Reviews - NVIDIA (BFG) GeForce GTX 295 - the champ is back!

[drunkenmaster]

yup, though the 65nm part is 576mm, the 55nm is unlikely to be smaller than 450nm, probably closer to 500 than 450nm. Its supposed to be a 16% drop half node in size from 65 to 55nm, if you get lucky and everything lines up better than at 65nm, you might save some die space, if it doesn't quite shrink perfectly(which considering this was supposed to be out 4 months ago and has taken several respins to even get out) you probably will get less than a 16% die size reduction which will put it at a little above or below double the size of an ati core. Power per square MM is far higher on the ATI core, because its half the size, thats really where cooling largely comes into play.

But Webby is also right, its also down to ATi's quite strangly and stupidly low fan speed at stock, you can bump it from 25%(where it hangs for me) to 37% or so without really being able to hear it anymore but get a decent drop in temps, up to 45% and you get a ok drop and not to much noise 50% fan and you'll go from 95c load to 70c load but will be fairly damn noisey.

THe thing is ATi fans are by default very slow which does result in high temps which ISN'T a bad thing, just isn't a great thing either. I haven't seen mention how loud the 295 fan is, it might be 3 times as loud and thats why its cooling better, I really don't know.

Either way, I'm going to sue someone next gen if there aren't some 4850x2 style heatsinks on the cards from release, or at least someone gets a 3rd party cooler out for next gen cards basically on release. its ridiculous waiting 6 months for a cooler only for the next gen to be almost upon you

[hurleybird]

Unfortunately, there is a beta driver out for the 4870X2 that gives monster improvements, invalidating this review. With the 9.1 beta driver the 4870X2 and the GTX295 are neck and neck, with Nvidia generally winning at lower res/AA/aniso, and AMD taking the crown at higher res/AA/aniso. It's a shame that a lot of work went into this review for nothing, but that's how things currently stand.

I can't post any URLs because of the rules on this board, but check out computerbase.de and HardOCP.

[moogle]

I can't post any URLs because of the rules on this board, but check out computerbase.de and HardOCP.

We've seen them (or someones posted some of the results here). The 9.x drivers right?
You need a few more posts before you can post urls anyway. Spam protection stuff

[hurleybird]

I didn't see any mention of the new drivers in this thread, so I figured I'd give them a mention. Is Hexus getting sloppy (using beta drivers from NVIDIA but not AMD) or is AMD PR just that bad?

[kalniel]

I didn't see any mention of the new drivers in this thread, so I figured I'd give them a mention. Is Hexus getting sloppy (using beta drivers from NVIDIA but not AMD) or is AMD PR just that bad?

I'd like to think AMD were nice enough not to try and spoil a competitors launch by splashing out beta drivers, but I think in this case they aren't officially available to the public so it's fair enough not to review with them - the nVidia beta drivers tested here are what you actually get with the card I assume.

The review is hardly 'for nothing', that's a faintly ridiculous statement.

[hurleybird]

I'd like to think AMD were nice enough not to try and spoil a competitors launch by splashing out beta drivers, but I think in this case they aren't officially available to the public so it's fair enough not to review with them - the nVidia beta drivers tested here are what you actually get with the card I assume.

The review is hardly 'for nothing', that's a faintly ridiculous statement.

If AMD were nice enough? NVIDIA has a history of ruining launches with "splashing out beta drivers" -- they started it when the Radeon 8500 launched -- so I guess you reap what you sow. As for the review being "for nothing", the numbers are wrong, or at the least will be wrong in a day or two when the 9.1 betas are publicly available. If that doesn't make a review "for nothing" I don't know what would. Again, I'm not blaming Hexus, for all we know it could be AMDs fault they didn't get the new driver, but that doesn't change the worthlessness of the review. Of course, only half the numbers are wrong, and all Hexus needs to do to recover this review is retest the other half.

[kalniel]

Well I'm sorry that you feel you've wasted your time reading this review for free. Perhaps you should cut your loses and rather than argue about it go and read another review that you feel better reflects your viewpoint?

