Review: HD 2900XT 512Mb vs 8800GTS 320Mb

[Blacklash]

Here's another one for those interested:

http://www.madshrimps.be/?action=getarticle&number=1&artpage=2590&articID=581

 

[sleepeeg3]

Not exactly in the same price range.

 

[HyperTension]

Sigh... Here we go again....

Decent product, possibly.

Worth it? Nope. Benchmarking a card that has 256Mb less memory, and less potential (hardware specs), and that 8800GTS 320 does as well as it does. Also, bear in mind that they did NOT overclock the GTS.

Why not? That point alone makes their review loose weight in my mind with regards to legitimacy. Good review, but it's just regurgitating what most of us already know.

 

[tickle_me_emo]

Ok that guy ran a bunch of demos

fear demo - pretty much a tie since all the cards got poor minimum frame rates

tomb raider demo - tie, all of them get over 50 FPS, though the GTS renders 80 more FPS than both 2900s

NFS Carbon demo - tie, though the 2900 gets a few more frames

Dirt demo - tie again, though they all have poor minimum frames

quake 4 demo - tie again, slight edge to the 2900 though

oblivion not a demo i think - 2900 wins, by a lot

synthetic benchmarks - who cares?

heat/power - they don't have a direct comparison with the 8800...

and that normalizing chart at the end is a misleading piece of crap, winning by 1-2 frames average frames per second doesn't matter


So there's 5 ties and the 2900 wins at oblivion, that's not very good for a hotter, louder, more expensive card, and where's the stalker testing ?

Though it looks like in Europe the 8800 is the same price as the 2900, so it's not that bad of a deal over there.

 

[Snowdog]

Though it looks like in Europe the 8800 is the same price as the 2900, so it's not that bad of a deal over there.

Actually it is not, that is what I thought the table said at first, but you have to look around to find:

"Additionally we set the price of both cards to the same number to see which card would be the better deal if both were indeed available at the same €273 price tag."

So in essence they are saying the 2900 would have the best bang/buck if it was the same price as the GTS. Which is just a tad weird. Back in reality they are not the same price and it loses the bang/buck pretty badly.

 

[tickle_me_emo]

Actually it is not, that is what I thought the table said at first, but you have to look around to find:

"Additionally we set the price of both cards to the same number to see which card would be the better deal if both were indeed available at the same €273 price tag."

So in essence they are saying the 2900 would have the best bang/buck if it was the same price as the GTS. Which is just a tad weird. Back in reality they are not the same price and it loses the bang/buck pretty badly.

heh, i missed that

so the only legimiate positive comment the guy listed was the free game bundle

 

[Trepidati0n]

Actually it is not, that is what I thought the table said at first, but you have to look around to find:

"Additionally we set the price of both cards to the same number to see which card would be the better deal if both were indeed available at the same €273 price tag."

So in essence they are saying the 2900 would have the best bang/buck if it was the same price as the GTS. Which is just a tad weird. Back in reality they are not the same price and it loses the bang/buck pretty badly.

Even if the cards sold for the same price...the 2900XT still costs more...unless you think electricity is free.

 

[JMke]

Not exactly in the same price range.

that was indeed the point, cheaper NVIDIA card matching performance of more expensive ATI gear.

Sigh... Here we go again....

where is that?

Nope. Benchmarking a card that has 256Mb less memory, and less potential (hardware specs), and that 8800GTS 320 does as well as it does. Also, bear in mind that they did NOT overclock the GTS.

Why not? That point alone makes their review loose weight in my mind with regards to legitimacy. Good review, but it's just regurgitating what most of us already know.

how exactly does one loose weight by not OCing? I think GTS OC has been covered well enough in the past. Point is to see if, even with OC, the ATI can catch up.. end result: barely.

and that normalizing chart at the end is a misleading piece of crap, winning by 1-2 frames average frames per second doesn't matter

where did it say that it does? All it does is summarize the benchmarks results... where is stated that 1-2 FPS is a large win?

So there's 5 ties and the 2900 wins at oblivion, that's not very good for a hotter, louder, more expensive card, and where's the stalker testing ?

that's indeed not very good; article doesn't claim otherwise.

Which is just a tad weird. Back in reality they are not the same price and it loses the bang/buck pretty badly.

more like suggestive reviewing, I hope, if ATI decides to lower their price to make HD 2900 more competitive, they have to at least match the GTS 320's price.

-----

I get the impression you guys think that piece was PRO ATI? If you look at last page in the price/performance overview, you can see that the GTS holds a healthy 30% lead over the HIS and 24% over the Jetway model; so GTS 320 still remains best bang for the buck.

