Moratorius
n00b
- Joined
- May 16, 2010
- Messages
- 12
I'm usually updating the thread at overclock.net first; I posted in these two forums since I've found both forums to be rather helpful during my lurking in the past. I was just curious what people thought I should do... it seems to be a pretty big deal as the 465 is an unreleased card, but I'm not too keen on doing anything that'll void my warranty. I'll fulfill any requests for information that won't void my ability to return the card, but that's the extent of it.
Photos hosted by chatch15117, thanks!
--------------------
5/24/2010:
--------------------
Now I'm just waiting for the Galaxy rep to process a RMA (and play some Starcraft 2 on ultra while I still have the card, heh). Here are the 3DMark Vantage benchmarks for the card in 465/470 modes using the beta 257.15 drivers.
465:
470:
Decent gains
--------------------
5/22/2010:
--------------------
I've attempted to backup and compare the two bios, but it seems I'm only able to download the v70.00.21.00.03 bios (it's the attached bios I saved from GPU-Z in the very beginning of this thread, corresponding to the 470 specifications):
Interestingly, running pciblocks on the saved bios image right after I saved the bios (still running w/ 465 specs according to the bios screen when I had booted up) results in a gpu mismatch:
So it seems the hardware id alternates between the 465 and 470, but both GPU-Z and nvflash can only save the 470 bios despite identifying that the bios isn't the one in use...
--------------------
5/21/2010:
--------------------
PCB Backside:
I turned on my computer this afternoon and upon booting into Windows, I was confronted with the default low resolution that accompanies unloaded drivers. Windows forced me to wait for it to download its own drivers:
Something caught my eye as I rebooted (my phone's camera is terrible, but convenient):
460?
After the reboot, I found that Windows had installed the 197.41 drivers. So I decided I'd check all the information programs again and run another 3dmark vantage.
No real changes (though GPU-Z 0.4.3 has more information now)... so I proceeded to update my drivers again... but I was forced to downgrade back to 197.41 by Windows again (Device Manager told me it was a Code 1 error).
Again, something caught my eye as I rebooted:
470 now? My card is Jekyll & Hyde! The information programs all agreed that it was an actual 470 this time though.
The 3dmark vantage score also seems to be normal for a stock 470.
I've rebooted a number of times since, and the card alternates between 465 and 470 specs. What could be causing this?
--------------------
Original Post (5/20/2010):
--------------------
My Galaxy GTX 470 GC arrived today.

I had a bit of difficulty installing the NVIDIA drivers, because the card couldn't be recognized by the installer. I had to resort to changing the PCI device id of the GTX 470 line in the NV_DISP.INF to get the drivers to install.
However, the card seems to be identified as a GTX 465 with matching hardware details (e.g. 1GB memory, lowered clocks).
Is it possible that the factory mis-flashed the card with a GTX 465 bios? Or have I really received a GTX 465 in the GTX 470's cooling shroud?
Benchmarks (197.75):
Photos hosted by chatch15117, thanks!
--------------------
5/24/2010:
--------------------
Now I'm just waiting for the Galaxy rep to process a RMA (and play some Starcraft 2 on ultra while I still have the card, heh). Here are the 3DMark Vantage benchmarks for the card in 465/470 modes using the beta 257.15 drivers.
465:
470:
Decent gains
--------------------
5/22/2010:
--------------------
I've attempted to backup and compare the two bios, but it seems I'm only able to download the v70.00.21.00.03 bios (it's the attached bios I saved from GPU-Z in the very beginning of this thread, corresponding to the 470 specifications):
Interestingly, running pciblocks on the saved bios image right after I saved the bios (still running w/ 465 specs according to the bios screen when I had booted up) results in a gpu mismatch:
So it seems the hardware id alternates between the 465 and 470, but both GPU-Z and nvflash can only save the 470 bios despite identifying that the bios isn't the one in use...
--------------------
5/21/2010:
--------------------
PCB Backside:
I turned on my computer this afternoon and upon booting into Windows, I was confronted with the default low resolution that accompanies unloaded drivers. Windows forced me to wait for it to download its own drivers:
Something caught my eye as I rebooted (my phone's camera is terrible, but convenient):
460?
After the reboot, I found that Windows had installed the 197.41 drivers. So I decided I'd check all the information programs again and run another 3dmark vantage.
No real changes (though GPU-Z 0.4.3 has more information now)... so I proceeded to update my drivers again... but I was forced to downgrade back to 197.41 by Windows again (Device Manager told me it was a Code 1 error).
Again, something caught my eye as I rebooted:
470 now? My card is Jekyll & Hyde! The information programs all agreed that it was an actual 470 this time though.
The 3dmark vantage score also seems to be normal for a stock 470.
I've rebooted a number of times since, and the card alternates between 465 and 470 specs. What could be causing this?
--------------------
Original Post (5/20/2010):
--------------------
My Galaxy GTX 470 GC arrived today.

I had a bit of difficulty installing the NVIDIA drivers, because the card couldn't be recognized by the installer. I had to resort to changing the PCI device id of the GTX 470 line in the NV_DISP.INF to get the drivers to install.
Code:
%NVIDIA_DEV.06C4.01% = Section012, PCI\VEN_10DE&DEV_06C4
%NVIDIA_DEV.06C4.01% = Section011, PCI\VEN_10DE&DEV_06C4
NVIDIA_DEV.06C4.01 = "NVIDIA GeForce GTX 470"
However, the card seems to be identified as a GTX 465 with matching hardware details (e.g. 1GB memory, lowered clocks).
Is it possible that the factory mis-flashed the card with a GTX 465 bios? Or have I really received a GTX 465 in the GTX 470's cooling shroud?
Benchmarks (197.75):
Last edited: