MSI 580 Lightning

[SF3D]

It can "crash" during the load of GT3 and performance is then stock / low power 3D. There is no way to get 700fps at 1650 clock. I am sorry

[teurorist]

but then it also was maybe the OCP ? or ocp is directly shut down?

[cowie]

ah thats stock clocks for gt4 even gt2 is way low for those clocks i get same gt2 at 1150 based on sb @5.4
sucks man you done killed it...

[loopy83]

CPU was on stock for this result... we didnt care about the performance and the card didnt switch to 2D i think, because 1700MHz where nearly stable too.
Maybe it was an temperature issue, but we just saw, that the GPU was able to handle those clocks.
And beware of the memory clocks, I think this was the limiting factor in this run beside the CPU.

However... card is dead

[Kabauterman]

CPU was on stock for this result... we didnt care about the performance and the card didnt switch to 2D i think, because 1700MHz where nearly stable too.
Maybe it was an temperature issue, but we just saw, that the GPU was able to handle those clocks.
And beware of the memory clocks, I think this was the limiting factor in this run beside the CPU.

However... card is dead

to see the scaling... CPU on Stock is the reason for low GT4





we did also a straight 1450 run in 3DMark11 what results in 9700 Points (wothout overlocked GFX Memory) but no screen from this

[T0lsty]

it`s a bullsh(i)t scores in 03.
i can hit 721 fps on stock gpu and cpu clocks

it will scale even on stock 2600k. but not 720 fps

[karandiru]

Bandwidh very low (192gb/s.) my friend,maybe the problem pci-e speed x8

[Kabauterman]

Yes we just saw that we used 8x after the Testing

score is as well but clocks aren't ...

and with Clock Switching in the Tests I think there is no reason for the OCP to come

[G H Z]

maybe the problem pci-e speed x8

For sure that didn't help. Same thing happened with 5870 Lightning when too cold, didn't see that at all on 480 though

[[DRACO]]

Hello

I have the chart that I played in the moa 2011 in Istanbul, and the results he obtained are below the other participants to have this bug in the graph,

And I thought that was trouble for other things, motherboard, drivers, bios ect ...

But that is not asymptomatic and many things have changed and I think this model can accurately esatr engineren defective when a sample,

Been experimenting with a twin 580gtx frozr2 and I found that there are some estimates that lihting 580gtx not do it right, well mine.

I have done tests and found in the 3DMark03 test very low fps GT2 780 to 820 fps instead would have to give much more from 1300 to 1500 fps in default,

What could be the problem? I believe that any calculation engine of the Gpu chiset 580gtx is malfunctioning,

MSI P67 MARSHAL
I7 2600K @ 4700MHZ AIR
580GTX TWIN FROZR2 (DEFAULT)
KINSTON 2100 (DEFAULT)




MSI P67 MARSHAL
I7 2600K @ 4700MHZ AIR
580GTX LIGHTING (DEFAULT)
KINSTON 2100 (DEFAULT)



MSI P67 MARSHAL
I7 2600K @ 4700MHZ AIR
580GTX LIGHTING (DEFAULT)
KINSTON 2100 (DEFAULT)



As you can see sometimes to results as 70% less than normal,

I've tried everything and now I give up and I think that is defective you are saying?

Thanks

[karandiru]

Do you use riva tuner during the test..?

[[DRACO]]

Do you use riva tuner during the test..?

not even I have it installed,

The RivaTuner does not use

Greetings

[G.Foyle]

We've had the same thing in Istanbul. FPS at the start of GT3 were ok, but after the bench the average was <700 fps.

[DStealth]

Hello

I have the chart that I played in the moa 2011 in Istanbul, and the results he obtained are below the other participants to have this bug in the graph,

And I thought that was trouble for other things, motherboard, drivers, bios ect ...

But that is not asymptomatic and many things have changed and I think this model can accurately esatr engineren defective when a sample,

Been experimenting with a twin 580gtx frozr2 and I found that there are some estimates that lihting 580gtx not do it right, well mine.

I have done tests and found in the 3DMark03 test very low fps GT2 780 to 820 fps instead would have to give much more from 1300 to 1500 fps in default,

What could be the problem? I believe that any calculation engine of the Gpu chiset 580gtx is malfunctioning,

MSI P67 MARSHAL
I7 2600K @ 4700MHZ AIR
580GTX TWIN FROZR2 (DEFAULT)
KINSTON 2100 (DEFAULT)

[IMG][/IMG]


MSI P67 MARSHAL
I7 2600K @ 4700MHZ AIR
580GTX LIGHTING (DEFAULT)
KINSTON 2100 (DEFAULT)

[IMG][/IMG]

MSI P67 MARSHAL
I7 2600K @ 4700MHZ AIR
580GTX LIGHTING (DEFAULT)
KINSTON 2100 (DEFAULT)

[IMG][/IMG]

As you can see sometimes to results as 70% less than normal,

I've tried everything and now I give up and I think that is defective you are saying?

