choice.
Now you're telling me that I need to be impartial and not use the latest publicly available drivers on one IHV's products in order to disable an advantage that they have because the other team haven't adopted dual core support in to their drivers just yet. When ATI get around to adding dual core support in to their drivers, we'll use them without a doubt. Until then, we're in the "7800 GTX vs no product" situation again.
What if it takes ATI a year to get dual core support in to their drivers? Long after single core CPUs are (likely to be) history. Anyone buying a system today would be unwise buying a single core CPU unless they aren't going to use it for anything other than gaming. Most people who would drop $1000/£650 on a CPU would use it in their main computer, which also doubles as a gaming machine too.
Again, there's no doubts that we'll look at ATI's new products in the future when we get samples from other board partners. If and when they release new drivers that add dual core support, we'll revisit ATI's newer dual-core ready drivers with another product review.
We can only tell it like it is now, and this is how it is.
I'm not telling you to use any other drivers, I just think it would have been prudent of you to mention the dual core support in the drivers and the possible impact of that, just as I stated before with regards to the ATi cards using SM1.1, 2.0 etc. Uninformed readers may take the performance advantage to be down to the cards alone, without regard for the fact that you are testing on a dual core platform. All the reviews from reputable sites (I can't remember if I read a review from bit-tech but I do count you as reputable) stated the SM situation and explained the impact to gamers. Also I would be dead interested in seeing the performance gains tested as I said, by disabling a core, I'm sure others would be too. :)
when I get the time, I'll do something along those lines. I've got a general article planned in a couple of weeks - I'll look and see if I can include something in there at that time.
Thats not what the review was about, the review serves to compare that sample of ATI's best offering to Nvidia's best offering, if one didn't support DirectX 9, would it be fair to run only DirectX 8 tests? "No", I hear you say.
This is the top end, all a reviewer can do is let them off the leash and see what the best they can acheive is. A second core does very little for a single threaded game anyway (99.99% are), all the second core does is take off the background load that system services are using.
Originally Posted by Xen0phobiak Thats not what the review was about, the review serves to compare that sample of ATI's best offering to Nvidia's best offering, if one didn't support DirectX 9, would it be fair to run only DirectX 8 tests? "No", I hear you say.
This is the top end, all a reviewer can do is let them off the leash and see what the best they can acheive is. A second core does very little for a single threaded game anyway (99.99% are), all the second core does is take off the background load that system services are using.
edit: oops I'm a bit late.
You haven't read this thread have you? If you had you would see that I'm not talking about the game being multithreaded, it is the DRIVERS that are multithreaded. Also even though the DX8.0 vs DX9.0 example is rediculous, if you did do that you would still inform your readers that you were doing it and why, I just didn't like the complete lack of mentioning the fact that the nvidia drivers are multithreaded. http://techreport.com/onearticle.x/8459
I think I agree with bigz here, we are trying to reflect the game on best playable settings for a gaming rig, which everyone in this forum seem to have agreed that the X2 is the way to go for gaming now.
I also think that the fact that ATI hasn't taken in the use of dual core processors means the 7800GT is more futureproof and, overall, better.
I don't believe that either company are 'better' than the other - they both have their plus points. It will be interesting to see the way that games go over the next 12 months or so. If the Sapphire X1800XL came down in price, it'd be as good a purchase as the XFX or BFGTech cards, but we can only write a review based on its price at the time of publishing.
Originally Posted by bigz I don't believe that either company are 'better' than the other - they both have their plus points. It will be interesting to see the way that games go over the next 12 months or so. If the Sapphire X1800XL came down in price, it'd be as good a purchase as the XFX or BFGTech cards, but we can only write a review based on its price at the time of publishing.
I actually meant the GT being better (imo) than the x1800 - edited!
That will all depend on the way that games go. I've said this to one of the graphics firms in an email exchange the other week - I don't particularly think that either solution is the 'Radeon 9700' of this generation. I don't think there is a Radeon 9700 this time around. The reason I mention Radeon 9700 is that when it came out (and until the launch of GeForce 6) there was no competition for the architecture that it was based around. GeForceFX was a poor architecture in comparison to the Radeon 9700 and it got spanked in just about every department, and rightly so.
There is no doubt that the Radeon X1800XL has some nice features, including superior dynamic branching performance, better filtering quality, and the option for HDR+AA in all games that support HDR (and are patched to use hardware-based AA at the same time). The GeForce 7800 GT has a feature set that isn't too far off it and I'd say that NVIDIA have got the stronger multi-GPU platform at the moment by a long shot, but we'll have to wait and see how good CrossFire on X1800 series is when CrossFire cards are available.
The reason that the GeForce 7800 GT (as a single card solution) is a better buy at the moment is because the price differential is huge (~25%) and you can buy a pre-overclocked GT that outperforms or matches the X1800XL in just about every situation for a lot lot less. NVIDIA have a well-refined architecture behind GeForce 7, but there are some things missing that only become apparent when you look at ATI's architecture.
Admittedly, High Quality AF is not all that useable in the most demanding games out at the moment, and Adaptive AA isn't quite as efficient as Transparency SuperSampling from a real world gaming perspective. The time demos seem to show it up as quite useable though. Having said that, like HDR was on GeForce 6 hardware - it's good to have those features if the game is capable of using them without too much of a performance hit or, of course, you don't mind dropping the res/details to turn them on.
It's back and forth all the way, neither is the performance leader or significantly 'better' than the other at this MSRP price point, so the one with the lowest price wins. Simple as. :)
Originally Posted by MrWillyWonka I think I agree with bigz here, we are trying to reflect the game on best playable settings for a gaming rig, which everyone in this forum seem to have agreed that the X2 is the way to go for gaming now.
I also think that the fact that ATI hasn't taken in the use of dual core processors means the 7800GT is more futureproof and, overall, better.
I will say AGAIN I don't care about the use of the X2 really, as long as people are made aware of the fact that the nvidia drivers are multi-threaded and the ATi ones are not. As an example, whenever processors were benched on websites, ie P4 Vs Athlon XP or the non X2 64s, in tests that took advantage of the P4's HT capability like some video encoding, this is always stated in the results, ie something along the lines of "The P4 shows a performance advantage here and this is probably due to it's Hyperthreading capabilties which the software is taking advantage of".
I don't understand why it is so hard for you guys to understand what I'm trying to say. :(
Also as a side, ATi will be introducing multithread capable drivers soon as well so that negates the whole "futureproof" idea.
As it stands for the vast majority of gamers, drivers that support multithreading are a non-issue as they are still using single core cpus.
Comments 26 to 48 of 48
ReplyEdited to state SC:CT patched :p
I also upgraded the memory this time around too. That made some games react totally differently (like BF2 for example) because of the extra system memory, but I've not found much of a difference between X2 4800+ and FX-57 when I was playing around with 1GB of memory on an X2. All games that are currently available are all single threaded anyway, so I don't believe that the difference is as big as you're making it out to be.
The Ati card you tested is basically running like the nvidia card in those results with the /ONECPU switch. I would say that those are reasonably significant performance gains.
Sorry I couldn't find more I'm at work right now, but I have read that the performance gains can be even larger than that.
We're using drivers that are publicly available for NVIDIA hardware from an official source and we're using a CPU that was chosen by the majority of bit-tech readers who chose to vote in a recent poll in the hardware forum. You can't actually get the ATI drivers unless you actually buy an X1800XL. That's the only part of the whole review that is not publicly available and I challenge anyone to prove that the settings that we've reported do not fit inside the boundaries that we set out.
In fact, the only one that did was on the Sapphire Radeon X1800XL in F.E.A.R. where the average frame rate was 49.5 fps but the minimum frame rate was 18 fps and the game was a lot smoother than on the XFX 7800 GT Extreme scoring 60.4 fps average and 18 fps minimum. As I said in the review, I'd take the Sapphire's smoothness over the higher average (but noticeably choppier) frame rate that was returned by the XFX card any day.
/ponders Gosh, I must be NVIDIA's PR bitch because I used a dual core CPU.
