Zotac GeForce GTX 295 1792 MB

[wolf]

CHOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOP

 

[DarkMatter]

I'm always amazed at Nvidia cards' overclockability, and this one is even more impressive, being a dual card. But as much as I'm amazed, I'm also confused, because that demostrates that every Nvidia card could have much higher stock clocks and would thus perform and look better. Because 95% of the people won't OC and when looking at benchmarks that certainly hurts it's image, compared to what it could be, I mean.

Not that it affects too much the informed consumer, there are plenty of factory OCed cards for the same price as non overclocked ones for the ones who don't want to overclock, but still... I dunno.

 

[wolf]

I'm always amazed at Nvidia cards' overclockability, and this one is even more impressive, being a dual card. But as much as I'm amazed, I'm also confused, because that demostrates that every Nvidia card could have much higher stock clocks and would thus perform and look better. Because 95% of the people won't OC and when looking at benchmarks that certainly hurts it's image, compared to what it could be, I mean.

Not that it affects too much the informed consumer, there are plenty of factory OCed cards for the same price as non overclocked ones for the ones who don't want to overclock, but still... I dunno.

mmm, its great that they all OC so well, and honestly i think nvidia choose clock speeds to purposely keep them in line with ATi counterparts.

I do doubt that we could see a 4870X2 boast such nice gains from overclocking, and its really interesting to see these GT200b cores oc so well on 1.0375v !

currently i run 1.125v thru mine for 710/1566/2302 rock stable, which should be give an edge on the numbers in the TPU review, and with more volts this card will do 756/1620/2322! which is just stellar for what this card is, and will easily best GTX 280 SLi at those speeds (unless memory limited)

oh and FYI my 295 is the Zotac Model too, very pleased with it

 

[DarkMatter]

mmm, its great that they all OC so well, and honestly i think nvidia choose clock speeds to purposely keep them in line with ATi counterparts.

I do doubt that we could see a 4870X2 boast such nice gains from overclocking, and its really interesting to see these GT200b cores oc so well on 1.0375v !

currently i run 1.125v thru mine for 710/1566/2302 rock stable, which should be me an edge on the numbers in the TPU review, and with more volts this card will do 756/1620/2322! which is just stellar for what this card is, and will easily best GTX 280 SLi at those speeds (unless memory limited)

oh and FYI my 295 is the Zotac Model too, very pleased with it

Yeah those Zotac cards own. I've seen pretty impressive OCs on them. It wouldn't surprise me if taking an average of different vendors, they would be the best OCers.

 

[wolf]

Yeah those Zotac cards own. I've seen pretty impressive OCs on them. It wouldn't surprise me if taking an average of different vendors, they would be the best OCers.

I am so pleased with this card its not even funny, far and away the most powerful graphics subsystem Ive ever owned.

It just makes me so eager so see what Nvidias next best is that can par this card alone, as a GTX280/285 does to a 9800GX2 now that drivers have matured.

also, im not sure where on TPU i read it but there seems to be confusion over exactly what the cards in a GTX295 are, with most reviewers saying " they are between GTX260's and GTX280's in SLi", the easiest way to think of it is this folks.

The GTX260 has 28 ROPS, 448-Bit mem interface, 576/1242/999 mhz and 192 or 216 sp's
The GTX280 has 32 ROPS, 512-Bit mem interface, 602/1296/1100 mhz and 240 sp's

each core in a GTX295 is as follows:

28 ROPS, 448-Bit mem interface, 576/1242/999 mhz and 240 sp's

so in every single way these are pretty much GTX260's but with all sp's unlocked, the easyest way to think of it is as a GTX 260 core 240 model, as oppose to core 216 or 192.

 

[DarkMatter]

It's funny, but I never thought of it that way, GTX260 core 240, and it's exactly that, indeed lol. I usually do as most people do: "a GTX260, but almost a GTX280... that is, with 240 SPs, but only 28 ROPs..." And you end up telling them the whole specs, while they are already instructed on what a GTX260 216 is.

It doesn't matter to me anyway, I never have names on my head, I have the full specs and much more. When I hear GTX260 everything comes to my mind and so is with any GPU that I know. In my head it's not the GTX260 216, it's the "7x4 ROP, 7x64bit, 9 cluster x 3 TPC x 8 SP = 216 SP and ..." card. I don't know how to explain it lol, but yeah, I know I'm sick.

 

[spearman914]

It's funny, but I never thought of it that way, GTX260 core 240, and it's exactly that, indeed lol. I usually do as most people do: "a GTX260, but almost a GTX280... that is, with 240 SPs, but only 28 ROPs..." And you end up telling them the whole specs, while they are already instructed on what a GTX260 216 is.

It doesn't matter to me anyway, I never have names on my head, I have the full specs and much more. When I hear GTX260 everything comes to my mind and so is with any GPU that I know. In my head it's not the GTX260 216, it's the "7x4 ROP, 7x64bit, 9 cluster x 3 TPC x 8 SP = 216 SP and ..." card. I don't know how to explain it lol, but yeah, I know I'm sick.

Please elaborate the numbers. lol

 

[newtekie1]

I'm always amazed at Nvidia cards' overclockability, and this one is even more impressive, being a dual card. But as much as I'm amazed, I'm also confused, because that demostrates that every Nvidia card could have much higher stock clocks and would thus perform and look better. Because 95% of the people won't OC and when looking at benchmarks that certainly hurts it's image, compared to what it could be, I mean.

Not that it affects too much the informed consumer, there are plenty of factory OCed cards for the same price as non overclocked ones for the ones who don't want to overclock, but still... I dunno.

I believe nVidia purposely leaves a large overclocking headroom on their cards when picking stock clocks for several reasons.

1.) To make sure the card is stable, even in the worst conditions, such as cases with little airflow.
2.) To impove yeilds. Lower stock clocks means more cores that pass quality control.
3.) To make their partners happy. NVidia has a running history with their partners for allowing overclocked versions of their cards. If nVidia started putting out cards with the clock speeds already pushed to the max, their partners wouldn't be happy.

 

[wolf]

It's funny, but I never thought of it that way, GTX260 core 240, and it's exactly that, indeed lol. I usually do as most people do: "a GTX260, but almost a GTX280... that is, with 240 SPs, but only 28 ROPs..." And you end up telling them the whole specs, while they are already instructed on what a GTX260 216 is.

It doesn't matter to me anyway, I never have names on my head, I have the full specs and much more. When I hear GTX260 everything comes to my mind and so is with any GPU that I know. In my head it's not the GTX260 216, it's the "7x4 ROP, 7x64bit, 9 cluster x 3 TPC x 8 SP = 216 SP and ..." card. I don't know how to explain it lol, but yeah, I know I'm sick.

i know exactly how you feel, im probably not sooooooo in depth but im thinking about every aspect of the card when i think of it, not just something as simple as GTX260 (especially with the amount of variants now, 192sp, 216, then the new 55nm's....)

but so many people ask me about computer gear, im sort of the go to guy for my group of mates. when i got my GTX295 i had to find a way to explain what they were, cos they arent GTX260's and they aren't GTX280's, so the GTX260 core 240 was born.

my guess is they wont be released as a single card because they would perform too close to a GTX280 i imagine.

and @ newtekie, you make some very good points on card clockspeeds. cheers to you both

 

[Drizzt5]

Very nice review!

 

[DarkMatter]

I believe nVidia purposely leaves a large overclocking headroom on their cards when picking stock clocks for several reasons.

1.) To make sure the card is stable, even in the worst conditions, such as cases with little airflow.
2.) To impove yeilds. Lower stock clocks means more cores that pass quality control.
3.) To make their partners happy. NVidia has a running history with their partners for allowing overclocked versions of their cards. If nVidia started putting out cards with the clock speeds already pushed to the max, their partners wouldn't be happy.

Yeah, yeah, I know, but my comment came precisely after thinking of that. Except point three, which is the one that matters most IMO, others don't make too much sense anymore. Let's face it, GT200 is as stable running at 660 mhz as it is at 600mhz, no matter the conditions. In average conditions any Nvidia card is stable with a 10-15% OC and with good (not even excelent) airflow even 20% is rock solid. If we factor in voltmods the margin is even greater, and power consumption and temperatures never really get much higher, don't get dangerous. So the cards could have come with 10% more clocks without any issues.

Clocks to improve yields is a different story and in the case of GT200, might be a fair reason, because at the time they were finalizing the specs they had a lot of problems that were soon fixed, but late for the specs probably. There I can start understanding.

But I think it is certainly because of the partners that they clock them low, so that the latter look like heroes with their outstanding clocks. Because prices of factory OCed cards aren't usually much more expensive than the competitor's cards, I guess Nvidia thinks that at the end of the day the customer gets the same. But IMO it hurts the image of the cards, GT200 fiasco at launch would have never happened if the cards were running a safe 10% higher. No more "Ati has released a slightly faster card for $100 less for example", it would have been a "10% slower card for $100 less", which at the high end wouldn't be that bad. Nowhere in the media or people's heads you will see that the GTX260 is >>10% faster than the HD4870 when both are at their max OC.

 

[wolf]

Nowhere in the media or people's heads you will see that the GTX260 is >>10% faster than the HD4870 when both are at their max OC.

i hear you there man, but maybe if they did clock it faster nvidia would maybe try justify a higher price, as at stock speed they trade blows well, but as you say, at mac OC a GTX260 will move well ahead of a 4870.

however, since the GTX295 holds the absolute crown, and temps are proven to be very manageable, i would have loved to have seen higher clocks.

even something like 625/1350/1100 would have just further solidified the crown. i'm willing to wager any GTX295 out there would run at those speeds without so much as a hiccup.

 

[MoonPig]

Great work Wizz, one of your earlier reviews actually helped me pick my last Video Card.

Do you think there is any chance you could take a screen shot of the setting you are using when you next review a card?

Thanks for the review.