Thanks

Yea we have such a problem when 3dm03 arived , solution was very simple - just set desktop res to 1024*768 and then run it, when you have bigger desktop resolution gt2 is bugging

[[DRACO]]

Yea we have such a problem when 3dm03 arived , solution was very simple - just set desktop res to 1024*768 and then run it, when you have bigger desktop resolution gt2 is bugging

hello brother

Probe and today all resolutions and also probe in case 1024 x 768,

But still the same problem, this chart gives a result sometimes low and sometimes much lower,

Greetings

[amlett]

Hi guys. I've been reading you several months and your work is awsome. I've boughtt a 580 lightning and I'd like ask you for some help for pushing it. First of all, I'm not planning use LN2 (my wife would kill me), but I'll put on the card an EK VGA Supreme waterblock with a XSPC RX360 rad just for the card.

I've played with it one hour this morning because I've got not much time these days, and this is the best result I've got on air:

core +200
mem +100
aux +50
fan 100%
no switches up, no unlocked bios (VID for 832mhz 0.997)

3dMark 2011 @ 990/2300: P7087 (6801 GPU score) (i7 3,8ghz HT)



I hope I'll mount the block tomorrow if I can (today is very difficult), set a pair of delta fans blowing to the PCB, and try to see how far can I push it. Any advices?

Can I turn on the voltaje switches safetly on water? what about using the unlocked 1.21v bios too?
I used 1.21v with a 470 and this block reaching 925/2180mhz (53º), but I don't know if with the switches I'll be too much volts for water...

thanks!

[loopy83]

The switches just add 0,04V... so I think its safe for water
But first try afterburner... just to max out the card.
I think these voltage switches are good for 24/7 OC or CB/CBB issues under LN2.

[amlett]

Thanks for the rep.

Finally, aganist my wife's desires, yesterday I was able to mount the EK block. I just could try 1000/2300 with the MSI AB maxed out and the max temp was 53º, which it's very nice. Today after work I'll try were can it goes with the switches turned on.

I have a couple of doubts, the PWM ClockTuner and PLL switches, what are they for? I mean, would I see any improvement turning them on?
And finally, is ther any kind of pencil vmod that I can do? I did the pencil vmod for memory voltage with the 470 and I got better mem clocks, but I don't know if there is a similar "light" vmod for this card. I have no soldering vmods experience and also, I don't know if a more agressive vmod can be done just with water cooling. Where can I find tutorials about vmods for the lightning?

[sorry for my english]

[Kabauterman]

vCPU, vPLL and vMEM Switch just add Voltage. Nothing else. 0,04V in each case.

OCP Switch increase OCP from 500A to 550A (no need for this on Water or Air)

PWM Switch changes the Frequency of the PWM (Voltage Regulation/Mosfets)
with higher Freq you got cleaner Voltage output but your Mosfets get hotter.
And even the Stock Freq of the PWM is a lot more than used on other Cards or Boards.

Extreme Cooling switch is useless on Air/Water

[amlett]

Thanks kabauterman!

I'll try today just with the vCore Vmem and vPLL switches turned on.

[loopy83]

I think you will see no gain from the VPLL switch... but its worth a try...

[Kabauterman]

Thanks kabauterman!

I'll try today just with the vCore Vmem and vPLL switches turned on.

Plase test first with Afterburner, you don't need to switch on PLL and MEM if u don't get any gain in Clocks...

We tested 4 Cards and not even one scales well on vMEM with Air cooling...

Just to switch on the V-Switches is not the right way...

[amlett]

Yep but it seems that my mems won't go over 2300mhz with +100mv, maybe it's its top frecuency on air. ok I'll focus on the vcore switch and the msi AB, at the moment I can reach 1ghz with +200mv, I hope I can push a little more with that 0.04 extra volts...

Thanks guys for the help.

[G.Foyle]

And even the Stock Freq of the PWM is a lot more than used on other Cards or Boards.

Ranked by Vcore ripple under load:
1. MSI Lightning with pwm clock tuner ON: ~55 mV peak-to-peak
2. ASUS Direct CU 2: ~60 mV peak-to-peak
3. MSI Lightning with pwm clock tuner OFF: ~70 mv peak-to-peak
Measured on a decoupling capacitor right under the core.

[amlett]

Well, it seems that no matter how much voltage I put in the card can't go over 1ghz. I've tried with the vcore switch (also with OCP switch) and an unlocked bios (1.21) and even with temps lower than 55º it can't pass the ghz. The voltmeter reads 1.26v on load.

I was expecting better OC on water

Anyhow, thanks to all for your help.