View Full Version : Fermi In Action
Ancient76
12-11-09, 02:44 PM
http://www.tomshardware.co.uk/forum/page-276100_15_0.html
http://img16.imageshack.us/img16/8425/residentevil5gtx300n.png
http://img705.imageshack.us/img705/5503/stalkerclearskygtx300n.png
http://img705.imageshack.us/img705/2432/farcry2gtx300n.png
http://img686.imageshack.us/img686/8986/crysisgtx300.png
Specifications
GTX 380
CUDA cores 512
core clock 730MHz
shader clock 1600MHz
Memory 1100MHz 2GB
TDP(watts) 225w
GTX 360
CUDA cores 480
core clock 650MHz
shader clock 1400MHz
Memory 1000MHz 1.8GB
TDP(watts) 195w
Opinions :)
JasonPC
12-11-09, 02:52 PM
Faked nice try though.
LydianKnight
12-11-09, 03:00 PM
Veeery fake, I'd say. No matter how good (or bad) NVIDIA's engineering skills are, no way a GTX 360 (if that's the final codename) can beat a 5970, just a crap fake ¬¬
LydianKnight
12-11-09, 03:02 PM
Oh and 480 cores for a GTX 360? the other thread mentions 384, so... bleah, fck' that xD
Ninja Prime
12-11-09, 03:08 PM
480 cores wouldn't even work for that matter. Faked by some NVfanboy that didn't even know wtf he was doing, lol.
XMAN52373
12-11-09, 03:57 PM
480 cores wouldn't even work for that matter. Faked by some NVfanboy that didn't even know wtf he was doing, lol.
Why couldn't 480 cores work? Each shader group has 32 cores in them. 1 disabled shader group would equal 480 total cores. Last time I looked 16x32=512 and 15x32=480.
Ninja Prime
12-11-09, 05:04 PM
Why couldn't 480 cores work? Each shader group has 32 cores in them. 1 disabled shader group would equal 480 total cores. Last time I looked 16x32=512 and 15x32=480.
IIRC they are coupled to ROPS in pairs, so 64 would work, but not 32, might be wrong though.
Other than that, 2GB ram with 384 bit interface doesn't work. NV would never launch a mid range within 8% of the performance of their high end, and the bench numbers don't even match their own spec numbers. ;)
XMAN52373
12-11-09, 05:14 PM
IIRC they are coupled to ROPS in pairs, so 64 would work, but not 32, might be wrong though.
Other than that, 2GB ram with 384 bit interface doesn't work. NV would never launch a mid range within 8% of the performance of their high end, and the bench numbers don't even match their own spec numbers. ;)
Well, if BSoN specs are to believed, it wouldn't be 8% of the performance if they disable 1 ROP group and 1 shader group and use the same clock as reported by Theo at BSoN for a GTX360. Other than that issue, you are correct that 2GB and a 384 bit, let alone 320bit buses don't work.
Also concider the GTX275, same ROPs as the GTX260, same bus aswell, but with GTX285 shader count.
Redeemed
12-11-09, 07:06 PM
I'm calling BS as well. No way that the GTX380 is faster than the 5970.
CaptNKILL
12-11-09, 07:52 PM
The performance of the nvidia cards doesn't look too suspicious to me... its the performance of the 5970. There's no way that its only 20-40% faster than the 5870.
So yeah, these are probably total BS.
Heinz68
12-11-09, 08:19 PM
Opinions :)
Posted by "successful_troll" at Tom's Hardware Forum (http://www.tomshardware.co.uk/forum/page-276100_15_0.html)
This benchmarks is not fake, is from the tomshardware review about the GTX 300 series.
Looking at those benchmarks the GTX 3xx seems to be more powerful than we expected, I am surprised.
Here is the link "successful_troll" gave for the benchmarks in the same forum thread.
http://www.tomshardware.co.uk/nvidia-GTX300,review-31769.html
@Ancient76
I'm not going to bother with my opinion on the benchmarks, but since you posted them, why don't you give us your opinion about them?
|MaguS|
12-11-09, 08:32 PM
Posted by "successful_troll" at Tom's Hardware Forum (http://www.tomshardware.co.uk/forum/page-276100_15_0.html)
Here is the link "successful_troll" gave for the benchmarks in the same forum thread.
http://www.tomshardware.co.uk/nvidia-GTX300,review-31769.html
@Ancient76
I'm not going to bother with my opinion on the benchmarks, but since you posted them, why don't you give us your opinion about them?
That link is broken and when you go to the Toms Hardware main page it does not even mention a review or even anything regarding the GTX300.
That link is broken and when you go to the Toms Hardware main page it does not even mention a review or even anything regarding the GTX300.
Coz there isn't one :)
CaptNKILL
12-11-09, 08:50 PM
If you compare that picture to all of the real ones from Tom's Hardware, there are some slight differences.
The logo is the same size on all of the real benchmarks I looked at, where as its larger on the GT300 one.
Also, all of the benchmark images from Tom's Hardware have a black transparent background on the borders. The fake image uses white. ;)
Besides... why the hell would Tom's make a benchmark chart for a graphics card that won't be released for several months and is still under NDA? Further more, how would it leak?
Heinz68
12-11-09, 11:30 PM
That link is broken and when you go to the Toms Hardware main page it does not even mention a review or even anything regarding the GTX300.
I knew that, I'm still waiting for the Ancient76 opinion, since "Fermi In Action" is his thread.
He asked everybody for an opinion on the benchmarks without giving his.
Doesn't some international version of toms hardware always leak reviews early, like dailytech is to annandtech? The reviews are always crap and you come away with more questions than you had, but I think that's the idea behind these, it's not the official Toms benchmarks.
The nvidia slides could be right, because surely nv cherry pick benchmarks and have probably used old drivers for the 5970. Besides the 5970, I would still expect the 5870 to be a bit closer to the GTX 360 than that.
Atomizer
12-13-09, 04:49 AM
Man I wish I knew the real performance numbers, the wait is killing me!...Oh wait thats my shoe
leekirlew
01-06-10, 03:28 PM
Doesn't some international version of toms hardware always leak reviews early, like dailytech is to annandtech? The reviews are always crap and you come away with more questions than you had, but I think that's the idea behind these, it's not the official Toms benchmarks.
The nvidia slides could be right, because surely nv cherry pick benchmarks and have probably used old drivers for the 5970. Besides the 5970, I would still expect the 5870 to be a bit closer to the GTX 360 than that.
what makes you so sure that 360 is gonna be ahead of the 5870? judging by the performance leap and previous releases both camps have been pretty close. i expect the 5870 to sit between the 360 and 380 until amd "update" their drivers :afro:
jesus thats an old sig =p
methimpikehoses
01-06-10, 03:55 PM
I'll stick with my 295's.
mailman2
01-06-10, 04:16 PM
I'll stick with my 5870 :)
I might get another 5870 :D
I might get another 5870 :D
+1 but i'll only have 2 :o
slaWter
01-07-10, 11:05 AM
I'll get rid of the 5970 and get two Fermis :)
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.