I was going to buy a 550ti video card to replace my old 8800Gt which is broken. I was trying to compare and see if the new card would be better in general. I found this comparison that says "The GeForce 8800 GT 512MB will be a bit (more or less 17%) better at anisotropic filtering than the GeForce GTX 550 Ti." How can a 5 year old card be better than a card released recently? Is this texel rate and filtering something I would notice in videos or games? Just curious as to what it really means.
Thanks
http://www.hwcompare.com/9815/geforce-88…|||You've just provided a perfect example of why hwcompare.com isn't worth bothering with.
Many people don't realize that the results from HWcompare don't reflect any real-world testing of graphics cards. Their numbers are purely THEORETICAL, it says so right on the page!!!
There are many different specs for graphics cards- the memory interface size, the type of memory, the core and memory clock speeds, the number of stream processing units- but all of those specs have less impact upon performance than the specific GPU a card is based upon! It's impossible to accurately evaluate cards by just looking at the numerical specs, because the GPU models aren't numerical values. Same thing with CPUs- a 2.4Ghz Core 2 Duo E4600 wipes the floor with a single-core 3.6Ghz Pentium 4- the Core 2 Duo internal architecture is much more advanced and efficient, picking the older CPU because it's running at 3.6Ghz would be silly.
Here's the direct comparison:
http://www.anandtech.com/bench/Product/6…
With graphics card numerical specs, sometimes a higher value in one area is offset by a lower value in another. For example the GeForce 9800GT is a 256-bit GDDR3 card. Meanwhile the Radeon HD 4770 is a 128-bit card. You'd think with half the memory bandwidth it would be slower, but it's faster. That's because the 4770 uses GDDR5 memory, and it turns out that a smaller pipeline to faster GDDR5 memory worked just as well as the bigger pipeline to slower GDDR3 memory. So the 4770 beats the 9800GT but it doesn't beat the GTS 250 (also a 256-bit GDDR3 card) because that one has a more powerful GPU.
What HWCompare does is like comparing automobiles based upon manufacturer's brochures, without ever taking them out on the track! They're just guessing based upon differences in particular technical subsystems, often without accounting for technological advances over different generations. A complete waste of time.
The results from videocardbenchmark.net are somewhat better- that shows how various cards did on benchmark tests. Now those results are sometimes misleading because the tests often favor newer generation cards over older ones. Passmark scores show you which cards ran the test suite software best, not which cards actually performed best in game titles like Crysis, Call of Duty, Battlefield Bad Company 2. What really counts is which cards get higher fps in Age of Conan running at 1600x900 or 1280x1024 at various detail settings, not which cards were fastest at drawing the same polygon over & over in test A, or had the highest memory throughput in test B. But even those results are better than HWcompare, because you're at least seeing some actual results, even if the tests aren't the ones you're interested in.
The best source for performance data is card reviews. Check sites like Anandtech, Hot Hardware, Tom's Hardware, Techspot, Guru3D... those show how cards perform at different resolutions and detail settings in actual game titles. They also typically compared that performance to competing cards in the same price bracket, or earlier models. They also provide power consumption information and some context of why different models were introduced.
http://www.techspot.com/review/240-ati-r…
For reference, you can see the GeForce 9800GT (the updated 8800GT) in the chart above. It's slightly better than the Radeon HD 5670. The very top card on that chart (Radeon HD 5750) is approximately equal to the GTS 450. That's below the GeForce GTX 550 Ti, which competes with the Radeon HD 5770.
http://www.techpowerup.com/reviews/Zotac…
http://www.tomshardware.com/reviews/rade…
http://www.techspot.com/review/359-nvidi…
Always look for real reviews- those tell the story.|||The 550 TI was meant to be a new but low performance video card for its times, the 8800 GT is still a very good graphics card if you compare it to current cards.
In the end though, 550 TI is a bad video card, your better off not upgrading or get a 460 gtx which should be the same money as a 550 TI but better performance.|||That's a load of crap. The GTX 550 Ti will stomp an 8800GT in every possible way.
Sheff is an idiot. The 8800GT gets outrun by a Radeon 6670. The GTX 550 Ti easily spanks them both. The 8800GT is a fairly low end card these days.
C-Man nailed it, albeit a bit wordy. Comparing raw hardware specs on video cards means NOTHING. Benchmarks are everything. If you want to know how a video card actually performs, look at tests that show how it actually performs.
No comments:
Post a Comment