So you don't think there has been demand for $150-250 cards since march?
There most certainly has been demand for it. But when you have A) obligations to OEM's and partners to supply X number of chips in their notebooks, you need to prioritize GK107. And when you have a $500 video card that sells out nonstop for the first 3 months, you need to prioritze for that video card. Nvidia has been producing the chips that they contractually obligated themselves to (GK107) and the chips that make the most money (GK104). It isn't until a month ago (or so) that there were enough wafers in production with respect to demand for what is in the supply line to get out any other chip.
GK106 should sell more chips than GK104 (unit demand will be greater at <$250), but when the starting price for GK106 based cards will be at $250 a card tops (it may very well end up less), and moving down from there, and when coupled with the same amount of vram and a pcb similar to the GTX670 with a die size only 2/3 the size of GK104, Nvidia is making more per wafer selling GK104 than they would be selling GK106.
Last edited by a moderator:
The MSI card isn't OC'd, as someone on another pointed out from their GTX660 Ti naming, the reference version is 2GD5 and OC version is 2GD5/OC
Nah, their 660GTX Ti's are named like that too
http://www.msi.com/product/vga/N660Ti-PE-2GD5.html
http://www.msi.com/product/vga/N660Ti-PE-2GD5-OC.html
edit:
Or ye, the cooler on Ti isn't reference, but clocks are despite "PE" name
Wow, in fact i understand the confusion, i was read some 660TI review and they was just indicate MSI Power Edition. ( and tested the card who run at 1097mhz turbo boost ( well even more if we follow the review ).
A1xLLcqAgt0qc2RyMz0y
Veteran
Tweaktown jumps the gun again
NVIDIA GEFORCE GTX 660 2GB Reference Video Card Review
http://www.tweaktown.com/reviews/49...0_2gb_reference_video_card_review/index1.html
NVIDIA GEFORCE GTX 660 2GB Reference Video Card Review
http://www.tweaktown.com/reviews/49...0_2gb_reference_video_card_review/index1.html
Eh... Tweaktown. No mention of what drivers or system they are using at all this time and again, horrible games lineup.
EDIT: Looking at the Dirt 3 results, they are still using Cat 12.3 for all radeons with the exception of the 7970 GHz edition. This time they haven't even mentioned it. What scumbags...
EDIT: Looking at the Dirt 3 results, they are still using Cat 12.3 for all radeons with the exception of the 7970 GHz edition. This time they haven't even mentioned it. What scumbags...
I dont understand they do 10 pages of games tests without AA or AF, and then 1 page of test with AA and AF ?
Many results are missing on the benchs, you dont know why...
I will finish by ask me if someone hack their site for put strange review online..
Many results are missing on the benchs, you dont know why...
I will finish by ask me if someone hack their site for put strange review online..
Last edited by a moderator:
A1xLLcqAgt0qc2RyMz0y
Veteran
I agree, TT's reviews are pretty much useless.
Also, will any vendor really release a GTX660 at reference clocks?
I suspect that they will all be factory overclocked just like the GTX660ti
Just a random sample: GTX 660Ti Unigine vs GTX 670
Tweaktown early review, which was called useless by the same actors: +22.9%,
Anandtech review: +22.5%
Why it should be different for the GTX 660?
There are obvious flaws in the TT review, but I fail to see how it is pretty much useless. Their results should be a good indication of how the GTX 660 compares to other Kepler based cards.
Real numbers are terribly inconvenient, aren't they?
Tweaktown early review, which was called useless by the same actors: +22.9%,
Anandtech review: +22.5%
Why it should be different for the GTX 660?
There are obvious flaws in the TT review, but I fail to see how it is pretty much useless. Their results should be a good indication of how the GTX 660 compares to other Kepler based cards.
Real numbers are terribly inconvenient, aren't they?
I always skip over those almost useless benchmarks like 3DMark or Unigine Heaven. Real games in realistic settings are the real deal. And there you can already see inconsistencies in TT own numbers (as the HD7970GE sometimes show much more than clock speed proportional gains). Comparing to numbers gathered with a wild mix of driver versions is far from perfect.
add resolutions like 1920x1080 and if the benchmarks set up 2 identical machines with different videocards, I would say none would be able to tell difference between 3 different pricegroups of cards.
If the gamer cant notice difference when gaming, does the "new" card has any value then?
I see a lot of people upgrading from for example 5850 or 6870 or 480 or 460 or such and they are dissapointed due to the benchmarks people presents dosnt add value in actual practical use.
I waited from my 6850 set up to this die shrink to get one single card a 7970 and I waited until price went way down, so half the price vs release and I got tremendous value and now are waiting for next die shrink unless amd pull a rabbit out of the hat with SI.
I'm talking Kepler to Kepler comparison for a reason. Use the TT numbers as a derating factor, then apply it to whatever review site you feel best supports your biases (let's be honest: that's how it is for most people). End up with a score that's going to be pretty close to reality.
Assuming they used the same drivers for all Kepler cards, that would work, but even this is not clear.
iMacmatician
Regular
From VideoCardz: "NVIDIA To Release Four GK106 Graphics Cards."
EDIT: And more: "NVIDIA Kepler GK106 GPU Detailed."
5 SMXes, 3 GPCs (2 + 2 + 1).
EDIT: And more: "NVIDIA Kepler GK106 GPU Detailed."
5 SMXes, 3 GPCs (2 + 2 + 1).
Last edited by a moderator:
