Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Difference between GPU's?
#1
Alright, so I'm starting to get awfully curious about the difference between hardware. What I mainly want to know if a couple of the main differences between my current ASUS EAH6670/DIS/1GD5 Radeon HD 6670
Link: http://www.newegg.com/Product/Product.as...6814121442

and the new one I'm getting: SAPPHIRE 100355-1GOCL Radeon HD 7850
Link: http://www.newegg.com/Product/Product.as...6814202004

Now, I notice that they're both 1 GB of DDR5 Ram, but the 6670 is only 128-bit, and the 7850 is 256-bit. What's the difference between those, and whats the benefit/downside to having the 256-bit?

Also, the 6670 has 480 stream processors and the 7850 has 1024. What does having more do, and what exactly are "Stream processors"?

Last but not least, the 6670 is PCI Express 2.1 x16 whereas the 7850 is PCI Express 3.0 x16. what do those mean, and what's the benefit/downfall of the 3.0?
Reply

Sponsored links

#2
http://www.hwcompare.com/12040/radeon-hd...n-hd-7850/
You can use this site to get an idea how fast the 2 cards are.
All the technical terms translate into the speed you can get in games.
Reply
#3
http://www.bit-tech.net/hardware/2012/03...7850-2gb/3

The above real benchmark on actual games help giving an idea... the HD 6670 does not figure in it just because ir is under the category to even try to run those games under those quality level.

PS: For the difference between the two cards performance will not be easy to get a direct comparative test between them, honestly would be unfair and senseless. It situates somewhere between the Nvidia GT 440 and GT 450, what means it would struggle to play most games at 1080p but can do at low presets.

The first digit on the card (both Nvidia and AMD/ATI) indicates the generation and not clearly indicates the relative performance, the second digit is very important, it indicates the card category and has a major impact. The third digit in Nvidia was dumped and substituted by indicators, when exist (for example LE indicates a lesser card while Ti is a better card than the normal one), So x50-x70 indicates mainstream cards, the x50 as mainstream- and x70 mainstream+, the x80 at the top and x90 indicates a dual card (SLI). bellow x50 the card is considered entry level.

At AMD/ATI the codification is a bit more complex for the 3rd and 4th digits are confuse and I could never understand them well, what I know for sure is the higher the best.\

http://www.tomshardware.com/reviews/gefo...214-3.html here is a comparative benchmark
Imagination is where we are truly real
Reply
#4
(02-18-2013, 03:29 AM)xenoriddley Wrote: Alright, so I'm starting to get awfully curious about the difference between hardware. What I mainly want to know if a couple of the main differences between my current ASUS EAH6670/DIS/1GD5 Radeon HD 6670
Link: http://www.newegg.com/Product/Product.as...6814121442

and the new one I'm getting: SAPPHIRE 100355-1GOCL Radeon HD 7850
Link: http://www.newegg.com/Product/Product.as...6814202004

Now, I notice that they're both 1 GB of DDR5 Ram, but the 6670 is only 128-bit, and the 7850 is 256-bit. What's the difference between those, and whats the benefit/downside to having the 256-bit?

Also, the 6670 has 480 stream processors and the 7850 has 1024. What does having more do, and what exactly are "Stream processors"?

Last but not least, the 6670 is PCI Express 2.1 x16 whereas the 7850 is PCI Express 3.0 x16. what do those mean, and what's the benefit/downfall of the 3.0?

The number of bit (the bus width) of the memory is related to the actual bandwith between GPU and Memory:
Bandwith (in GB/s)= Bus width * Memory Clock (in GHz) /8
Generally a more powerful GPU requires an higher bandwith (makes faster calculation so needs to exchange data with the memories more often and at an higher speed).

Stream processors are the "cores" of the GPU, are the units that make the calculation, so an higher number means that is more powerful (if the frequency is the same, higher frequency means faster calculation if the number of stream processor is the same).

PCI-Ex 3.0 is 2 times faster than PCI-Ex 2.0 and is related to the speed at which the GPU can exchange data with the CPU and the system memory, obviously higher is better but the PCI-Ex 2.0 is enough fast even for the fastest GPU you can buy at the moment, so there's no real advantage in the 3.0, anyway to use the higher speed of the 3.0 you need a compatible motherboard and a compatible CPU, otherwise it will work at 2.0. I don't really know what's the difference between 2.0 and 2.1, but probably nothing interesting.

(02-18-2013, 06:22 AM)nosisab Ken Keleh Wrote: http://www.bit-tech.net/hardware/2012/03...7850-2gb/3

The above real benchmark on actual games help giving an idea... the HD 6670 does not figure in it just because ir is under the category to even try to run those games under those quality level.

PS: For the difference between the two cards performance will not be easy to get a direct comparative test between them, honestly would be unfair and senseless. It situates somewhere between the Nvidia GT 440 and GT 450, what means it would struggle to play most games at 1080p but can do at low presets.

The first digit on the card (both Nvidia and AMD/ATI) indicates the generation and not clearly indicates the relative performance, the second digit is very important, it indicates the card category and has a major impact. The third digit in Nvidia was dumped and substituted by indicators, when exist (for example LE indicates a lesser card while Ti is a better card than the normal one), So x50-x70 indicates mainstream cards, the x50 as mainstream- and x70 mainstream+, the x80 at the top and x90 indicates a dual card (SLI). bellow x50 the card is considered entry level.

At AMD/ATI the codification is a bit more complex for the 3rd and 4th digits are confuse and I could never understand them well, what I know for sure is the higher the best.\

http://www.tomshardware.com/reviews/gefo...214-3.html here is a comparative benchmark

AMD codification is extremely simple, higher number = higher performances. Generally if two GPUs share the first two digits means that they are based on the same chip and probably are very similar in therms of amount of RAM and bandwith, the one with the higher third digit has slighltly more stream processor or an higher frequency. The fourth digit is always zero. The "GHz Edition" that you can find on some of the latest GPUs means just that the GPU is clocked at 1GHz (or more if there's OC or turbo).
FX-6300@4.4GHz | Noctua NH-U9B SE2| Asrock 990FX Extreme3| Sapphire HD7850 2GB| 2*4GB AMD Performance Memory 1600MHz Cas8| Intel X-18M 80GB G1| Seagate Barracuda 7200.12 500GB| WD Caviar Green 2TB| Samsung S24B300 (LG Flatron L1720P)| Antec Three Hundred| Gigabyte GT-E8000| OCZ StealthXStream 2 600W| Gigabyte GM-M6800
Reply




Users browsing this thread: 1 Guest(s)