I just like to see impartiality, and ommitting the fact that nvidia is using both cores in it's drivers (which I think is fantastic by the way) is not painting a fair picture of what is really happening.
Looks pretty good so far
http://www.xtremesystems.org/forums/showpost.php?p=1075680&postcount=72
Now you're telling me that I need to be impartial and not use the latest publicly available drivers on one IHV's products in order to disable an advantage that they have because the other team haven't adopted dual core support in to their drivers just yet. When ATI get around to adding dual core support in to their drivers, we'll use them without a doubt. Until then, we're in the "7800 GTX vs no product" situation again.
What if it takes ATI a year to get dual core support in to their drivers? Long after single core CPUs are (likely to be) history. Anyone buying a system today would be unwise buying a single core CPU unless they aren't going to use it for anything other than gaming. Most people who would drop $1000/£650 on a CPU would use it in their main computer, which also doubles as a gaming machine too.
Again, there's no doubts that we'll look at ATI's new products in the future when we get samples from other board partners. If and when they release new drivers that add dual core support, we'll revisit ATI's newer dual-core ready drivers with another product review.
We can only tell it like it is now, and this is how it is.
This is the top end, all a reviewer can do is let them off the leash and see what the best they can acheive is. A second core does very little for a single threaded game anyway (99.99% are), all the second core does is take off the background load that system services are using.
edit: oops I'm a bit late.
You haven't read this thread have you? If you had you would see that I'm not talking about the game being multithreaded, it is the DRIVERS that are multithreaded. Also even though the DX8.0 vs DX9.0 example is rediculous, if you did do that you would still inform your readers that you were doing it and why, I just didn't like the complete lack of mentioning the fact that the nvidia drivers are multithreaded. http://techreport.com/onearticle.x/8459
I also think that the fact that ATI hasn't taken in the use of dual core processors means the 7800GT is more futureproof and, overall, better.
I actually meant the GT being better (imo) than the x1800 - edited!
There is no doubt that the Radeon X1800XL has some nice features, including superior dynamic branching performance, better filtering quality, and the option for HDR+AA in all games that support HDR (and are patched to use hardware-based AA at the same time). The GeForce 7800 GT has a feature set that isn't too far off it and I'd say that NVIDIA have got the stronger multi-GPU platform at the moment by a long shot, but we'll have to wait and see how good CrossFire on X1800 series is when CrossFire cards are available.
The reason that the GeForce 7800 GT (as a single card solution) is a better buy at the moment is because the price differential is huge (~25%) and you can buy a pre-overclocked GT that outperforms or matches the X1800XL in just about every situation for a lot lot less. NVIDIA have a well-refined architecture behind GeForce 7, but there are some things missing that only become apparent when you look at ATI's architecture.
Admittedly, High Quality AF is not all that useable in the most demanding games out at the moment, and Adaptive AA isn't quite as efficient as Transparency SuperSampling from a real world gaming perspective. The time demos seem to show it up as quite useable though. Having said that, like HDR was on GeForce 6 hardware - it's good to have those features if the game is capable of using them without too much of a performance hit or, of course, you don't mind dropping the res/details to turn them on.
It's back and forth all the way, neither is the performance leader or significantly 'better' than the other at this MSRP price point, so the one with the lowest price wins. Simple as. :)
I will say AGAIN I don't care about the use of the X2 really, as long as people are made aware of the fact that the nvidia drivers are multi-threaded and the ATi ones are not. As an example, whenever processors were benched on websites, ie P4 Vs Athlon XP or the non X2 64s, in tests that took advantage of the P4's HT capability like some video encoding, this is always stated in the results, ie something along the lines of "The P4 shows a performance advantage here and this is probably due to it's Hyperthreading capabilties which the software is taking advantage of".
I don't understand why it is so hard for you guys to understand what I'm trying to say. :(
Also as a side, ATi will be introducing multithread capable drivers soon as well so that negates the whole "futureproof" idea.
As it stands for the vast majority of gamers, drivers that support multithreading are a non-issue as they are still using single core cpus.
-
« Previous
-
1
-
2
-
Next »
Discuss in the forums