[matty-hodgson]

and all Hexus needs to do to recover this review is retest the other half.

so you expect Hexus to retest all their reviews whenever AMD or Nvidia release a new driver update?

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

[shadowmaster]

I think this review uses the latest driver (catalyst 9.x)-

http://translate.google.com/translat...istory_state0=

[hurleybird]

so you expect Hexus to retest all their reviews whenever AMD or Nvidia release a new driver update?

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

No, only when the drivers *significantly* change the competitive situation should a video card need to be retested. Hexus and everyone else retested the Core 216 when big bang 2 came out, and the 9.1 driver seems to give even bigger improvements than that. The contrast between how Hexus has treated AMD and NVIDIA with respect to drivers is striking, but again, it's possible that AMD PR is at fault so I won't go blaming them for that yet. The sad thing is that, no matter who is to blame, this review should have tested with 9.1 beta to begin with and shouldn't need to have been retested. And I wasn't quite honest when I said that this review is for nothing, it's worse than that, giving people the wrong idea is worse than saying nothing at all.

[kalniel]

Hexus and everyone else retested the Core 216 when big bang 2 came out, and the 9.1 driver seems to give even bigger improvements than that.

Maybe if AMD didn't feel the need to bang out drivers every month regardless then they could rightly claim some attention when they bring out a new one, but if nVidia only release big driver updates once in a while then you can't blame sites being rather more eager to retest when that happens than every month.

And I wasn't quite honest when I said that this review is for nothing, it's worse than that, giving people the wrong idea is worse than saying nothing at all.

You've already made your point, now you're almost trolling.

[hurleybird]

Maybe if AMD didn't feel the need to bang out drivers every month regardless then they could rightly claim some attention when they bring out a new one, but if nVidia only release big driver updates once in a while then you can't blame sites being rather more eager to retest when that happens than every month.

While NVIDIA has less frequent WHQL releases, it is constantly pushing out new beta drivers. Still, driver release frequency shouldn't determine whether or not to retest an exceptional driver that delivers a massive increase above and beyond ordinary improvements. To say otherwise is a very thin rationalization that you'd expect from an NVIDIA (I hate the word, but I have to say it here) fan boy. Yes, I know you use a Radeon, and I don't think you're a fan boy, just that you used some poor logic.

You've already made your point, now you're almost trolling.

No. It would be very easy to look at how Big Bang II was handled and how Cat 9.1 was handled, and start shouting about how biased Hexus is. However, despite how things currently look, I have been consistent in not assigning blame until more information is available -- for all we know AMD didn't send Hexus the driver, or sent it far to late to do anything about. That much should tell you that I'm not a troll. Despite that, no matter who is at fault, a review that has the wrong numbers and the wrong conclusion isn't going to help buyers, it's going to hurt them and contribute to people making the wrong choice. That's why the review is worse than nothing, and I challenge you to refute that logic. A simple disclaimer about cat 9.1 at the end of the review would solve that (but might hurt traffic), but a retest is necessary for the review to actually be helpful.

[moogle]

A simple disclaimer about cat 9.1 at the end of the review would solve that, but a retest is necessary for the review to actually be helpful.

Don't worry, there are other reviews that have or will have the 9.1 stuff updated. Just because it's not like that in this review I think you shouldn't be so concerned about it

Not telling you to go away, just it's not something worth arguing about on this review.

[hurleybird]

Don't worry, there are other reviews that have or will have the 9.1 stuff updated. Just because it's not like that in this review I think you shouldn't be so concerned about it

Not telling you to go away, just it's not something worth arguing about on this review.

Yeah, but you know how these things start. I make a statement, someone doesn't like it and responds, I respond to his criticism, and the thing kind of balloons. I think I've probably said all I need to though.

[moogle]

Yeah, but you know how these things start. I make a statement, someone doesn't like it and responds, I respond to his criticism, and the thing kind of balloons. I think I've probably said all I need to though.

I know. That "someones wrong on the internet" cartoon springs to mind
But hey if you live in the UK keep going I say. 14 more replies and you get yourself some Scan free shipping

Although I'm guessing you're from USA.

[matty-hodgson]