Geforce GTS 450 appears on shelves

[Eastcoasthandle]

don't be stupid. that 5770 is the same speed as a normal one.

An equal comparison is needed here. It doesn't make sense to provide performance results using a OC EDITION 450 against a SILENT EDITION 5770. Then you reply to me for point it out with name calling. No need for the unprovoked argument.

Well, according to their graph the passive 5770 is only 4 degrees hotter than the fanned 450; perhaps it overclocks really well? But yeah, strange comparison.

The temps are telling aren't they? But it I find it odd to use a silent edition for performance comparison review when the gist of such a card is passive cooling.

[trinibwoy]

An equal comparison is needed here.

"Equality" is primarily based on price and the 450 is cheaper. How is that an unfair comparison?

[Newblar]

"Equality" is primarily based on price and the 450 is cheaper. How is that an unfair comparison?

well a giant passive heatsink isnt cheap...
its apples to oranges

compare a 5770 oc to 450gts oc

[Krizby87]

ok so if I remove the default cooling on some card and slap on a passive cooler, does that make it any different if the clocks are the same ? From the benchmarks I see GTX 450 and HD 5770 are pretty much equal, now it just comes down to brand preferences, whatever floats your boat. Here in France the HD 5770 is selling for 175 eur, GTX 460 for 250 eur, hopefully the GTX 450 comes at 150 then it's good for me

[Eastcoasthandle]

"Equality" is primarily based on price and the 450 is cheaper. How is that an unfair comparison?

ok so if I remove the default cooling on some card and slap on a passive cooler, does that make it any different if the clocks are the same ? From the benchmarks I see GTX 450 and HD 5770 are pretty much equal, now it just comes down to brand preferences, whatever floats your boat. Here in France the HD 5770 is selling for 175 eur, GTX 460 for 250 eur, hopefully the GTX 450 comes at 150 then it's good for me

I think you are implying if I'm aware that HSF combo does more then cool the GPU? Yes, I am aware of that. So it always a good idea to make sure that not only the GPU has active cooling but the video card circuitry as well when providing comparison results. Which is why an OC actively cooled video card vs passively cooled stock video card is not a good comparison.

[Ejizz]

^^^
Why didn't they overclock it as well then?

[Eastcoasthandle]

^^^
Why didn't they overclock it as well then?

That's one reason people don't find the review valid. Could they have OC an actively cooled 5750 as well? Sure...

[Ejizz]

^^^
Because Nv bought the review, they own it, it's only fair they get to make there cards look better than they actually are.

[~CS~]

Just wait a month or so ,if trend continues the card will be 100$,gtx460 768 was 200$ at launch,now 160-170$.

[B-Shot]

FTW OC them all passive cooled (same HS) and see what happens, then they will have red apple-v-green apple and maybe the 450 wont fall on it's face at high res.

[Krizby87]

I think you are implying if I'm aware that HSF combo does more then cool the GPU? Yes, I am aware of that. So it always a good idea to make sure that not only the GPU has active cooling but the video card circuitry as well when providing comparison results. Which is why an OC actively cooled video card vs passively cooled stock video card is not a good comparison.

nah I mean if the video card is cooled well it doesn't matter if it has a passive or active cooler. You can put a 120mm fan right next to the passive cooled card and that makes it way better cooled than the default HSF, and that the way I prefer over the small little cooler.

Anyways most people would agree that if the GTX 450 is selling cheaper than HD 5770 then it's a better deal.

[trinibwoy]

I'm still trying to understand why in the world it matters if one solution is passive or if one is factory overclocked? Even if they used a normal 5770, those run from $125-$180. How is it unfair to compare the 5770 to a $130 card?

/boggle

[cegras]

It's not like the comparison is fair or not. The main point is that the GTS450 draws more power for marginally less performance, and that anyone who can buy a 5770 can OC it to clear the difference for roughly the same price.

[Calmatory]

I'm still trying to understand why in the world it matters if one solution is passive or if one is factory overclocked? Even if they used a normal 5770, those run from $125-$180. How is it unfair to compare the 5770 to a $130 card?

/boggle

Maybe not from a performance point of view, the numbers do not change. But think about it. People look for numbers, do you think that overclocked edition vs. silent edition is anywhere near apples to apples comparison number wise?