 

[tickle_me_emo]

where did it say that it does? All it does is summarize the benchmarks results... where is stated that 1-2 FPS is a large win?

that chart is very misleading and pointless

it's misleading because you took out the tomb raider score

and it's pointless because who really gives a damn if the 2900 gets 3-4 more average FPS?

 

[HyperTension]

I may as well cut and paste....

If your going to include results of an overclocked 2900XT, to balance it out, include the results of an overclocked GTS320. Why would you NOT include that in your spiffy graphs and charts.

Hmm probably because the author didn't want those results to skew the image that he / they wanted to present. I'm not throwing stones, but before you declare a certain card a winner / better value, whatever, make all things consistant / and let the results speak for themselves.

The XT has more memory, improved architech., later release, etc, etc. and the 320 is the only card they tested against it. Why was the 640GTS not included?. The XT was designed to go against the 640GTS. By overclocking the XT, and benchmarking the 320GTS only (and at stock speeds, not even an OC2), you are inherently skewing the results.

This leads to my general dismay at the way reviews (especially video cards) are skewed. Pro ATI / nVidia, whatever. I don't care, just give us correct non-horseshit result (see Tom's / Inq for specifics).

 

[tickle_me_emo]

I may as well cut and paste....

If your going to include results of an overclocked 2900XT, to balance it out, include the results of an overclocked GTS320. Why would you NOT include that in your spiffy graphs and charts.

Hmm probably because the author didn't want those results to skew the image that he / they wanted to present. I'm not throwing stones, but before you declare a certain card a winner / better value, whatever, make all things consistant / and let the results speak for themselves.

The XT has more memory, improved architech., later release, etc, etc. and the 320 is the only card they tested against it. Why was the 640GTS not included?. The XT was designed to go against the 640GTS. By overclocking the XT, and benchmarking the 320GTS only (and at stock speeds, not even an OC2), you are inherently skewing the results.

This leads to my general dismay at the way reviews (especially video cards) are skewed. Pro ATI / nVidia, whatever. I don't care, just give us correct non-horseshit result (see Tom's / Inq for specifics).

did you read the conclusion

Let’s wrap things up:

+ Decent high-end performance
+ Free games bundle
+ Overclocking headroom
+ Good CrossFire scaling with most games

- Price a bit on the high side
- Noisy in 3D mode
- High Power Consumption
- CrossFire texture bugs (driver issue)

ignore those green #'s they're pointless

 

[Citizen86]

I don't think he was exactly saying it was a good buy, although that whole chart at the end was rather strange... why make them the same price and see which did better? I would think that the whole point of benchmarking a more expensive card to a cheaper card would be to find out if the more expensive one is worth it... which they show it isn't worth every penny. Then they do some weird graph showing the opposite with basically a footnote.

*shrug*

 

[JMke]

that chart is very misleading and pointless

Pointless "how" ? comparing in % the performance difference of the games tested, what were they thinking! OMG!

Hmm probably because the author didn't want those results to skew the image that he / they wanted to present. .

I'm pretty sure the "image they want to present" is not you are referring to. If you can attentively read the last page of that article, something is askew it clearly states what products if best bang for the buck and why.

Comparing OCs is also no base whatsoever to decide upon, sample difference is high among cards; OC results for HD 2900 were included, tested with 3 units, to show you how much can be gained.

OC performance and scaling with GTS has already been covered extensively, but should not be taken into account price/performance wise, unless close to 99% of the units out there reach the OC tested.

 

[JMke]

I don't think he was exactly saying it was a good buy, although that whole chart at the end was rather strange... why make them the same price and see which did better?

*shrug*

more like suggestive reviewing, I hope, if ATI decides to lower their price to make HD 2900 more competitive, they have to at least match the GTS 320's price.

the same price was a footnote...

 

[jmackay]

That graph is confusing, because it looks like they are priced the same, where as they are saying if they were priced the same, the 2900 would be a good value.

That part isn't done very well and will just confuse most people who aren't paying close attention to what they are doing.

They should have just flat out said: "If this card was $100 cheaper, it would be a good card, however we have to recommend the 320mb 8800gts instead." or something similar.

 

[Citizen86]

Ah, the editor of MadShrimps. I actually really liked that OC article on the 8800's.

In reference to your conclusion with that graph, YES you said that you took the price out of the equation, but people can go to the games they want to see and look at the beautiful graphs that clearly explain exactly how each card did to see the straight comparison. I'm not really sure what I'm arguing here... Oh well. Nice site anyways.

 

[JMke]

They should have just flat out said: "If this card was $100 cheaper, it would be a good card, however we have to recommend the 320mb 8800gts instead." or something similar.

point taken, thank you, that is more straightforward; added this under the chart:

"If the ATI HD 2900 XT was €100 cheaper, it would be a very good deal, but as it stands, the Geforce 8800 GTS 320Mb is the better choice."

