facebook rss twitter

EVGA doubles-up on GTX 570 memory

by Navin Maini on 21 June 2011, 13:26

Tags: EVGA

Quick Link: HEXUS.net/qa6f7

Add to My Vault: x

Please log in to view Printer Friendly Layout

Boosting memory capacity is one way of differentiating your wares, and EVGA has looked to the GTX 570 to do just that.

 

 

Rather than the 1,280MB GDDR5 you'd find on a vanilla GTX 570, EVGA's new outing debuts with a total of 2,560MB - running at 950MHz (3,800MHZ effective). The GPU is clocked at 732MHz, with the shaders running at 1,464MHz. There clearly aren't any factory-fresh overclocks part of this equation.

It's also said that the PCB is an in-house design from EVGA - shorter than the reference offering from NVIDIA. The company's online store has the new GTX 570 flavour priced at just shy of $400 USD.

HEXUS Forums :: 9 Comments

Login with Forum Account

Don't have an account? Register today!
Posted by aidanjt - Tue 21 Jun 2011 13:52
Simple question…. Why?
Posted by watercooled - Tue 21 Jun 2011 14:10
Might make a difference at very high resolutions but I'd say it's probably just an e-peen thing. If this card is smaller, combined with lower power draw/noise it might be a good buy though, if the price is right.
Posted by brasco - Tue 21 Jun 2011 14:13
aidanjt
Simple question…. Why?

Undoubtedly it's just a bigger number for the less tech savvy to sell more units :rolleyes:

Having said that I regularly run out of vram when using CUDA, but that's surely too small a minority to worry about.

brasc
Posted by krazy_olie - Tue 21 Jun 2011 14:18
useful for gpgpu cuda/opencl etc

high resolution/texture quality + vsync etc.

Depends if thee is enough bandwidth to utilise it though.

Probably more of a future proofing thing really
Posted by aidanjt - Tue 21 Jun 2011 14:19
brasco
Undoubtedly it's just a bigger number for the less tech savvy to sell more units :rolleyes:
Indeed.

brasco
Having said that I regularly run out of vram when using CUDA, but that's surely too small a minority to worry about.
Yeah, at least that problem can be rectified by memory management. And nV should make compute-specific cards easier to get a hold of for hardcore CUDA users.

SEE NEWER »