Now to the point. Why on earth do these sites even compare OC edition cards to vanilla cards? People care about numbers, so doesn't the OC edition quite much skew the numbers? But hey, it's no secret that the companies themselves ship the OC edition cards to the sites to be reviewed. Even THEY understand that it skews the numbers for them.

[trinibwoy]

It's not like the comparison is fair or not. The main point is that the GTS450 draws more power for marginally less performance, and that anyone who can buy a 5770 can OC it to clear the difference for roughly the same price.

Agreed. That certainly wasn't the "main point" that was being argued though.

Maybe not from a performance point of view, the numbers do not change. But think about it. People look for numbers, do you think that overclocked edition vs. silent edition is anywhere near apples to apples comparison number wise?

Now to the point. Why on earth do these sites even compare OC edition cards to vanilla cards? People care about numbers, so doesn't the OC edition quite much skew the numbers? But hey, it's no secret that the companies themselves ship the OC edition cards to the sites to be reviewed. Even THEY understand that it skews the numbers for them.

I'm not sure what numbers you're referring to. Price is the only number that matters. Are you saying that reviewing a $130 reference GTS 450 is more "fair" than reviewing a $135 factory overclocked one? Why is that? More than half the 450's on newegg are at higher than reference clocks.

[Eastcoasthandle]

Agreed. That certainly wasn't the "main point" that was being argued though.

No, that was the main point. Like I said in the previous post, you don't have to OC just a 5770 they could have OC a 5750. As for price, you can accompany that with heat and power consumption since this isn't an enthusiast level video card. HTPC, OEM replacements, etc would also consider that segment of cards.

[Calmatory]

I'm not sure what numbers you're referring to. Price is the only number that matters. Are you saying that reviewing a $130 reference GTS 450 is more "fair" than reviewing a $135 factory overclocked one? Why is that? More than half the 450's on newegg are at higher than reference clocks.

So you'll always get the cheaper card?

People check the performance. This review indicates that GTS 450 would actually be somewhat matching HD5770, while in reality it isn't. Sure, it is actually clear that it's a OC edition, but that doesn't change the fact that they're comparing "GTS 450" to "HD 5770" here. It just boosts the GTS 450 numbers.

I'm not even interested in this particular case, since for me these cards are worthless as I have no plans to buy any discrete cards for forthcoming years as I don't play any modern games. But what I hate is this current trend of making these reviews. It isn't hard to see that Nvidia and AMD are sending out samples with intent to make their cards look better than competitors. Heck, what was the buzz with GTX 280 and 1+n sites doing more or less the same review at exactly the same time, with more or less exactly the same games tested? Then, not only that, but the websites which run the tests with their "test suite"(which usually consists of timedemos) and add the results to a "database" and then one can compare the tests against other cards found in the "database". Just look at this thread "Nvidia cards usually gain performance with driver releases", that should say lots about the credibility of such test "databases".

Guess people don't care about things like small inaccuracies when a driver release can bring up the performance by some 5-10 %, possibly multiple times. And yet difference of 15-25 % between cards is considered a "X COMPLETELY DESTROYS Y IN TEST Z!!". It's just pathetic.

I partly understand that TweakTown is comparing products of the same price range, but I claim that the product was brought to them by Nvidia, with the intent to make their product look better than the actual product they ship to the market is.

Isn't the idea of these reviews to give a consumer an idea of how the product performs? In that case, one could possibly think that accuracy of the results is very important, yet it seems that most of the sites aren't really caring about it. If Nvidia or AMD tells them to use X, Y and Z to test their hardware, the sites will gladly do so, and then the result database with old and not-to-date results. Not to mention that some sites have been using old versions of drivers for Nvidia or AMD in order to make the product look inferior compared to competitors offering(who more or less provided the new heardware).

Is there any site which actually cares to provide accurate and up to date reviews?

[trinibwoy]

No, that was the main point. Like I said in the previous post, you don't have to OC just a 5770 they could have OC a 5750. As for price, you can accompany that with heat and power consumption since this isn't an enthusiast level video card. HTPC, OEM replacements, etc would also consider that segment of cards.

Sure they could have chosen to compare it to an OC 5750 but I'm not sure that would change the outcome or conclusion. There's a 710Mhz 5750 for $135 AR and a 720Mhz one for $170 at the egg.

And of course you have to consider heat, power consumption etc. But that's what reviews are for. However, you don't select cards for comparison according their heat or power consumption. They are chosen based on price. Why are you guys acting like this is such a foreign concept?