 

[tickle_me_emo]

Pointless "how" ? comparing in % the performance difference of the games tested, what were they thinking! OMG!

way to not quote the rest of my post

you choose to not put in the tomb raider demo (you should have also gotten rid of the quake 4 score as well since it doesn't really stress the cards)

and the chart shows in favor of ATI when infact the games were a lot closer than it suggests

 

[JMke]

way to not quote the rest of my post

you choose to not put in the tomb raider demo (you should have also gotten rid of the quake 4 score as well since it doesn't really stress the cards)

and the chart shows in favor of ATI when infact the games were a lot closer than it suggests

Tomb Raider Anniversary edition is quite taxing game engine those cards were doing 200+FPS 1600x1200 4xAA max detail -- could have also been Q3A and seen large differences

The game chart shows in performance favor of the ATI because, in most benchmarks it comes in front; maybe just by a few FPS, but when the average is 33FPS, only a few FPS extra translates into a few % extra too. you'll have a hard time spotting the difference between 100FPS average or 110FPS average, it's more noticeable when it's 33 vs 40 fps, but also look at the minFPS, they tend to annoy us the most . Under the chart all detail is shown too for those looking 1cm under it Nothing incorrect is shown there.

It's also the reason that, IF (if!) both cards were priced the same, the HD 2900 XT would make a better deal

 

[jmackay]

point taken, thank you, that is more straightforward; added this under the chart:

"If the ATI HD 2900 XT was €100 cheaper, it would be a very good deal, but as it stands, the Geforce 8800 GTS 320Mb is the better choice."

Wow, ty for that addition (though you might want to change it to slightly less than 100, like 90 or 80 or something so it matches the difference in price in €'s. I just remember when I originally read that review I was blown away when I saw that and thought both cards were priced the same, then realized that you were doing an "If it was" scenario.

 

[JMke]

although google finance tells me 1 U.S. dollar = 0.73 Euros, companies tend to just swap the $ with € in front of the prices, hoping to get away with a nice 25+% profit

 

[tickle_me_emo]

Tomb Raider Anniversary edition is quite taxing game engine those cards were doing 200+FPS 1600x1200 4xAA max detail -- could have also been Q3A and seen large differences

The game chart shows in performance favor of the ATI because, in most benchmarks it comes in front; maybe just by a few FPS, but when the average is 33FPS, only a few FPS extra translates into a few % extra too. you'll have a hard time spotting the difference between 100FPS average or 110FPS average, it's more noticeable when it's 33 vs 40 fps, but also look at the minFPS, they tend to annoy us the most . Under the chart all detail is shown too for those looking 1cm under it Nothing incorrect is shown there.

It's also the reason that, IF (if!) both cards were priced the same, the HD 2900 XT would make a better deal

you just pick and choose what points you want to respond to

i'm over it now since you actually picked a side and said the 2900xt has poor value

 

[Citizen86]

you just pick and choose what points you want to respond to

i'm over it now since you actually picked a side and said the 2900xt has poor value

lol, this conversation leaves me feeling a little used and abused also....

 

[Blacklash]

If you follow his link over below the "Oblivion" chart you will see a stock GTS 640Mb beats the HD 2900XT stock and overclocked.

XT was 43 and 48FPS respectively. The GTS 640Mb is 54FPS @ 1600x 4xAA|16xAF.

I'd say overall it makes a strong argument for the GTS 320Mb. If you play 1600x and up with 4xAA or greater I'd go 640Mb.

I thought it was an interesting read. I do wish he had overclocked the GTS 320Mb. As others mentioned, I would have preferred an overclocked 640Mb card to contrast with the HD 2900XT instead.

An 8800GTS 640Mb @ 580-621 would rough a HD 2900XT up in most all titles with 4xAA or greater active @ 1600x.

EDIT:

Here's overclocked 8800GTS 320Mb cards in "Oblivion" from the same site. At their best overclock they are even or 1 FPS behind the overclocked HD 2900XT @ 1600x 4xAA|16xAF.

http://www.madshrimps.be/?action=getarticle&number=12&artpage=2365&articID=541

47 or 48FPS for most of them. The overclocked XFX card does 44FPS out of the box. (580|900)

 

[JMke]

these individual runs can't really be 100% compared between articles, trouble with manual runthroughs
but yes, the GTS 640 will definitely outrun the HD 2900 XT.

you just pick and choose what points you want to respond to

and what points did I not respond to? like I said:

I get the impression you guys think that piece was PRO ATI? If you look at last page in the price/performance overview, you can see that the GTS holds a healthy 30% lead over the HIS and 24% over the Jetway model; so GTS 320 still remains best bang for the buck.

"picking sides" is so passé, I'd rather for the best price/performance balance