Search
Sign in with
Sign up | Sign in

Why such poor performance for the 7950 GX2?

Tags:
Last response: in Graphics & Displays
Share

Hi,

I am the (previously) proud owner of a Leadtek 7950 GX2, and have been debating on upgrading to an 8800GTX (or possibly an R600 if it ever comes out). I looked at the Toms Hardware VGA charts to see what sort of performance increase I would be expecting.

I was amazed at how poorly the 7950 GX2 performed compared to other cards.

For example, Prey at 1600x1200 (which is the res i like to play at, well 1680x1050), 32-bit, 4xAA, 8xAF performed only 1FPS faster than an AGP 7800GS+, and was a lot slower than other supposedly poorer performing cards.

And it's not just that game, similar results come up in Half Life 2 Episode 1.

Are these results skewed by anything, or does the card really perform that slowly relatively. I was always led to believe that this card should perform at about the level (maybe a touch slower) of the X1900XTX, which seems to outdo it by a mile in basically every benchmark on the VGA Chart.

If the results are skewed, say for example by driver revisions, then isn't THG painting a biased view of all cards by not keeping them all as up-to-date as possible,

If they aren't skewed........ Dammit :!:

I must admit, on the whole I've been happy with the performance of my card, and I have overclocked it fairly successfully, getting a score of about 13500 on 3D Mark 05. DirectX 10 was previously the main reason for me to upgrade, but if my card performs that badly relatively, then it's definately time for me to upgrade.

Does anyone have any comments regarding the performance of this card, or the charts in general.

My specs are this....

Core 2 Duo E6600
Leadtek 7950GX2
Intel 975XBX M/B
Corsair 2Gb XMS DDR2
XClio 550W PSU (this will go if I upgrade GPU)
Viewsonic VX2025
Logitech G15 K/B & G5 Mouse

Cheers,

More about : poor performance 7950 gx2

Register or log in to remove

I wouldn't go so far as to say THG reviews a complete and utter joke, but they're definitely shady. They can be useful sources sometimes of information but must be viewed with discretion.

The same goes for the CPU and GPU charts. If something looks wrong, it probably is.

Sadly, that sounds just like how I'd describe the Inquirer.
Related ressources
Ask the community

A link to the article would be a big help.

But some of the specs are: X1900XTX vs. 7950 GX2

ATI: Core Clk 650MB / Pixel Clk 16 / Fill Rate 10400 / Texture Clk 16 / Fill Rate 10400 / Mem Clk 1550 / Mem Bus 256b / Mem Bandwidth 49.6 GBs
Nvidia: Core Clk 1gig / Pixel Clk 32 / Fill Rate 16000 / Text Clk 48 / Fill Rate 24000 / Mem Clk 1200 / Mem Bus 2X256b / Mem Bandwidth 76.8 GBs

With the exception of Memory Clock Speed... the 7950GX2 should be the performer here.

Quote:
So why would THG have the results as so poor?



I've seen some charts on some sites, I think even THG, saying that the 7950GX2 outperforms the 7900GTX. Which is completely untrue.

Quote:
A link to the article would be a big help.

But some of the specs are: X1900XTX vs. 7950 GX2

ATI: Core Clk 650MB / Pixel Clk 16 / Fill Rate 10400 / Texture Clk 16 / Fill Rate 10400 / Mem Clk 1550 / Mem Bus 256b / Mem Bandwidth 49.6 GBs
Nvidia: Core Clk 1gig / Pixel Clk 32 / Fill Rate 16000 / Text Clk 48 / Fill Rate 24000 / Mem Clk 1200 / Mem Bus 2X256b / Mem Bandwidth 76.8 GBs

With the exception of Memory Clock Speed... the 7950GX2 should be the performer here.


There is no particular article, these are the the VGA charts that THG put out and are supposedly updated on a regular basis. Though by the sounds of it, there's no point in looking at it anyway. Nevertheless, here is a linky.

THG VGA Chart

Thanks for the replies anyhoo, tbh I generally check lots of different sources before I buy hardware, but it was the THG VGA Chart that confused me. I always thought of THG as a reputable hardware reviewer. I've been told! :oops: 

Cheers

Quote:
A link to the article would be a big help.

But some of the specs are: X1900XTX vs. 7950 GX2

ATI: Core Clk 650MB / Pixel Clk 16 / Fill Rate 10400 / Texture Clk 16 / Fill Rate 10400 / Mem Clk 1550 / Mem Bus 256b / Mem Bandwidth 49.6 GBs
Nvidia: Core Clk 1gig / Pixel Clk 32 / Fill Rate 16000 / Text Clk 48 / Fill Rate 24000 / Mem Clk 1200 / Mem Bus 2X256b / Mem Bandwidth 76.8 GBs

With the exception of Memory Clock Speed... the 7950GX2 should be the performer here.


The 7950 GX2 does not have an actual effective core clock of 1ghz :roll: Same for the mem clock of 1200mhz.

Those specs can be very misleading for noobies who dont understand how it all works.

The actual effective clock for the core is 500mhz and mem is 600mhz.

GEEEEEZ I try to save a little time in typing and get a bomb! Thanks for clearing up exactly what I just read.

So for the inconvienience to others... its 2 x 512mb and 2 x 600 which makes an effective 1 GB & 1200MHz :?

the 7950GX2 is easy faster than the 7900GTX in the same way a 7900GT SLI is faster than a single 7900GTX, tahts all it is is two 7900GT's in SLI on the same card.

It reads more like this though

500/1200 G71
500/1200 G71

the reason you say 1200 is because its alot easier than sayin 600x2

if u want my opinion.. the tom hardware site itself is a joke. Tom's Hardware is a sell out. The site has just recently been sold to some guy who doesnt know anything about pc's. Their review charts i would say are twisted a bit. I think they are payed off to bump a couple cpu or gpu's up where they dont belong, misleading the consumer. I mean come on! If they are scum enough to put annoying google adds right smack in the middle of a thread post, they will do anything for some extra money $. My advice to everyone is boycott toms hardware site, stick to the forums, and read other articles from other sites. Let them see their website hit counter drop so low it will bring tears to there eyes, i would love thaT!

Quote:


So for the inconvienience to others... its 2 x 512mb and 2 x 600 which makes an effective 1 GB & 1200MHz :?


no your wrong.. i got a dual core running at 2.33ghz does that mean i got 4.66 ghz?> nooo its doesnt!! so GEEEZZZ he does have a huge point, that is why the performance is so low ong the 7950gx2.. you notice its 600mhz x 2!! if it was just a single 1200mhz it would surly out perform the 600 x 2 since most games arent made for a dual gpu which i guess would need some type of multithreaded code

Quote:
if u want my opinion.. the tom hardware site itself is a joke. Tom's Hardware is a sell out. The site has just recently been sold to some guy who doesnt know anything about pc's. Their review charts i would say are twisted a bit. I think they are payed off to bump a couple cpu or gpu's up where they dont belong, misleading the consumer. I mean come on! If they are scum enough to put annoying google adds right smack in the middle of a thread post, they will do anything for some extra money $. My advice to everyone is boycott toms hardware site, stick to the forums, and read other articles from other sites. Let them see their website hit counter drop so low it will bring tears to there eyes, i would love thaT!


I dont i stick with Guru3d and fireing squad.

Quote:
Not really. Unless I'm terribly mistaken it doesn't make an effective 1GHz at all. It's just two 500MHz cores in SLi, and everyone knows SLi does not double performance.


I have 2 games on hand that SLI more than doubles the performance on. Even though this is rare there are some games that scale extremely well.


I am going to work right now and when I get back tonight around 6:30pm I will be glad to show you if you want some pics. :wink:

Really? Wow, I don't see how that's even possible (edit: outside of driver shenanigans), but I know better than to dismiss something without thoroughly investigating. I'd be quite interested.

The point was, though, two 500MHz cores in SLi is not the same as a 1GHz core. IMO it'd be somewhat closer to describe it as a 500MHz core with twice the pipelines etc but that isn't equivalent either.

I once considered this card - the 7950GX2... I read the reviews of it... pitted against other Nvidia cards and the X1900XTX... I saw as it outperformed the X1900XTX with its 650 Clk... I in my simple understanding of mathmatics deduced that cutting a workload in half and spread to 2 cores yeilds outputs (EFFECTIVELY) twice as fast - which would have seemed like what had happened.

I may be way off here... and at this point I don't give a shit.

Enjoy your day :) 

in a perfect world it would give it double the preformace.

But the data is transmitted between the cores, loosing some due to distance, then one core has to process the data, so agian loosing some preformace, thats why its not double.

Well it looks as though the updated VGA charts are a little more accurate, the 7950GX2 is way up the top in all benchmarks (that I've looked at).

I'd like to think I had a little something to do with it! 8)
Register or log in to remove
 
Ask the community
Read discussions in other Graphics & Displays categories
!