Hello community!


Like the title of this thread.. ;-)

There is the GTX465..
IIRC, GTX460 came little bit later..

What would be the better choice (higher RAC) for S@H CUDA?


Thanks!

---

nVIDIA specs:

'GTX460'

CUDA Cores: 336
Core: 675 MHz
Shader: 1350 MHz
RAM: 1800 MHz


'GTX465'

CUDA Cores: 352
Core: 607 MHz
Shader: 1215 MHz
RAM: 1603 MHz

---

I can't speak for the 465, but I know what my 460 does. I've got it running at 920mhz, 1900mhz memory speed. It knocks out 2 WU's at a time in about 15-18 minutes.

---

The GTX465 and GTX460 1GB performance is very close to each other but the GTX465 does use a lot more power so IMO I'd go with the GTX460 1GB (the GTX460 768MB is slower).

Cheers.

---

I am running a 460-768 now on my AMD Quad

---

The point of view 465's are in fact bios limited 470's

http://forums.overclockers.co.uk/showthread.php?t=18181584

---

As far as I can understand it, the 465 is based on the same core as the 470's and 480's, and therefore suffer the same problems with heat being on the older core design. The 460's are a different revision, and therefore run cooler and have a bit more overclocking potential.

I'm actually looking at a 460 for myself, but still trying to justify the cost of it. I do a bit of gaming as well, but not enough to 'really desperately' need it. The 470's and 480's are still far too expensive for my liking, and I cannot justify the cost on one of those at the moment. Maybe someone else can give a bit more info?

The EVGA (SC version) seems to be a good bet. The version that is on sale on a couple of websites here comes with an unexceptional 2 year warranty, but you can extend it to 10 years when you register (but I don't have any details on that yet.)

regards, Giz.

---

A proud GPU User Server Donor!

This is the 460 I am running:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814121391

Stay fairly cool at about 55*C while crunching 2 WU's at 920mhz and 1.06v core. That's with the fan running around 60% (no louder than my case fans). She idle's around 30-32*C at stock clock, and upper 40's while crunching at those stock speeds. Even OC'd the 36% that it is, it only runs 6-8*C warmer. Only had it about a month now, but so far I am very happy with it.

---

This is the 460 I am running:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814121391

Stay fairly cool at about 55*C while crunching 2 WU's at 920mhz and 1.06v core. That's with the fan running around 60% (no louder than my case fans). She idle's around 30-32*C at stock clock, and upper 40's while crunching at those stock speeds. Even OC'd the 36% that it is, it only runs 6-8*C warmer. Only had it about a month now, but so far I am very happy with it.

Ok, now there is an active link. It's an Asus, Me I like the 1GB MSI Cyclone OC, Which costs about $5 less and I like It cause the fan looks like It's easy to clean.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

There is the GTX465..
IIRC, GTX460 came little bit later..

What would be the better choice (higher RAC) for S@H CUDA?

These threads on BOINCstats might prove helpful:

GeForce GTX 460 1 GB

NVIDIA Launches GeForce GTS 450

---

Something to remember, when I was browsing some results crunched by the 480 or 470, I notice big differences between Standard CUDA App. and the Lunatics Unified Installer, the standard app. appears to be a bit faster on WU's with a ~0.4AR, but I don't know how many WU's are run at a time.

And on lower AR's, ~0.01- 0.1AR, there is little difference, but also hard to find, cause these AR's are normally done by CPU.
So this makes it even more difficult, finding differences.

The SSSE3 optimized app. from Lunatics are 3x as fast, compaired on a Q6600 (stock) CPU, compaired to the Standard App.

Anyway, IMO the NVidia's 400 series, especially the 460, 465, 470 and 480 are all, very good, making much less error's (Find Triplets in a row), -9 (overflow) and a lot of other error's, most important, these are not very sensible to high temps!
These cards give the possebillity the run much more tasks together, then the 200 series, although, from the 260 and up, are still faster, running 1 WU at a time.

Just an observation.......

---

Fred, Jason has said there is little to no improvement in speed for Fermi GPUs in this app but his main concern was all the -9 overflows caused by people trying to use the V12 app and also to help with the -12 and -1 errors. This was the first priority and to this end he has done a fantastic job. Now he is working to find optimization for speed increases. If I have read him correctly, there are a lot of things he needs from Nvidea first to accomplish this though. I'm confident these speed increases will come soon.

---

PROUD MEMBER OF Team Starfire World BOINC