New GPU - Ever see such terrible soldering? HIS IceQ 7950

[Puffnstuff]

If you think the solder job is bad just wait until you need warranty service from his who have contracted it out to a 3rd party outfit in CA. They rejected my card for warranty because a circuit had failed...gee why did I send it in to them in the first place.

 

[Abwx]

Seems that there s a lot of solder flux that remained on the PCB ,
in principle it can be cleaned using a teethbrush wetted with
trichlorethylene , wich what is used generaly to dissolve said matter.

Generaly too much solder flux can stick to a component
connection and act as in insulator between said pins and
the solder , resulting in cold solder with no electric contact.
In such cases one has to use a soldering iron and solder again
pins that look apparently soldered but are insulated and also
put solder on traces that were also flux covered , as seems the
case with this card.

 

[KingFatty]

My 3L I went 1150/1600 in UH 4.0 without any artifacts but that was high as I was willing to try due to no vrm temp sensors. To me imo this not a loud card either which does matter to me to a degree.

UH 4.0
With stock fan settings temps maxed (1150/1600) at 69C.
With fan profile it was like either 62 or 64C max at 1150/1600.
http://forums.anandtech.com/showthread.php?t=2341362

Do you recall what voltage you needed to get up to 1150?

My card seems to be limited to 1100 at 1.25v because artifacts appear unless the fan is cranked to full. But were you using like 1.3v to reach 1150?

I think the IceQ cooler is fantastic, with great VRM cooling and a reinforcing plate etc. The card is nicely built. But all this fancy cooling goes to waste if I can't even beat 1100 - I think another 7950 with crappy cooling, and not even having any VRM cooling, could beat 1100? Maybe I'm doing something wrong with the overclocking. My technique was to open the catalyst control panel, and go under the section for overclocking, and click on the agreement to open the options. Then I set the power slider to +20%. Then close catalyze control center. Open MSI afterburner, confirm that the power slider is set to +20%. Then adjust the voltage to 1.25, and increase the frequency to 1100, apply.

 

[Fastx]

1100/1600 I am at 1.18v at 1150/1600 I am at 1.21v or 1.22v.


Misc
I never used CCC OD for overclocking and left it disabled. I also read awhile to leave it disabled when oc'ing with Trixx fwiw.

I set PB to 20+ of course.
I have UPLS disabled.
Stock/auto fan settings per above oc's running UH 4.0

What are your temps at 1100 1.25v under load at stock/auto fan settings and also your VRM temps if you have VRM Temps sensors?

I never used MSI AB for oc'ing (only Trixx) but only just for monitoring in game OSD for GPU temps, GPU core clocks/usage, FPS.

IMO I will not go over 1.25v on oc's.

KF imo I would return that card for a refund.

 

[Subyman]

How do you think they lowered the price on these cards? Looks like the quality of a cheap fan controller.

 

[KingFatty]

Is there any chance I would get artifacts by leaving the memory clocks at stock, and raising *only* the core?

My approach in testing overclock was to raise core and voltage first, leaving memory speed stock. But I assumed the resulting artifacts were because the core couldn't handle it, so I didn't even bother increasing the memory speeds.

Now I'm wondering, is it possible that the artifacts would go away if I overclocked *both* the core and memory speeds, sort of thinking that they would better match each other? Instead of just increasing the core speed and leaving memory stock? I would think no way, but I just want to double-check before returning the card.