Gpu 102

J

joeswm8

New Member
i just read the GPU 101 or whatever, and I understand that the more pixel pipelines, the better performance, but how do you ultimately determine the true performance of a card?

like is it the pixel pipelines x core clock = performance level ?

or pixel pipelines x memory clock = performance level ?

and how exactly does the RAM play into the equation?



i now have three examples:

and also, I am now looking at three cards (no DX10 yet):

XFX 7900GS Extreme
http://www.newegg.com/Product/Product.asp?Item=N82E16814150202
20 pixel pipelines / 525 core clock stock / 1550 memory clock stock / 256mb Gddr3 RAM

XFX 7900GT
http://www.newegg.com/Product/Product.asp?Item=N82E16814150216
24 pixel pipelines / 470 core clock stock / 1370 memory clock stock / 256mb Gddr3 RAM

XFX 7950GT Extreme
http://www.newegg.com/Product/Product.asp?Item=N82E16814150189
24 pixel pipelines / 570 core clock stock / 1460 memory clock stock / 512mb Gddr3 RAM


how would I evaluate the performance level of these three cards?
and wouldnt the 7900GT essentially equal the performance of the 7950GT?

thanks a lot
 
^^^^^^ Wrong wrong wrong. Their info is completely wrong. Most of their stuff is nowhere near accurate!
 
ADE said:
^^^^^^ Wrong wrong wrong. Their info is completely wrong. Most of their stuff is nowhere near accurate!
Click to expand...

well.. if youre so awesomely brilliant and know that.. could you enlighten us and tell us whats the best way then?

i'd say gpureview (wich is the theoretical performance) and benchmarks
 
ADE said:
^^^^^^ Wrong wrong wrong. Their info is completely wrong. Most of their stuff is nowhere near accurate!
Click to expand...
Completely wrong my buttocks. Why don't you tell us what every single specification of all the cards are, then?
 
Gee I would be happy to but I'm saying they screwed up a lot of info for me before. I would Love to say but sadly I can't seem to trust too much of their info.
 
They had info for one card and put it for a completely different card, When doing an RMA I looked it up and saw that my card info was wrong and tried to send it back and got in a huge mess over GPUreviews wrong info, and well, there are a few more but I think my point is said.
 
lol i just look at benchmarks.
http://www23.tomshardware.com/graphics.html
just choose the two cards you're debating... and see which ones better for the game you want to play.

generally you get what you pay for as long as you go to newegg to get it. ive even seen some really good deals on cards at best buy... i mean $199 for a 7900gs is a really good deal... its still cheaper online though.

as you can tell, gpu's seem like theyre much more complicated than cpu's... well im here to tell you they arent, they just change a whole lot more often than cpu's do... aside from dual core lately.

just understand one thing: directX 7,8,9 cards are a WHOLE lot different than directX 10 cards. a WHOLE lot. they changed the architecture, and nvidia even has a seperate driver download for the 8800 series... they need different drivers, therefore the architecture is completely different. nvidia's dx10 solution uses stream processors... 128 of them to be specific... or 96 in omega's case. it outperforms any other card on the market by a LONG shot. the 8800gtx is more than twice as powerful as the previous most powerful gpu on the market, the 7900gtx... the 7950gx2 doesnt count because it wasn't a single gpu :D

id get an 8800 but frankly... i dont need it right now for the games i play... my 7800gt plays them all on highest settings without a hitch. aside from some AA, which my old 7900gtx (yes i sold it) couldnt do with HDR anyways... cause nvidia sucks... which is why im probably going to choose ATI this time... if not for reliability and longevity, but for their avivo video converter! lol plus i have a tv tuner card made by ATI and im very satisfied with it.
 
Last edited:
Well the best think to tell you are that benchmarks are benchmarks. If you want the truth behind a graphics card you need real time real world experience, not a graph. If you want benchmarks look into 3D Mark 06. Its a benchmark for gamers. But overall you can't trust web sites on benchmarks cuz they all say different things. Generally 2 is better than 1, heir number = better graphics card --That only in general, not all the time!-- but it comes down to this--BENCHMARKS ARE NOTHING CUZ IT WONT TELL YOU REAL WORLD RESOLTS-- That's about all I can say.
 
Games are, but that in the form of a website benchmark doesn't truly reflect what YOU will be able to do overall when YOU yourself play it . I'm just saying they can be very misleading in many ways.
 
maroon1 said:
gpureview.com is good website to compare cards a stock speed.
Click to expand...

yeah but it doesnt answer my questions, and I really want them answered so I can learn and not just look at benchmarks all the time

can someone please directly answer the pixel pipeline, memory, core questions please?

maroon1 said:
But but I'm very sure that 7900GS Extreme will perform similar to 7900GT at stock speed
Click to expand...

the 7900GT has 4 more pixel pipelines, so it will easily outperform the 7900GS, right?
 
how do you ultimately determine the true performance of a card?

like is it the pixel pipelines x core clock = performance level ?
or pixel pipelines x memory clock = performance level ?
and how exactly does the RAM play into the equation?

i have three examples:

XFX 7900GS Extreme
20 pixel pipelines / 525 core clock stock / 1550 memory clock stock / 256mb Gddr3 RAM

XFX 7900GT
24 pixel pipelines / 470 core clock stock / 1370 memory clock stock / 256mb Gddr3 RAM

XFX 7950GT Extreme
24 pixel pipelines / 570 core clock stock / 1460 memory clock stock / 512mb Gddr3 RAM

how do you determine the performance of these three cards? by looking at what, in what order?

and wouldnt the 7900GT=7950GT because of same pixel pipelines, and similar core and memory?
 
You determine it by experience. You also must take into account overclocking. That just makes thing more difficult, but I see what your saying. This is a though question to answer. I say you determine it by how well it plays games and to do that you look at the reviews. There really isnt any mathematical way to find out this stuff, unless your a computer genius or something. Look at the cards and then look at their score on 3D Mark 06 and that's the best way to see real results since the 3D Mark 06 uses the same thing for every card it tests there for seeing which card does the best in the test.
 
You must log in or register to reply here.
Back
Top