People check the performance. This review indicates that GTS 450 would actually be somewhat matching HD5770, while in reality it isn't. Sure, it is actually clear that it's a OC edition, but that doesn't change the fact that they're comparing "GTS 450" to "HD 5770" here. It just boosts the GTS 450 numbers.

So your problem is that reviews of factory overclocked cards give the wrong impression because they are faster than reference cards? That's unbelievably silly since people don't shop based on whether the products are reference or not. The factory overclocked cards are real products available for sale, not some shady marketing trick. This review indicates that Galaxy's OC edition 450 sometimes matches a stock 5770, that's all.

I partly understand that TweakTown is comparing products of the same price range, but I claim that the product was brought to them by Nvidia, with the intent to make their product look better than the actual product they ship to the market is.

So the factory overclocked editions aren't actual products shipped to the market and available for sale in volume? They're just gimmicks to fool people into thinking the 450 is faster than it really is? I could understand if it was something like the 480 lightning or 5970 ares you were complaining about but you're really not making any sense here complaining about cheapo cards anybody can buy, some of which aren't any more expensive than the reference options.

[bamtan2]

some of you guys are going way overboard.

they compared regular video cards you can buy in stores. they didn't hide the names or the prices. they didn't overclock one. this is apples to apples.

[Ejizz]

^^^
Don't worry your disoriented and dizzy for a reason, I believe it's called 'spin', and good spin always has room reserved for 'plausible deny-ability'... Nv are masters at 'spin'.
However PR problems can arise when 'spin' is used a little too often, and 'plausible deny-ability' becomes less and less plausible, which can then motivate the more 'perceptive' populace to speak out against such 'spin', which then feeds a viscous cycle of spin/speaking out, and that was how 'Nvidia bashing' was born.

[JohnJohn]

Can someone point me out to some reference model full review?, NOT Oced.

Most people are going to buy the reference model most of the time, how can they compare it with other video cards when the only reviews available are Oced models..?

This is the closest thing available i say:


http://www.techpowerup.com/reviews/P..._Sonic/28.html


Reference model is on par with 5750

[RAW-Raptor22]

In other words, reference model is about equal to an 8800Ultra...

Took long enough...

[Spectrobozo]

so, the reference GTS 450 is equivalent to a (1 year old and similarly priced) reference HD5750, while the 5750 uses +- 70% of the power used by the GTS 450!?
not the greatest thing you would expect from 1 year of evolution, but.. oh, well is at least one more option, and seems to be a good overclocker to, also the support for physx and the performance in some games is quite nicer,

[Monstru]

Can someone point me out to some reference model full review?, NOT Oced.

Most people are going to buy the reference model most of the time, how can they compare it with other video cards when the only reviews available are Oced models..?

It's not in english, but it is reference design and stock clocks

Gigabyte GTS 450 OC - enter GF 106 - Google Translate

not the greatest thing you would expect from 1 year of evolution

A few funny Nvidia facts, to spice this up a little bit:

-13 September 2009 was the launch date for HD 5750 and HD5770
-29 Octomber 2007 was the launch date of 8800GT, the first G92 implementation.


Nvidia managed to finnaaaaaaaaaaaaaaally replace G92 in just a little bit under 3 years. All hail the immortal G92 and his countless re-incarnations

[tajoh111]

It's not in english, but it is reference design and stock clocks

Gigabyte GTS 450 OC - enter GF 106 - Google Translate



A few funny Nvidia facts, to spice this up a little bit:

-13 September 2009 was the launch date for HD 5750 and HD5770
-29 Octomber 2007 was the launch date of 8800GT, the first G92 implementation.


Nvidia managed to finnaaaaaaaaaaaaaaally replace G92 in just a little bit under 3 years. All hail the immortal G92 and his countless re-incarnations

Man is there a bottle neck there. It seems upping the clocks don't do as much wonders as overclocking the memory. Considering it took so long to make a suitable g92 replacement, it took way too long and considering the legacy, it not nearly as strong a product.

This product is going to look stale in 2 or 3 months.

Its like they removed the good that made the gtx 280 and gtx 8800 and replaced it with this CGPU crap that makes this architecture so much less efficient.

This architecture needs a clean up pronto, ALA 2900xt --> 3870.

The gts is an ok product, but it needs to go down about 10 bucks to make it competitive.